Observational survey of financial difficulties among patients with multiple myeloma and chronic lymphocytic leukaemia treated at US community oncology clinics (Alliance A231602CD)

Abstract

Objectives

To estimate the proportion and correlates of self-reported financial difficulty among patients with multiple myeloma (MM) or chronic lymphocytic leukaemia (CLL).

Setting

Sixty-six US community and minority oncology practices affiliated with the National Cancer Institute Community Oncology Research Programme (NCORP).

Participants

A total of 521 patients (≥18 years) with MM or CLL consented and 416 responded to a survey (completion rate=79.8%). Respondents had a MM diagnosis (74.0%), an associate degree or higher (53.4%), were White (89.2%), insured (100%) and treated with clinician-administered drugs (68.0%).

Study design

Observational, theoretical model and protocol-based patient survey administered between May 2019 and June 2020.

Primary and secondary outcome measures

Financial difficulty was assessed using a single-item measure, the EORTC QLQC30: ‘Has your physical condition or medical treatment caused you financial difficulties in the past year?’ and using an ‘any-or-none’ composite measure of 22 items assessing financial difficulty, worries and the use of cost-coping strategies. Multivariable logistic regression models assessed the association of financial difficulty with diagnosis, socioeconomic and treatment characteristics.

Results

About 16.8% reported experiencing financial difficulty using the single-item measure and 60.3% using the composite measure. Most frequently endorsed items in the composite measure were financial worry about having to pay large medical bills related to cancer and difficulty paying medical bills. Financial difficulty using the single-item measure was associated with having MM vs CLL (adjusted OR (aOR), 0.34; 95% CI, 0.13 to 0.84; p=0.02), having insurance other than Medicare (aOR, 2.53; 95% CI, 1.37 to 4.66; p=0.003), being non-White (aOR, 2.21; 95% CI, 1.04 to 4.72; p=0.04) and having a high school education or below (aOR, 0.36; 95% CI, 0.21 to 0.64; p=0.001). Financial difficulty using the composite measure was associated with having a high school education or below (aOR, 0.62; 95% CI, 0.41 to 0.94; p=0.03).

Conclusions

US patients with MM and CLL report financial difficulty, especially those with low socio-economic status. Interventions are needed to mitigate patients’ financial difficulty.

Trial registration number

NCT03870633.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study was a theoretical model and protocol-based. Financial difficulty was conceptualised as multidimensional based on previously published literature and pilot interviews with patients, physicians and caregivers.

• The primary outcome variable was a previously developed and empirically tested measure. The secondary outcome measure was a composite aiming to assess how treatment impacted patients’ daily lives and informed decisions to delay or forego care and use cost-coping strategies.

• The study relied on medical chart abstraction to establish key independent variables including patient cancer diagnoses, treatment history and current comorbidities.

• The study had strong engagement among diverse community and minority cancer treatment sites despite the completion of recruitment coinciding with the initial stages of the COVID-19 pandemic.

• Small sample sizes limited our ability to examine differences in the experience of financial difficulty by geography, race and ethnicity, and blood cancer type, and measured confounders could bias the parameter estimates.

Background

The haematologic malignancies multiple myeloma (MM) and chronic lymphocytic leukaemia (CLL) represent a small percentage of all cancers affecting persons in the United States.¹ Treatment varies for these cancers from watchful waiting to intensive chemotherapy with or without a haematopoietic stem cell transplant at specialised centres and has experienced significant advancement resulting in greater survivorship over the past two decades.^{2 3} Current MM treatments often involve multiple therapies, including chemotherapy, targeted drugs, transplant and CAR-T.³,⁹ In CLL, targeted treatments, such as Bruton’s tyrosine kinase inhibitors (BTKi’s: ibrutinib, acalabrutinib and zanubrutinib), B-cell lymphoma (BCL)-2 inhibitor (venetoclax), phosphoinositide 3-kinase(PI3K) inhibitors (idelalisib, duvelisib), and their combinations have become standard of care over older chemotherapeutic regimens.²¹⁰,¹³

The costs of these activities and their coordination over prolonged periods of time and regular and ongoing management by specialists may deplete MM and CLL patients’ financial resources, interfere with their and their caregivers’ ability to work, and make it difficult to afford other necessities.⁷¹⁰,¹⁸ Some of these costs comprise direct medical expenses, such as paying for physician visits, transplants^{4 16 19} and other treatments, including the high prices of targeted agents.⁷,⁹¹⁵ Other costs are indirect, such as lost income due to an inability to work and travel costs.⁴⁶,8 10 12 21 22 There are numerous terms used to express the cancer patient cost burden associated with their diagnosis and treatment, including financial toxicity, financial stress, financial hardship, financial distress, economic burden, economic stress, economic hardship or economic distress.¹⁷ The intent of these terms is to describe the stress and hardship arising from the financial burden of cancer treatment, and their multiplicity belies their complexity, indicating both the presence of stress and hardship and their correlates, including various cost-coping strategies, such as skipping medication, taking less medication or not filling a prescription.¹⁷²³,³³ In this study, we use the term ‘financial difficulty’ to encompass the multidimensional nature of a patient’s cost burden. Financial difficulty has been previously defined and includes an inability to pay for basic necessities such as food and utilities, the presence of medical debt and high out-of-pocket spending relative to income and the use of cost coping strategies.^{17 25 34 35}

This Alliance/NCI protocol-based study sought to describe self-reported financial difficulty among US patients treated in the community with MM or CLL and identify factors associated with such difficulty. Our prespecified hypotheses are based on the theoretical model of Yabroff (2018) and were that patient self-reported financial difficulty is associated with treatment, diagnosis and comorbidities, and socioeconomic characteristics including patient sex, race and ethnicity, education and the presence and type of insurance coverage.^{1733 36},³⁹

