Effect of Digital Health Coaching on Self-Efficacy and Patient-Reported Outcomes in Individuals with Acute Myeloid and Chronic Lymphocytic Leukemia: A Pilot Randomized Controlled Trial

Abstract

Introduction

Promotion of self-efficacy can enhance engagement with health care and treatment adherence in patients with cancer. We report the outcomes of a pilot trial of a digital health coach intervention in patients with leukemia with the aim of improving self-efficacy.

Methods

Adult patients with newly diagnosed acute myeloid leukemia (AML) and chronic lymphocytic leukemia (CLL) were randomized 1:1 to a digital health coach intervention or standard of care. The primary outcome of self-efficacy was measured by the Cancer Behavior Inventory (CBI) score.

Results

A total of 147 patients (37 AML, 110 CLL) were enrolled from July 2020 to December 2022. In the AML cohort, there was a mean increase in CBI score of 7.03 in the digital health coaching arm compared to a mean decrease of −3.57 in the control arm at 30 days (p = 0.219). There were no significant associations between the intervention and other patient-reported outcomes for patients with CLL.

Conclusion

There were numerical, but not statistically significant increases in self-efficacy metrics in AML patients who received digital health coaching. Although this trial was underpowered due to enrollment limitations during a pandemic, digital health coaching may provide benefit to patients with hematologic malignancy and warrants further investigation.

Introduction

The treatment landscape for leukemia has rapidly evolved over the past several years with the introduction of novel targeted and immune-based therapies that have improved event-free and overall survival (OS) of patients with both acute and chronic leukemias [1–3]. Yet each of these new treatments, in addition to the more traditional radiation and chemotherapy, is associated with risk for acute and chronic toxicities that can significantly affect a patient’s mood, energy, and quality of life [4–7]. Such toxicities require assessment and early intervention to reduce the burden of treatment-related side effects and long-term complications of therapy. Importantly, routine symptom monitoring in patients with metastatic solid tumor malignancies has been shown in a randomized controlled trial (RCT) not only to lead to significant improvements in physical function, symptom control, and quality of life but also to improve OS compared to usual care alone [8]. These findings not only underscore the importance of monitoring adverse events and sequelae related to treatment but also necessitate further studies investigating the optimal method to track and detect these therapy-related toxicities.

Digital or eHealth technologies are emerging as methods to monitor individuals with chronic disease. In the last few years, various electronic modules, surveys, and smartphone applications have been generated and tested in small cohorts of cancer patients as strategies to track and monitor symptoms and patient-reported outcomes (PROs) [9, 10]. Most of the published literature has reported on solid tumors such as breast cancer [11, 12], but more recent studies have evaluated the role of digital technology in PROs in hematologic malignancies as well [13–15]. Digital health coaching, which provides patient-oriented health education to achieve personal health-related goals utilizing an integrated mobile web or mobile experience, has been proposed as a strategy to promote healthy behaviors and improve self-efficacy. Few studies have evaluated the use of digital health coaches in improving self-efficacy and PROs in cancer patients [16].

In the context of this study, an established digital health coaching program was engaged to provide both interpersonal interaction as well as e-modules designed for patients with cancer diagnoses to better manage symptoms such as pain, fatigue, depression, anxiety, and navigate more effectively through the healthcare system. Pilot studies of this digital coaching intervention in small cohorts of patients with prostate cancer and patients undergoing stem cell transplant have established the feasibility of the intervention in the cancer population with modest improvements in self-efficacy scores over time [17–19]. In this larger single-center, pilot RCT, patients with the two most common types of adult-onset leukemia, acute myeloid leukemia (AML) and chronic lymphocytic leukemia (CLL), were separately randomized to a digital health coach intervention arm, or standard of care symptom monitoring with the aim of improving self-efficacy through a digital health coach program.