Methods

Study design

A231602CD was a multicentred observational, prospective theoretical model and protocol-based study of patients diagnosed with the blood cancers MM or CLL and receiving treatment at a National Cancer Institute Community Oncology Research Programme(NCORP) site (clinicaltrials.gov study identifier NCT03870633). The protocol is appended to this study. NCORP, a programme of the National Cancer Institute (NCI), is a national network for cancer clinical trials and care delivery studies that is comprised of 7 research bases and 46 community sites across the US, 14 of which are designated as Minority/Underserved community sites. NCORP sites are consortia of researchers, hospitals, practices, medical centres and other groups that provide healthcare services.⁴⁰ The study was administered through the Alliance for Clinical Trials in Oncology research base, and the NCI Central Institutional Review Board (CIRB) served as the IRB of record. All NCORP community sites were invited to participate in the study. Patient sample size was prespecified based on a power calculation of the precision of the CI for a proportion. NCORP site staff recruited patients based on study eligibility criteria and a limited medical record review, and they obtained written, remote, verbal or electronic consent. Participants were mailed a $20 gift card on completion of the 60-min telephone survey administered by the study team. The study was open to patient recruitment by the sites on 14 May 2019.

Participating patients

Study eligibility was restricted to adult patients (≥18 years of age) who (1) had been prescribed or recommended to receive drug-based anticancer therapy, whether administered orally or by infusion, within the prior 12 months; (2) were not currently enrolled in a clinical trial in which a drug was supplied; and (3) were able to read and comprehend English or Spanish. Patients with a psychiatric illness or other mental impairment that would preclude their ability to give informed consent or respond to the telephone survey were excluded from the study.

Survey design

The study design was approved by the Alliance/NCI and conducted per protocol.

A comprehensive literature review and extensive pilot interviews with patients, caregivers and physicians conducted in Summer and Fall 2018 informed the formal study design. The study employed a theoretically grounded patient financial assessment survey comprising multiple domains, including financial difficulty, patient socioeconomic indicators, and health and well-being (online supplemental eAppendix 1). Questionnaire items assessing socioeconomic indicators, health and well-being were drawn from the Health and Retirement Survey, a validated national survey, and questionnaire items assessing financial difficulty and cost coping behaviours were drawn from other previously published study instruments (online supplemental eAppendix 1 one for a guide to items and the protocol for their sources).45 7 13 15,17 23 24 26 29 30 33 36 38 41 All items had closed-ended responses, with recall periods of either ‘now’ or ‘in the past 12 months.’ In some cases, the recall period of previously published items with a recall of 1 or 3 months was modified to ask patients about their experience in the past 12 months. These changes were reviewed by a survey methodologist in accordance with guidelines published by Stull et al.⁴⁴ This change enabled comparison and pooling of data across items and domains. In several cases, we relied on the structure of previously tested, validated questions and changed a small number of words to make it more specific to the cancers we studied; for example, we asked a question about financial difficulty paying for oral chemotherapy, which modified a previously fielded question that asked about medication generally. We pilot-tested the complete survey in 10 patients and made minor revisions to a few items to improve their clarity. Pilot testing of the survey occurred over the phone and was conducted in Fall 2018.

Self-reported socioeconomic information included sex, race, education level and insurance type. NCORP site staff abstracted information from the medical records of consented patients: date of diagnosis, treatment history and current treatments, including treatment initiation dates, current comorbidities, sex and date of birth. The medical abstraction applied to the past 12 months, with some exceptions, such as date of diagnosis, which may have been outside this window. Exposure to patient-administered and clinician-administered anticancer drugs was measured (see online supplemental etable 2 in the Supplement for full list of anticancer drugs that were considered patient-administered and clinician-administered and their frequency of use).

Outcome measures

The study was powered to assess responses to a single item from the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire – Core 30⁴²,⁴⁴ (EORTC QLQ-C30) with a modified recall period: ‘Has your physical condition or medical treatment caused you financial difficulties in the past year?’ The EORTC-QLQ-C30 is a cancer-specific instrument developed in the 1990s by international researchers. The standard measure uses a recall period of 1 week. However, due to the typical periodicity of medical billing and the chronic duration of treatment for the target patients, we modified the question to use a 12-month recall period. The choice was also informed by previously published work on recall periods for survey design^{35 44} and approved by the Alliance/NCI study review board. Based on previously published scoring of the instrument and approval by the study review board, participant responses were dichotomised (ie, ‘Not at all’ or ‘A little’ classified as No, ‘Quite a bit’ and ‘Very much’ classified as Yes).

A secondary composite measure was created to capture additional aspects of financial difficulty based on theoretical relevance, assessed in previously published studies of MM and CLL patients and based on patient, caregiver and physician pilot interviews. The composite measure included the following topics (see online supplemental eAppendix 1 in the Supplement for full details): Difficulty Paying Medical Bills (Questions 1, 2, 21), Delaying or Foregoing Medical Care (Questions 4, 6, 6 a-6f), Financial Worry (Questions 9, 10, 11, 12, 20), Cost-Coping Strategies (Questions 16 a-c, 16 f) and Treatment-Related Debt or Bankruptcy (Questions 17 and 18). The composite measure was approved by the Alliance/NCI study review committee and prespecified in the protocol. To score the composite measure, the study team prespecified the following procedure. To determine the number of factors underlying these 22 items, an exploratory factor analysis and a screen test were performed⁴⁵; the latter identifies the ‘elbow’ of a scree plot and retains all factors above the elbow. The scree plot (online supplemental file 2 in the Supplement) supported a unidimensional construct and the one-factor model had acceptable fit, χ²(209)=501.93, p<0.001, RMSEA=0.059, CFI=0.96. If the patient endorsed any of the 22 items, then they were categorised in a binary variable as having financial difficulty per the composite measure. To facilitate ease of interpretation of the composition measure, an ‘any-or-none’ scoring was applied.