Patients and Methods

Digital Health Coaching Program

The digital health coaching program consists of an intensive engagement model between the participant and a health coach via the electronic avenue(s) of the participants’ choice, including text, e-mail, phone call, or web-based platform. Health coaches are individuals with a baccalaureate level education or higher, often from allied health backgrounds. All received training in evidence-based health coaching using the National Board of Health and Wellness Coaching (NBHWC)-specific curriculum for which they are required to sit for certification once eligible. Coaching involves general health and wellness principles aligned to cancer-related physiologic and psychosocial sequelae such as pain and fatigue but does not include specific medical advice. Call scripts, checklists, and condition-specific notes can guide the health coaches along with general weekly focuses across health domains that are influenced by a cancer diagnosis, such as physical, mental, financial, and social health, as well as nutrition, physical activity, and sleep. Each interaction is personalized based on the patient’s specific health goals and guided by tiny steps, smaller, measurable goals that can be focused on between weekly calls, and that build toward the achievement of cumulative health outcomes. The program consists of 60 scheduled electronic touch points over 3 months, but the patient can engage their health coach as frequently as the participant wishes. Each week there is a topic of focus such as physical health, mental health, financial health, nutrition, exercise, or sleep. At the start of the program, the participant is asked to identify an overarching health goal aligned to the area of importance to them. The health coach uses a weekly call to cover content from the topic of the week and check in on the goal, setting smaller goals that build toward the achievement of the larger goal. Up to four “nudges” or content delivered by text or email (depending on the participant’s access and preferences) are additionally delivered per week. Nudges include information curated based on the participant’s needs, goals, or focus for the week, such as a message of encouragement or accountability toward weekly goal, evidence-based content on the focus for the week, or a specific follow-up to a request from the participant. E-modules are also available on web-based platforms as a complement to the health coach.

Participants and Setting

Individuals were enrolled at a large, National Cancer Institute-designated comprehensive cancer center in the Southwest United States. The leukemia center sees over 2,000 individuals with newly diagnosed leukemia annually. Eligible patients were adults aged 18 and older with AML or CLL diagnosed within the prior 3 months who could read and speak fluently in English. AML and CLL were chosen as the most common type of acute and chronic leukemias, respectively. Additional inclusion criteria required participants to have internet access with a device able to receive calls, texts, or emails, as well as electronic surveys. Patients were excluded if they were elected supportive care/comfort care only, had undergone treatment for previous diagnoses of leukemia, or were suspected to have less than 6 months to live. Analyses for patients with AML and CLL were conducted separately as determined a priori given that differences in disease acuity and/or intensity of initial therapies could affect participation and intervention outcomes. A sample size of 100 per arm within each disease cohort would have 80% power to detect an effect size of 0.398 using a 2-sample t test at a 0.05 significance level. Planned enrollment was 500 patients (250 with AML and 250 with CLL) with an assumed 20% (n = 50 per cohort) dropout rate.

Trial Design

During the 18-month enrollment period, eligible patients were randomized 1:1 to either the digital health coach intervention arm or standard-of-care arm within 7 days after completion of baseline surveys electronically via the secure Research Electronic Data Capture (REDCap) program. Treatment groups were not blinded to the study team members or participants due to the nature of the intervention. Two separate randomizations were conducted – 1 for patients with AML and another for patients with CLL – using a form of adaptive randomization to balance study arms with regard to marginal distributions of covariables that may impact outcome. In this randomization, called minimalization, before each participant is assigned a treatment group, the total number of participants in each group with similar covariate characteristics is totaled. These totals are based on the marginal sums of the covariates so that each covariate is considered separately. Participants’ treatment assignments are determined based on which group assignment would provide the best overall balance with respect to those covariates. Covariates used in this study were generation (Greatest Generation or earlier, Baby Boomers, Generation X, Millennials), race (Asian, Black, White, other), and ethnicity (Hispanic, non-Hispanic). Randomization was conducted using the Biostatistics Departmental Clinical Trial Conduct website.

The standard-of-care support service was a telephone triage line which participants could call when they experienced any concern and reach a clinical nurse. All study participants were asked to complete the study instruments at baseline, and on days 30, 60, and 90 after enrollment; after day 90, the monitoring period was complete. There was no remuneration for participating survey completion. The study was approved by the local institutional review board, and all patients were consented electronically by study staff at the Cancer Center. The full trial protocol is available in the supplemental content.

Outcome Metrics

The primary outcome was a change in self-efficacy as measured by the Cancer Behavior Inventory (CBI) v3 [20]. The CBI is a 27-item instrument completed by the patient that measures self-efficacy in coping with cancer across seven major domains: seeking and understanding medical information, emotional regulation, coping with treatment-related side effects, accepting cancer/maintaining a positive attitude, seeking social support, and spiritual coping. It uses a 7-point response scale ranging from 1 “Not confident at all” to 7 “Confident” with higher scores indicating higher self-efficacy. Version 3 of the CBI has been validated with 1,405 individuals with cancer across four cohorts [20].