Data collection

Medidata’s Rave Electronic Data Capture System (Rave) was set up centrally by the Alliance to collect patient self-reported survey responses and medical record abstracted items from each site recruiting patients in a standardised format. Rave is an electronic data capture system that the Alliance uses as its primary clinical data capture system to capture, manage and report data for all the clinical trials it runs. The survey was built in Rave. Within 8 weeks after patient registration, the patient survey was conducted as a centralised telephone interview by study team survey staff located at the University of North Carolina (UNC). The UNC research staff conducting the phone interviews made up to eight attempts to reach a participant the first time. A voicemail message was left on the first and fourth call attempt, as indicated in the call script. If someone other than the patient answered the phone, they were asked about a good time to reach the participant. If a participant was unable to complete the phone interview in a single phone call, the interview was continued later within the 8-week window (from the time of registration). Data collected were entered into Medidata Rave by the UNC study team survey staff, who were trained using a standard protocol to enter data into Rave. The study statistical and data management team also worked with the UNC study team survey staff to address any issues regarding data quality. For the medical record abstraction, clinical research associates at each site were trained using a standard protocol to identify each medical record item and manually enter medical record data into Rave. Ad hoc questions by the associates were answered in real time, and challenges in identifying or recording information into Rave were centrally managed and resolved by the study team. After data were collected for each recruited subject, the study data manager queried missingness and non-sensical entered values (eg, date of diagnosis happened prior to date of birth) and recontacted associates to resolve outstanding concerns.

Statistical analysis

Statistical analyses were prespecified in the approved protocol. Patient socioeconomic characteristics were examined descriptively overall, by cancer type and according to report of financial difficulty based on responses to the single-item measure and the composite measure. As all patients completed the survey in English, language was not included among the covariates examined.

Two multivariable logistic regression models were estimated separately for the two outcomes to assess the associations between patient characteristics and financial difficulty. The predictors for both models included sex (male/female), race (White/non-White), cancer type (MM/CLL), comorbidity (Charlson score=0/Charlson score ≥1), treatment (not clinician-administered/clinician-administered), education (High School Diploma, General Educational Development (GED), or below/above High School Diploma) and insurance type (Medicare/Medicare+Other/Other). The definition of the predictor variables is described in online supplemental eAppendix 2. The predictor variables were selected based on the theoretical model proposed by Yabroff (2018). Adjusted and unadjusted odds ratios (aORs and ORs, respectively), CIs and P values were calculated for each predictor. We used listwise deletion to account for missing data and conducted multiple imputation as a sensitivity analysis.⁴⁶

The relationship between the single-item measure and composite measure of financial difficulty was examined by calculating the proportion and CI of patients endorsing each item in the composite measure for the patients who endorsed the single-item measure and those who endorsed the composite and split by MM and CLL diagnosis.

For all models and comparisons, two-sided α=0.05 was used to determine statistical significance with no adjustment for multiplicity. Additionally, for statistical comparisons on the demographic variables by diagnosis and by endorsement of financial difficulty, we also applied the false discovery rate (FDR) correction to adjust for multiplicity.⁴⁷ All analyses were performed by the Alliance Statistics and Data Centre using SAS V. 9.4 with data frozen on 6 October 2021. STROBE reporting guidelines were followed.

Patient and public involvement

Patients were involved in the study conceptualisation, survey design and study conduct.

Results

Study recruitment ended on 15 June 2020 when the Alliance study monitor determined if the study had met its predefined patient recruitment goal of 500 total patients consented for survey participation. 521 patients from 66 affiliated practices clustered within 23 NCORP sites were registered to the study. Participating sites and the number of evaluable patients they accrued are listed in online supplemental etable 1 in the Supplement.

Of the 521 patients consented to the study, 416 completed all or part of the patient survey for a 79.8% completion rate (figure 1). Patients were enrolled from 66 affiliated practices associated with 23 NCORP community sites. There were 115 total affiliated practices among the 23 NCORP sites. Consequently, only 57.3% of enrolled affiliated practices enrolled a patient to the trial.

Figure 1. Legend: Consort Diagram. A description of patient study enrolment and survey completion.

CLL, chronic lymphocytic leukaemia; MM, multiple myeloma

Patients with MM represented 75% (n=308) of the full sample (n=416). Most respondents were male (56.5%; n=235), White (89.2%; n=371), had at least some education post high school (54.3%; n=222/409), and were treated with at least one clinician-administered therapeutic (68.0%; n=283) (table 1). All patients reported they were insured (table 1). Characteristics were similar by cancer types except CLL patients were older (71.2 vs 67.5; p<0.001), less likely to currently be taking clinician-administered therapies (26.9% vs 82.5%; p<0.001), and more likely to have Medicare plus another form of insurance (76.9% vs 63.3%; p=0.03) compared with MM patients. Due to the small sample of CLL patients recruited, we pooled all regression analyses across diagnoses.

Table 1. Socioeconomics and Treatment Characteristics of the 416 Patients Who Responded to All or Part of the Patient Survey.