Secondary outcomes explored the relationship between self-efficacy and physiologic/psychosocial PROs using the Functional Assessment of Cancer Therapy (FACT) and MD Anderson Symptom Inventory (MDASI) instruments. FACT-G (general) and FACT-Leu (for patients with leukemia) tools were used. The FACT-Leu is a 44-item measure of quality of life including questions targeted to physical (7 items), social (7 items), emotional (6 items), functional (7 items), and disease- and treatment-related (17 items) domains. Responses are measured from 0 to 4 with higher scores indicating better quality of life [21]. This instrument has been validated in both the acute and chronic leukemia populations [22, 23]. The MDASI is a multiple-symptom measure of the severity of cancer-related symptoms and the functional inference caused by symptoms; it has been validated and shown to be sensitive to disease- and treatment-related changes [24–27]. Patients rate the severity of 13 physical, affective, and cognitive symptoms from 0 “not present” to 10 “as bad as you can image” as well as interference with functioning ranging from 0 “did not interfere” to 10 “interferes completely.” Higher scores indicate worse symptoms and interference with daily life. It was estimated that the total time to complete all study instruments was 30 min.

Statistical Analysis

Analyses were conducted separately for the AML and CLL cohorts. Summary statistics including frequencies, percentages, means, standard deviations, and ranges were used to describe patient demographics, clinical characteristics, CBI, FACT-Leu, and MDASI by treatment arm within each disease cohort. A 2-sample t test was used to compare change between intervention arms at 30 days. CBI, FACT-LEU, and MDASI scores over time were assessed using linear mixed models which include time in 30-day increments and intervention as fixed effects, and intercept and slope as a random effect. All statistical analyses were performed using Stata/MP v17.0 (College Station, TX, USA).

Results

Demographics

From July 2020 to December 2022, a total of 823 patients met trial eligibility criteria. Of them, 305 patients (37%) expressed initial interest and were sent informed consent documents. Of these, 158 (52%) subsequently did not return the completed consent or declined. Ultimately, 147 patients were randomized after completion of the baseline survey, 37 with AML and 110 with CLL (Fig. 1). While this was less than the projected 500 patients, after the planned 18-month enrollment period, the trial was ended due to staffing shortages that emerged during the COVID-19 pandemic. At 30 days, 13 patients with AML (35%) were either lost to follow-up or withdrew from the study (five in the control arm, eight in the intervention arm). In total, 16 of the 37 patients with AML (43%) completed the final survey at 90 days (9/19 [47%] in the control arm, 7/18 [39%] in the intervention arm). In the CLL cohort, more patients completed the final survey (81 of 110, 74%) with similar representation in each group (43/55 [78%] in the control arm and 38/55 [69%] in the intervention). Of the 50 total patients that did not complete the study, 10 patients withdrew (4 stating they misunderstood the program, 2 due to lack of time, 2 because they were no longer interested, 1 because of health problems, and 1 unknown) and 1 patient passed away, while the other 40 were lost to follow-up.

Fig. 1.

CONSORT Diagram. A total of 823 patients met the trial’s strict inclusion and exclusion criteria. Out of these, 305 patients were interested in receiving the informed consent. A total of 157 participants signed consent, 10 of which did not complete the baseline surveys and therefore were removed from the study. Ultimately, 37 patients with AML and 110 patients with CLL were independently randomized to the control or digital health coach intervention arms. Twenty-six of the 37 patients with AML did not complete the final study instruments at the completion of the study on day 90. Twenty-nine of the 110 patients with CLL did not complete the final study instruments.

Participant demographics are presented in Table 1. The average age of participants with AML was 60.3 years (Table 1). AML participants were predominantly male (57%) and non-Hispanic (89%). Similarly, the average age of participants with CLL was 62.1 years. Of participants 52% were female and 93% were non-Hispanic. Demographics with respect to age, gender, ethnicity, alcohol use, employment, and insurance were equally distributed between the control and intervention arms for both the AML and CLL cohorts.

Table 1.

Demographics and clinical characteristics of AML and CLL cohorts

Characteristic	Total (n = 37)	Control (n = 19)	Digital health coach (n = 18)	p value*
AML cohort
Mean age (SD), years	60.3 (14.6)	60.2 (14.8)	60.5 (14.3)	0.976
Gender, n (%)				0.886
Female	16 (43)	8 (42)	8 (44)
Male	21 (57)	11 (58)	10 (56)
Ethnicity, n (%)				0.999
Hispanic	4 (11)	2 (11)	2 (11)
Not Hispanic	33 (89)	17 (89)	16 (89)
Alcohol use, n (%)				0.694
Yes	10 (27)	4 (21)	6 (33)
No	27 (73)	15 (79)	12 (67)
Employment, n (%)				0.999
Full time	16 (43)	8 (42)	8 (44)
Not employed	21 (57)	11 (58)	10 (56)
Insurance, n (%)				0.999
Commercial	1 (3)	1 (5)	0 (0)
Governmental	1 (3)	1 (5)	0 (0)
Managed care	23 (62)	11 (58)	12 (67)
Medicare	11 (30)	5 (26)	6 (33)
Self-pay	1 (3)	51 (5)	0 (0)

Characteristic	Total (n = 110)	Control (n = 55)	Digital health coach (n = 55)	p value*
CLL cohort
Mean age (SD), years	62.1 (10.3)	62.8 (9.9)	61.5 (10.7)	0.523
Gender, n (%)				0.849
Female	57 (52)	28 (51)	29 (53)
Male	53 (48)	27 (49)	26 (47)
Ethnicity, n (%)				0.999
Hispanic	8 (7)	4 (7)	4 (7)
Not Hispanic	102 (93)	51 (93)	51 (93)
Alcohol use, n (%)				0.461
Yes	60 (55)	33 (60)	27 (49)
No	50 (45)	22 (40)	28 (51)
Employment, n (%)				0.941
Full time	57 (52)	27 (49)	30 (55)
Not employed	51 (46)	27 (49)	24 (44)
Unknown	2 (2)	1 (2)	1 (2)
Insurance				0.626
Commercial	1 (1)	0 (0)	1 (2)
Governmental	4 (4)	4 (2)	4 (4)
Managed care	67 (61)	35 (64)	32 (58)
Medicare	36 (33)	17 (31)	19 (35)
Self-pay	2 (2)	1 (2)	1 (2)

AML, acute myeloid leukemia; CLL, chronic lymphocytic leukemia; SD, standard deviation.

*Determined by a 2-sample t test.

Engagement with the digital health coach through phone calls, emails, and text messages was similar between the AML and CLL cohorts (Table 2). Text messages were the most common mode of interaction with an average of 47.2 text messages (SD 27.7) in 90 days for participants with AML and 61 text messages (SD 34.5) for participants with CLL. Participants with AML spent an average of 2.9 h (SD 2.1) of interaction by all three modes of communication with the digital health coach over the 90-day trial period, while participants with CLL spent an average of 3.7 h (SD 2.0).

Table 2.

Interaction with the digital health coach

	AML (n = 18), mean (SD)	CLL (n = 55), mean (SD)
Total contacts, n	66.7 (45.5)	83.1 (39.3)
Phone calls	12.6 (8.4)	13.5 (4.9)
Emails	7.6 (8.9)	9.2 (6.2)
Text messages	47.2 (27.7)	61 (34.5)
Total time, min	176.5 (126.9)	224.9 (121.4)

AML, acute myeloid leukemia; CLL, chronic lymphocytic leukemia; SD, standard deviation.

Self-Efficacy by the Cancer Behavior Inventory (CBI)

In the AML cohort, there was a mean increase in total CBI score of 7.03 at 30 days in the digital health coaching intervention arm compared to a mean decrease in total CBI score of −3.57 in the control arm (Fig. 2a). This numerical increase in the intervention arm compared to the control arm was not statistically significant (p = 0.219). For each category of CBI questions, there was a modest trend toward improvement in the patient-reported metrics of self-efficacy in the intervention arm, although none of these reached statistical significance (Fig. 2b). The largest CBI score increase was in the category of managing side effects, with an increase of 3.20 with the digital health coach versus −0.4 in the control arm (p = 0.122).

Fig. 2.