Characteristics	MMNo. (%)	CLLNo. (%)	TotalNo. (%)	P value*	FDR-adjusted P value
No. of patients	308	108	416	NA	NA
Age in years, Mean (SD)	67.5 (9.79)	71.2 (8.01)	68.5 (9.49)	<0.001†	0.007
Sex				0.04	0.09
Female	143 (46.4)	38 (35.2)	181 (43.5)
Male	165 (53.6)	70 (64.8)	235 (56.5)
Race				0.33	0.48
White	272 (88.3)	99 (91.7)	371 (89.2)
Non-White‡	36 (11.7)	9 (8.3)	45 (10.8)
Geographic Region§				0.34	0.48
Northeast	20 (6.5)	9 (8.3)	29 (7.0)
Midwest	195 (63.3)	59 (54.6)	254 (61.1)
South	78 (25.3)	31 (28.7)	109 (26.2)
West	15 (4.9)	9 (8.3)	24 (5.8)
Education				0.21	0.42
High School Diploma/GED or below	132 (43.9)	55 (50.9)	187 (45.7)
Above High School	169 (56.1)	53 (49.1)	222 (54.3)
Missing	7	0	7
Home Ownership Status				0.31	0.48
Homeowner	255 (83.6)	96 (89.7)	351 (85.2)
Non-Homeowner	50 (16.4)	11 (10.3)	61 (14.8)
Missing	3	1	4
Current Employment Status				0.44	0.56
Employed	79 (25.8)	32 (29.6)	111 (26.8)
Unemployed	227 (74.2)	76 (70.4)	303 (73.2)
Missing	2	0	2
Reason for Unemployment				0.01	0.04
On Temporary Layoff	3 (1.3)	0 (0)	3 (1.0)
Retired	154 (67.8)	66 (86.8)	220 (72.6)
Taking Care of Home or Family	1 (0.4)	0 (0)	1 (0.3)
Unable to Work Due to Illness or Disability	66 (29.1)	8 (10.5)	74 (24.4)
Unknown	3 (1.3)	2 (2.6)	5 (1.7)
Reported Household Income				0.87	0.94
Less than $20,000	40 (14.4)	10 (10)	50 (13.3)
$20,000 to $39,999	52 (18.8)	19 (19)	71 (18.9)
$40,000 to $59,999	49 (17.7)	22 (22)	71 (18.9)
$60,000 to $79,999	42 (15.2)	15 (15)	57 (15.2)
$80,000 to $99,999	30 (10.8)	10 (10)	40 (10.6)
$100,000 or more	64 (23.1)	23 (23)	87 (23.1)
Missing	31	9	40
Insurance Type				0.03	0.08
Medicare only	18 (5.8)	5 (4.6)	23 (5.5)
Medicare+other insurance	195 (63.3)	83 (76.9)	278 (66.8)
Other¶	95 (30.8)	20 (18.5)	115 (27.6)
Charlson Comorbidity Index score				0.94	0.94
one or more	110 (35.7)	39 (36.1)	149 (35.8)
0	198 (64.3)	69 (63.9)	267 (64.2)
Treatment Received				<0.001	0.007
Clinician-Administered Therapeutic	254 (82.5)	29 (26.9)	283 (68.0)
Not a Clinician-Administered Therapeutic	54 (17.5)	79 (73.1)	133 (32.0)
Single-item Measure of Financial Difficulty				0.002	0.009
Yes	62 (20.1)	8 (7.4)	70 (16.8)
No	246 (79.9)	100 (92.6)	346 (83.2)
Composite Measure of Financial Difficulty				0.62	0.72
Yes	188 (61.0)	63 (58.3)	251 (60.3)
No	120 (39.0)	45 (41.7)	165 (39.7)

^*All P values come from a χ2 test unless otherwise noted.

^†Kruskal–Wallis P value.

^‡Non-White race captures patients who selected American Indian or Alaska Native, Asian, Black, or African American, Native Hawaiian or other Pacific Islander, Unknown, Not Reported.

^§Geographic Regions were divided according to the United States census divisions and regions: Northeast: Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut, New York, New Jersey and Pennsylvania; Midwest: Ohio, Michigan, Indiana, Wisconsin, Illinois, Minnesota, Iowa, Missouri, North Dakota, South Dakota, Nebraska and Kansas; South: Delaware, Maryland, Virginia, West Virginia, Kentucky, North Carolina, South Carolina, Tennessee, Georgia, Florida, Alabama, Mississippi, Arkansas, Louisiana, Texas and Oklahoma Washington DC; West: Montana, Idaho, Wyoming, Colorado, New Mexico, Arizona, Utah, Nevada, California, Oregon, Washington, Alaska and Hawaii.

^¶Other includes Private Insurance, Medicaid, Military Sponsored (including CHAMPUS & TRICARE), Veterans Sponsored, Not Otherwise Specified (NOS), and any combination of those listed here.

CLL, Chronic Lymphocytic Leukaemia; FDR, False Discovery Rate; MM, Multiple Myeloma; NA, Not Applicable.

Single-item measure of financial difficulty

Across the 416 patients, 16.8% (95% CI, 13.4% to 20.8%) endorsed financial difficulty by responding ‘Quite a bit’ or ‘Very much’ to the item ‘Has your physical condition or medical treatment caused you financial difficulties in the past year?’ The socioeconomic characteristics of those who reported financial difficulty based on this single item are summarised in online supplemental etable 3 in the Supplement.

After applying listwise deletion to the seven predictors and outcome variable used in the multivariable logistic regression model, the analytic sample consisted of 408 patients. The multivariable analyses are presented in table 2. Patients with CLL had significantly lower odds of reporting financial difficulty than patients with MM (aOR, 0.34; 95% CI, 0.13 to 0.84; p=0.02) as did patients who have above a high school education compared with patients with a high school education or below (aOR, 0.36; 95% CI, 0.21 to 0.64; p=0.001). Additionally, patients who did not have Medicare had significantly greater odds of reporting financial difficulty than patients with Medicare plus one other type of insurance (aOR, 2.53; 95% CI, 1.37 to 4.66; p=0.003) as did non-White patients compared with White patients (aOR, 2.21; 95% CI, 1.04 to 4.72; p=0.04). Listwise deletion is the most common missing data handling method; however, it has its limitations.⁴⁸ Thus, we have added multiple imputation as a sensitivity analysis, described the procedure and reported the results in online supplemental table 4. We obtained similar results when the analysis was restricted to the complete cases only.

Table 2. Multivariable Logistic Regression Results of Survey Respondent Self-Report of Financial Difficulty.