Box plots of mean change in CBI scores at 30 days in the AML cohort. The CBI is a validated 27-item instrument filled out by patients that measures self-efficacy in coping with cancer across seven major domains using a 7-point response scale ranging from 1 “Not confident at all” to 7 “Confident.” Higher scores indicate higher self-efficacy. CBI scores were measured at baseline and at 30 days post-intervention. Box and whisker plots were constructed to show a mean change in CBI score in both the intervention and control groups. a demonstrates a change in total CBI score, whereas b demonstrates the change in each individual domain of CBI score in patients with AML. The control group is shown in black (n = 14) and the digital health coach intervention group (n = 10) in blue. Dots indicated outliers.

In the CLL cohort, there were mean increases in the total CBI scores at 30 days in both the control and intervention arms (1.43 and 3.46, respectively) (Fig. 3a). These scores were not statistically different from one another (p = 0.556). There were no significant differences in any of the categories of CBI questions across the control or digital health coach arms in the CLL cohort (Fig. 3b).

Fig. 3.

Box plots of mean change in CBI scores at 30 days in the CLL cohort. The CBI is a validated 27-item instrument filled out by patients that measures self-efficacy in coping with cancer across seven major domains using a 7-point response scale ranging from 1 “Not confident at all” to 7 “Confident.” Higher scores indicate higher self-efficacy. CBI scores were measured at baseline and at 30 days post-intervention. Box and whisker plots were constructed to show a mean change in CBI score in both the intervention and control groups. a demonstrates a change in total CBI score, whereas b demonstrates the change in each individual domain of CBI score in patients with AML. The control group is shown in black (n = 48) and the digital health coach intervention group (n = 44) in blue. Dots indicated outliers.

Self-Efficacy and PROs over Time

Linear mixed models were used to assess the association between the digital health coaching intervention with various PROs and self-efficacy metrics over time through the 90-day trial period. There were positive regression coefficients between the digital health coach intervention and total CBI, FACT-G, and FACT-LEU scores in the AML cohort, which would be suggestive of improved self-efficacy (Table 3), but these were not statistically significant. There were small negative associations between the intervention and MDASI scores, which would indicate lessened symptoms from their malignancy, although again these data did not meet the predetermined significance threshold. In the CLL cohort, there were no significant associations between the intervention and the various self-efficacy metrics over time (Table 3).

Table 3.

Linear mixed models assessing association of digital health coach with self-efficacy and PROs over time

Self-efficacy or PRO metric	AML		CLL
	β coefficient of intervention (95% CI)	p value	β coefficient of intervention (95% CI)	p value
Total CBI	3.21 (−9.95, 16.38)	0.632	0.84 (−6.98, 8.66)	0.832
MDASI severity	−0.58 (−1.49, 0.34)	0.220	0.26 (−0.17, 0.70)	0.230
MDASI interference	−0.63 (−1.92, 0.67)	0.342	0.17 (−0.37, 0.72)	0.530
FACT-G total	8.17 (−0.93, 17.26)	0.078	−3.84 (−8.34, 0.67)	0.095
FACT-LEU total	11.32 (−2.44, 25.07)	0.107	−5.86 (−12.59, 0.87)	0.088

AML, acute myeloid leukemia; CBI, cancer behavior inventory; CI, confidence interval; CLL, chronic lymphocytic leukemia; FACT-G, functional assessment of cancer therapy general; FACT-LEU, functional assessment of cancer therapy for patients with leukemia; MDASI, MD Anderson symptom inventory; PRO, patient-reported outcome; SD, standard deviation.

Discussion

In this single-center pilot RCT, despite possible trends of improvement, there was no statistically significant difference in self-efficacy metrics in patients with newly diagnosed AML and CLL who received digital health coaching compared to those who received standard of care. While there was a numerical increase in CBI score of 7.0 points on average in the intervention arm for the AML cohort, the trial was underpowered due to enrollment limitations that overlapped with the COVID-19 pandemic. Enrollment was only 15% of the originally calculated sample for the AML cohort, and 44% for the CLL cohort. This low enrollment was compounded by a higher-than-expected dropout rate in the setting of the global pandemic. It is notable that the survey completion was higher in the CLL cohort than the AML cohort (74% vs. 43%) which may reflect that patients with newly diagnosed AML typically have more time-intensive inpatient and outpatient treatments which could limit their ability to engage in with a digital health coach and/or complete study instruments. During the pandemic, overall engagement with the healthcare system decreased. This, coupled with the time constraints of completing study surveys, likely played a role in the low participant retention. Unfortunately, these factors made it difficult to ascertain the precise effect that digital health coaching may have had in this population. While there was no remuneration, this could be a strategy considered in the future for survey completion to increase participant retention. Future studies should expect and account for high attrition rates, particularly with patients with advanced or acute life-threatening cancer.