Model	Unadjusted odds ratios (95% CI)	P value	Adjusted odds ratios (95% CI)*	P value
Single-item Measure of Financial Difficulty
Female	1.32 (0.78, 2.21)	0.30	1.09 (0.62, 1.91)	0.77
Non-White	2.97 (1.49, 5.90)	0.002	2.21 (1.04, 4.72)	0.04
Comorbidity	1.39 (0.82, 2.36)	0.27	1.53 (0.86, 2.75)	0.15
CLL	0.31 (0.15, 0.68)	0.003	0.34 (0.13, 0.84)	0.02
Clinician-Administered Treatment	1.72 (0.94, 3.15)	0.08	1.02 (0.49, 2.12)	0.96
Above HS Education	0.42 (0.24, 0.71)	0.001	0.36 (0.21, 0.64)	0.001
Medicare+Other†	NA (Reference)	NA	NA (Reference)	NA
Medicare	3.17 (1.21, 8.27)	0.02	2.17 (0.77, 6.07)	0.14
Other‡	2.50 (1.43, 4.37)	0.001	2.53 (1.37, 4.66)	0.003
Composite Measure of Financial Difficulty
Female	1.35 (0.90, 2.01)	0.15	1.28 (0.84, 1.94)	0.25
Non-White	1.84 (0.92, 3.69)	0.09	1.56 (0.75, 3.23)	0.24
Comorbidity	1.28 (0.84, 1.94)	0.29	1.27 (0.81, 1.97)	0.30
CLL	0.88 (0.56, 1.37)	0.57	1.21 (0.71, 2.09)	0.48
Clinician-Administered Treatment	1.52 (0.998, 2.32)	0.05	1.61 (0.97, 2.68)	0.07
Above HS Education	0.62 (0.41, 0.92)	0.02	0.62 (0.41, 0.94)	0.03
Medicare+Other†	NA (Reference)	NA	NA (Reference)	NA
Medicare	3.64 (1.21, 10.98)	0.02	3.00 (0.98, 9.21)	0.06
Other‡	1.47 (0.94, 2.32)	0.10	1.57 (0.97, 2.54)	0.07

^*The adjusted models include all predictors listed in the table.

^†The insurance variable was defined by two dummy variables with patients who have Medicare and another insurance (ie, Medicare+Other) as the reference group.

^‡Other insurance includes Private Insurance, Medicaid, Military Sponsored (including CHAMPUS & TRICARE), Veterans Sponsored, Not Otherwise Specified (NOS), and any combination.

CLL, Chronic Lymphocytic Leukaemia; HS, High School; NA, Not Applicable.

As an additional sensitivity analysis, we investigated other dichotomisations (Not at all/A little/Quite a bit vs Very much and not at all vs A little/Quite a bit/Very much) of the single-item measure. These results are summarised in online supplemental etable 5. Additionally, we provide the proportions of endorsement for each category of the primary endpoint for the overall sample and split by diagnosis in online supplemental etable 6. Due to the low endorsement of ‘Very much,’ the results of the Not at all/A little/Quite a bit vs Very much dichotomisation should be interpreted with caution. We found that for the Not at all vs A little/Quite a bit/Very much dichotomisation model, diagnosis and education were still significant predictors of financial difficulty, but the direction of the effects was reversed compared with the main analysis.

Composite measure of financial difficulty

All respondents who completed the single-item measure also completed the composite measure of financial difficulty (n=416 patients). More than half of respondents (n=251) affirmed at least one of the 22 items in the composite measure of financial difficulty (60.3%; 95% CI, 55.6% to 64.9%). The most frequently endorsed item was, ‘In the past 12 months, have you ever worried about having to pay large medical bills related to your cancer? (Part 1, Question 20).’ Other commonly endorsed items include difficulty paying medical bills in general and forgoing dental care. Table 3 presents the proportion and CI of patients endorsing each of the items in the composite measure of financial difficulty for those patients who endorsed the single-item measure of financial difficulty and for those patients who indicated financial difficulty according to the any-or-none composite.

Table 3. Number of Patients Who Endorsed the Items in Composite Measure of Financial Difficulty by Endorsement of Financial Difficulty of Single-item Measure of Financial Difficulty and Composite Measure.

Survey question	Endorsed Single-item measure (n=70)No.Proportion(95% exact CI)	Endorsed one or more composite items (n=251)No.Proportion(95% exact CI)
In the past 12 months, did you have problems paying or were unable to pay any medical bills? Include bills for doctors, hospitals, therapists, medication, equipment, nursing home or home. (Part 1, Question 1)	490.700(0.579, 0.804)	880.351(0.292, 0.413)
Do you or anyone in your family currently have medical bills that you are unable to pay at all? (Part 1, Question 2)	300.429(0.311, 0.553)	470.187(0.141, 0.241)
During the past 12 months, have you or someone in your family delayed medical care because you were worried about the cost (do not include dental care)? (Part 1, Question 4)	290.414(0.298, 0.538)	560.223(0.173, 0.279)
During the past 12 months, was there a time when you or someone in your family needed medical care but did not get it because you could not afford it? (Part 1, Question 5)	170.243(0.148, 0.360)	290.116(0.079, 0.162)
During the past 12 months, was there a time when you needed one of the following, but did not get it because you could not afford it? Prescription medicine (Part 1, Question 6 a)	180.257(0.160, 0.376)	410.163(0.119, 0.215)
During the past 12 months, was there a time when you needed one of the following, but did not get it because you could not afford it? Mental healthcare or counselling (Part 1, Question 6b)	60.086(0.032, 0.177)	120.048(0.025, 0.082)
During the past 12 months, was there a time when you needed one of the following, but did not get it because you could not afford it? Dental care (Part 1, Question 6 c)	330.471(0.351, 0.595)	790.315(0.258, 0.376)
During the past 12 months, was there a time when you needed one of the following, but did not get it because you could not afford it? Eyeglasses (Part 1, Question 6d)	220.314(0.209, 0.436)	460.183(0.137, 0.237)
During the past 12 months, was there a time when you needed one of the following, but did not get it because you could not afford it? Cancer-related medical care (Part 1, Question 6e)	130.186(0.103, 0.297)	210.084(0.053, 0.125)
During the past 12 months, was there a time when you needed one of the following, but did not get it because you could not afford it? Non-cancer related medical care (Part 1, Question 6 f)	190.271(0.172, 0.391)	310.124(0.086, 0.171)
If you get sicker or have an accident, how worried are you that you will not be able to pay your medical bills? (Part 1, Question 9)	400.571(0.448, 0.689)	660.263(0.209, 0.322)
How often in the last 12 months would you say you were worried or stressed about having enough money to pay your rent or mortgage? (Part 1, Question 10)	420.600(0.476, 0.715)	820.327(0.269, 0.389)
How often in the last 12 months would you say you were worried or stressed about having enough money to buy nutritious meals? (Part 1, Question 11)	370.529(0.406, 0.649)	660.263(0.209, 0.322)
How often in the last 12 months would you say you were worried or stressed about having enough money to pay household utilities such as water, gas, and electricity? (Part 1, Question 12)	41/690.594(0.469, 0.711)	74/2500.296(0.240, 0.357)
During the past 12 months, were any of the following true for you: You skipped medication doses to save money (Part 1, Question 16 a)	170.243(0.148, 0.360)	260.104(0.069, 0.148)
During the past 12 months, were any of the following true for you: You took less medicine to save money (Part 1, Question 16b)	160.229(0.137, 0.345)	290.116(0.079, 0.162)
During the past 12 months, were any of the following true for you: You delayed filling a prescription to save money (Part 1, Question 16 c)	280.400(0.285, 0.524)	460.183(0.137, 0.237)
During the past 12 months, were any of the following true for you: You used alternative therapies to save money (Part 1, Question 16 f)	40.057(0.016, 0.139)	180.072(0.043, 0.111)
In the past 12 months, have you or has anyone in your family had to borrow money or go into debt because of your cancer, its treatment, or the lasting effects of that treatment? (Part 1, Question 17)	310.443(0.324, 0.567)	470.187(0.141, 0.241)
In the past 12 months, did you or your family file for bankruptcy because of your cancer, its treatment, or the lasting effects of that treatment? (Part 1, Question 18)	20.029(0.004, 0.099)	3/2500.012(0.003, 0.035)
In the past 12 months, have you ever worried about having to pay large medical bills related to your cancer? (Part 1, Question 20)	600.857(0.753, 0.929)	1730.689(0.628, 0.746)
Please think about medical care visits for cancer, its treatment or the lasting effects of that treatment in the past 12 months. Have you ever been unable to cover your share of those visits? (Part 1, Question 21)	260.371(0.259, 0.495)	400.159(0.116, 0.211)

Note: Each cell presents the following information: number of patients who answered yes or always/usually/sometimes, the proportion, and the 95% Wilson score CI. If there was missing data, the total number of patients who answered the question was also provided.

The socioeconomic characteristics of those who endorsed the composite measure of financial difficulty are summarised in online supplemental etable 3 in the Supplement. Online supplemental etable 7 in the Supplement presents the proportion of patients endorsing each of the items in the composite measure of financial difficulty split by cancer type. For most items, a greater proportion of MM patients endorsed the item compared with CLL patients.

The logistic regression with the composite measure of financial difficulty as the outcome is presented in table 2 for the 408 patients with complete data. Education was significantly associated with the composite measure of financial difficulty such that patients with above a high school education had lower odds of reporting financial difficulty compared with patients with a high school education or below (aOR, 0.62; 95% CI, 0.41 to 0.94; p=0.03).

Relation between single-item measure and composite measure of financial difficulty

The single-item measure and the any-or-non composite measure of financial difficulty were dependent such that all 70 patients who endorsed the single-item measure endorsed one or more items of the composite measure. As a sensitivity analysis, we investigated differences between patients who indicated financial difficulty according to both the single-item measure and composite measure and those who did not. We created three non-overlapping groups: endorsed the single-item measure, endorsed the composite but not the single-item measure and did not endorse either measure. We tested for statistical differences in patient sociodemographic and comorbidity information by this three-group categorisation, and the results are presented in table 4 with the corresponding P value. If we assume endorsing the single-item measure indicates the most financial difficulty and not endorsing either measure indicates the least, then as financial difficulty increased, patients were significantly more likely to be younger, non-White, less educated, unable to work due to illness/disability, have MM rather than CLL and have some form of insurance other than Medicare.

Table 4. Patient Demographics and Treatment Characteristics by Self Report of Financial Difficulty by Measure.

Variable	No. of Patients (%)			P value*	FDR-adjusted P value
	EndorsedSingle-item Measure	EndorsedComposite Only	EndorsedNeither
No. of patients	70	181	165
Age in years, mean (SD)	64.3 (11.09)	68.0 (9.05)	70.8 (8.58)	<0.001†	0.004
Sex				0.43	0.47
Female	34 (49)	81 (45)	66 (40)
Male	36 (51)	100 (55)	99 (60)
Race				0.007	0.01
White	55 (79)	164 (91)	152 (92)
Non-White‡	15 (21)	17 (9)	13 (8)
Geographic region§				0.69	0.69
Northeast	5 (7)	11 (6)	13 (8)
Midwest	42 (60)	111 (61)	101 (61)
South	19 (27)	52 (29)	38 (23)
West	4 (6)	7 (4)	13 (8)
Education				0.002	0.007
High school diploma/GED or below	44 (64)	81 (45)	62 (39)
Above high school	25 (36)	98 (55)	99 (61)
Missing	1	2	4
Home ownership status				0.04	0.07
Homeowner	52 (75)	155 (86)	144 (89)
Non-Homeowner	17 (25)	26 (14)	18 (11)
Missing	1	0	3
Current Employment Status				0.09	0.13
Employed	26 (37)	47 (26)	38 (23)
Unemployed	44 (63)	134 (74)	125 (77)
Missing	0	0	2
Reason for unemployment				<0.001	0.001
On temporary layoff	0 (0)	3 (2)	0 (0)
Retired	18 (41)	96 (72)	106 (85)
Taking Care of Home or Family	0 (0)	0 (0)	1 (1)
Unable to Work Due to Illness or Disability	26 (59)	34 (25)	14 (11)
Unknown	0 (0)	1 (1)	4 (3)
Reported Household Income				<0.001	0.004
Less than $20,000	19 (29)	24 (15)	7 (5)
$20,000 to $39,999	18 (28)	37 (22)	16 (11)
$40,000 to $59,999	16 (25)	35 (21)	20 (14)
$60,000 to $79,999	6 (9)	20 (12)	31 (21)
$80,000 to $99,999	4 (6)	14 (8)	22 (15)
$100,000 or more	2 (3)	36 (23)	49 (34)
Missing	5	15	20
Insurance Type				0.003	0.008
Medicare only	7 (10)	12 (7)	4 (3)
Medicare+other insurance	34 (49)	123 (68)	121 (73)
Other¶	29 (41)	46 (25)	40 (24)
Charlson Comorbidity Index score				0.36	0.43
1 or more	29 (41)	67 (37)	53 (32)
0	41 (59)	114 (63)	112 (68)
Treatment received				0.10	0.13
Clinician-administered therapeutic	54 (77)	125 (69)	104 (63)
Not a clinician-administered therapeutic	16 (23)	56 (31)	61 (37)
Diagnosis				0.008	0.01
Multiple myeloma	62 (89)	126 (70)	120 (73)
Chronic lymphocytic leukaemia	8 (11)	55 (30)	45 (27)

^*All P values come from a χ2 test unless otherwise noted.

^†Kruskal–Wallis P value.

^‡White race captures patients who did not select American Indian or Alaska Native, Asian, Black, or African American, Native Hawaiian or other Pacific Islander, Unknown, Not reported.

^§Geographic Regions were divided according to the United States census divisions and regions: Northeast: Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut, New York, New Jersey and Pennsylvania; Midwest: Ohio, Michigan, Indiana, Wisconsin, Illinois, Minnesota, Iowa, Missouri, North Dakota, South Dakota, Nebraska and Kansas; South: Delaware, Maryland, Virginia, West Virginia, Kentucky, North Carolina, South Carolina, Tennessee, Georgia, Florida, Alabama, Mississippi, Arkansas, Louisiana, Texas, and Oklahoma and Washington DC; West: Montana, Idaho, Wyoming, Colorado, New Mexico, Arizona, Utah, Nevada, California, Oregon, Washington, Alaska and Hawaii.

^¶Other includes Private Insurance, Medicaid, Military Sponsored (including CHAMPUS & TRICARE), Veterans Sponsored, Not Otherwise Specified (NOS), and any combination of those listed here.

Finally, we report the correlations (online supplemental etable 8) of the exposures and outcomes so readers can see the relationships between all variables used in the main model.

Discussion

In this multicentred protocol-based study of MM and CLL patients conducted across 23 US NCORP sites and their 66 affiliates, 16.8% specifically reported experiencing financial difficulty over the past 12 months using a single-item measure, and 60.3% endorsed financial difficulty items using a composite measure with questions about financial worry and difficulty paying medical bills eliciting the most affirmative responses. This study had strong engagement and participation across 23 diverse NCORP Sites across the country and their 66 affiliates. Strong site engagement resulted in high patient recruitment and retention rates for this study (79.8%). Between May 2019 and June 2020, most participants (n=490) were recruited to the study, despite final recruitment timing being coincident with the initial stages of the COVID-19 pandemic.

This is the first national study based on Yabroff’s theoretical model to assess the proportion of patients with financial difficulties and its correlates among a large number of patients with MM and CLL treated at community practices in the US. The reports of financial difficulty using the simple measure overall diagnosis are smaller in magnitude, but the composite measure results are similar to those reported in a previous study of patients with MM fielded in the USA,⁷ a study of 40 patients with MM treated in a single hospital in Korea⁸ and a study of patients with CLL reporting financial difficulty in paying out of pocket for their medicines.¹³ Study results suggest financial difficulty reports were concentrated in patients with MM in our study, but small samples of CLL patients limited our ability to conduct disease-specific analyses. This is an important direction for future work.

The findings are important and highly relevant to patients, their families, treating physicians and policymakers.⁴,1820 21 23 25 26 29 31 33 34 38 39 Our study also adds to the growing published literature on financial difficulties associated with CLL and MM, haematological malignancies and solid tumour-based cancers in the USA^{17 19 21 49} and worldwide.⁵⁰,⁵⁴ Like previously published studies focused on MM and CLL, we focused on patients with prevalent disease rather than incident cancer. This focus is important because both treatment regimens and disease management are prolonged in these malignancies, in part because of treatment-related survivorship, and consequently our study’s results articulate the cost burdens of MM and CLL care accumulated over time.^{7 8 13 17}

We believe our use of single-item and composite measures of financial difficulty adds to the limited literature on the range of financial difficulties experienced by haematological cancer patients.^{55 56} We used a survey tool composed of previously validated items that were modified for this population and new questions that were evaluated for comprehension, which facilitates studying internal and external validity. Our use of the composite measure of financial difficulty sought to present a more holistic picture of how cancer diagnosis and treatment impact patients’ daily lives and inform decisions to delay or forego care and the use of cost-coping strategies. Findings from our study are in line with research that demonstrates how experiences of financial difficulty impact the use of cost coping strategies (including treatment nonadherence), feelings of distress, family members/caregivers.^{17 19 21 41 49} Comparison of the single-item and composite item responses is also informative. Interestingly, though patients who endorsed the single-item measure had a higher proportion of endorsement on items in the composite measure compared with patients who did not, some patients who reported financial difficulty using the composite measure did not always report financial difficulty on the single item measure. We also found that surveyed patients expressed worry or concern about paying for household utilities, cancer and non-cancer related care and indicated delaying or foregoing treatment for cancer and non-cancer related care in response to cost concerns; they also reported using cost-coping strategies for cancer and non-cancer related care. One interpretation of these combined results is that it is possible patients may be implementing cost-coping strategies without recognising the connection between these strategies and the financial difficulty they face. One implication of our study results is that future research and potential interventions may consider using both the single item and the composite measure elements to screen patients for financial difficulty. Use of these measures may also provide a way for physicians and sites to engage in specific discussions with patients and to identify resources that can help patients manage care and non-care related costs.^{1723 25},^{27 31 33}

More generally, these findings are consistent with other studies that highlight the importance of engaging the care team and financial navigators in providing support and resources.⁵⁷,⁵⁹ Our study results suggest that the potential financial costs of treatment for blood cancer should be an aspect of patient education and provider awareness when patients undergo diagnosis, select initial treatment regimens and proceed in disease management. Some work has endeavoured to pilot patient or physician-directed screening and financial counselling for patients after initial cancer diagnosis, specific preferences for types of assistance from patients and advanced financial planning to reduce delays in the initiation or continuity of treatment.⁶⁰ Previous studies have also identified barriers to providing and accessing financial navigation services that disproportionately impact vulnerable patient populations, including rural, minority and younger patients. More importantly, patient and provider awareness of the financial costs of cancer care requires more transparency regarding the costs of care, insurance benefit design and their interdependencies.^{61 62} Yet, opacity and complexity are defining characteristics of the current US medical system, and these measures are not readily available. Consequently, while patient and provider education might sound like a tractable fix, moving towards these goals systematically would require significant effort.1725 26 31 33 49 61,⁶³

Our results also suggest that patients with MM and CLL may be underinsured for treatment expenses.^{23 64 65} Results suggest that low socioeconomic position is associated with greater report of financial distress, yet, all survey respondents, including those reporting financial difficulty, were insured. Since the survey was fielded, insurance rates in the USA among older adults have generally remained stable over time, but commodity product inflation and interest rate rises have increased financial pressure on individuals and families, including seniors living on fixed incomes.⁶⁶ Consequently, our results reporting self-reported financial difficulties among this population are likely biased downwards in their representation of financial difficulty experienced contemporaneously. Medicare is relatively a generous coverage compared with commercial insurance, and it generally covers all for persons aged 65 and older and people with disabilities, though during our study period Medicare’s coverage for self-administered drugs was inadequate. Since our study began, there have been important changes to coverage for self-administered drugs for Medicare beneficiaries through the Inflation Reduction Act that should reduce out-of-pocket spending for this type of treatment over time.⁶⁴,⁶⁶

While these reforms may improve access and alleviate financial difficulty among some patients with MM and CLL undergoing chronic treatment with oral agents, they do not extend to patients who are commercially insured.⁶⁶ Related to this point, study respondents insured by a primary and secondary payer were less likely to report financial difficulty, and those having insurance other than Medicare were more likely to report financial difficulty. Although we were unable to probe specific elements of insurance coverage and benefit designs among survey respondents, it is possible being covered by non-Medicare (eg, commercial) insurance may render individuals more exposed to out-of-pocket costs associated with cancer treatment due to the common use of deductibles, coinsurance and copayments.^{5 7 9 10 15 17 20 23 25 26 29 30 32 33} This suggests that one important avenue for reform aiming to reduce financial difficulty among cancer patients would be a more generous coverage across insurance types.

Other system reforms may support improved patient access to the needed cancer treatment and mitigate financial difficulty.¹⁷⁶⁷,⁶⁹ These include reductions in the costs of cancer treatments itelf, or more public and philanthropic funds to defray the costs of cancer treatment for targeted patients in need. Philanthropic funds to defray patient out-of-pocket costs for cancer treatment are ubiquitous but complex as they are typically geographically and insurance-coverage restricted (eg, only non-Medicare eligible patients may qualify), are provided for some treatments and not others (eg, patient assistance programmes available for brand drugs, not transplantation costs) and depend on the largess of entities that may have idiosyncratic objectives or be short-lived.⁶⁹,⁷² Previous studies have emphasised that cancer patient-level reports of financial difficulties are also related to practice-level characteristics.^{17 73} Our study did capture additional practice-level characteristics of the sites enrolling patients, and future planned analyses will focus on those outcomes.

This study has several additional limitations. First, the sample itself was not representative of the national MM and CLL patient populations. Black or African-American patients make up 20.5% of the MM population and 5.8% of the CLL population, respectively.^{15 16} However, in this study, these populations only accounted for 7.8% of the MM sample and 7.4% of the CLL sample. Although our research team had identified Minority and Underserved NCORP Sites to participate in the study, seven were unable to recruit prior to the study closure due to site-related issues (eg, electronic health record conversions, staff shortages) and/or COVID-19 related delays. Moreover, while NCORP sites from across the country participated in the study, 60% of patients that participated in the survey were from the Midwest. Although this is the largest and most representative study of financial difficulty in US-based patients with MM and CLL to date, our findings may not be fully generalisable to the national CLL and MM communities due to these limitations. Similarly, patients with MM tended to have higher endorsement on all items in the composite measure of financial difficulty. Unfortunately, due to sample size limitations, we were unable to robustly assess disease-specific correlates of financial difficulty. This is an important direction for future work. Second, financial difficulty is a multidimensional concept. While we used two different methods for measuring financial difficulty (single-item measure and composite measure), these two methods are limited in capturing the full scope of financial burden. Moreover, we believe we are the only published study to date to have pursued the use of a composite measure any or none scoring approach in studying financial difficulty. Our composite measure construction was based on previously fielded surveys and pilot interview results. Additional mixed method approaches to assessing the full range of patients experiences and their diversity are needed. Third, we acknowledge that modifying the recall period or content of previously fielded questions would certainly limit the ability to compare data from this study to data from other studies which had used the original instruments; however, we thought it was important to collect the most salient data to this problem, and we confirmed comprehension of the items ahead of fielding the survey in the pilot phase. Fourth, this was a cross-sectional study and therefore we did not examine the dynamic implications of patient self-reported financial difficulty on health and non-health outcomes. This is an important area for future work. Fifth, there is a risk of confounding due to unmeasured variables and potentially measurement error of variables in the logistic regression models. Finally, although the study met its recruitment goals and achieved a high response rate within fifteen months of opening, data coding, cleaning and analysis took longer than expected due to COVID-associated labour challenges. We do not believe these delays alter the external validity of our study results.

Conclusions

This is the first multicentre study in the USA to systematically assess the proportion of patients with financial difficulties and its correlates among patients with MM and CLL treated in the community. We found that patients with blood cancers experience financial difficulty, especially among those with low socioeconomic status. All patients reporting financial difficulty were insured, suggesting underinsurance as a contributing factor. Future work should examine the long-term health and non-health ramifications of financial difficulty among patients, evaluate the impact of policies on patient-reported financial difficulty and pursue novel comprehensive interventions to mitigate blood cancer patient’s financial difficulty and improve treatment access and equity.

Data availability statement

Data are available upon reasonable request.