Regardless, this pilot study demonstrates that even with this small cohort, a numerical increase in multiple metrics of self-efficacy was observed, as well as numerically lower scores on symptom burden surveys, indicating that health coaching may provide benefit in this population. Based on the standard deviation of scores at baseline, the minimum clinical important difference would be a change of approximately 2.5 points. Interestingly, these effects seem to be most pronounced in those with newly diagnosed AML and not CLL. It is unclear if these subtle differences are because of the small sample size, or if differences between the two types of leukemia could contribute to these variations. This trial did not capture treatment regimens during intervention as many patients with CLL often undergo monitoring alone. As the acuity of acute leukemia is initially greater than chronic leukemia, patients are often sicker at presentation which may lead to greater benefit of a health coach intervention early during the clinical course. To this point, the stage of cancer has been linked to the amount of psychologic stress a patient experiences [28, 29], and the trial by Basch and colleagues that demonstrated an OS benefit from symptom monitoring in cancer patients only included patients with newly diagnosed metastatic cancer [8]. In contrast, a recent RCT looking at mobile app symptom tracking with electronic education resources in patients with known CLL and multiple myeloma found no significant difference in PROs between the intervention and control arms [14]. While this study also had a small sample size and was not looking specifically at self-efficacy and health coaching, these data suggest that perhaps patients with classically more acute and symptomatic cancer types such as AML may experience a more immediate benefit from these supportive digital health resources.

It is also possible that digital technologies and/or health coaching may provide increased initial benefit and then provide waning utility over time as patients become more accustomed with their cancer diagnosis and connected with supportive resources. Two trials examined eHealth programs providing education and electronic monitoring of symptoms and found that self-efficacy and FACT-G scores improved initially in the first 1–2 months but then no longer demonstrated benefit over standard of care at 3–4 months [30, 31]. Neither of these programs included health coaching, but there may be an optimal time to introduce such a resource in a patients’ cancer journey to maximize psychological benefit and self-efficacy.

Ideally, these technologies would be available to all individuals regardless of demographics, including race, ethnicity, socioeconomic status, etc. This study was limited by the requirement of fluency in the English language, which led to a predominantly homogenous trial population, and efforts to expand this resource to include other languages should be prioritized. Given the suboptimal enrollment in this pilot RCT, future studies are needed to investigate the role of digital health coaching on self-efficacy and other PROs that are critical for patients with leukemia, especially as treatments in this field evolve, and more patients are cured and experience survivorship challenges from chronic toxicities. Acute leukemia diagnoses may be optimally targeted to provide the most benefit, especially during the early phase of their treatment when there is overwhelming new information, external stressors, and adaptation required in their lives.

An important point in determination of how digital health coaches may integrate into routine clinical practice would be the cost of such an intervention. While cost was beyond the scope of this pilot RCT, future studies should capture both the finances and time needed from trained personnel to run a digital health coaching program. While many of the presumed clinical benefits (e.g., self-efficacy, patient independence, perceived symptom burden) are intangible and difficult to monetarily quantify, efforts to minimize the cost of healthcare systems will be critical if such a tool were to be broadly implemented.

In conclusion, this is the first known study evaluating the utility of a digital health coach on self-efficacy after a new diagnosis of either AML or CLL. While the study did not meet its enrollment goals and was underpowered to detect significant differences from the intervention, there were numerical increases in CBI scores and other markers of self-efficacy as well as decreased symptom burden scores in the AML digital health coach intervention arm that merit further investigation. Now that the pandemic has become a more routine part of daily life, future studies will focus on maintaining adequate enrollment to achieve the power to detect if there are significant improvements in self-efficacy and PROs with digital health coaching in newly diagnosed leukemia.

Statement of Ethics

The study was approved by the Institutional Review Board at the University of Texas MD Anderson Cancer Center (2018-0806). Written informed consent was obtained from all participants to participate in the study.

Conflict of Interest Statement

C.D. is supported by the LLS Scholar in Clinical Research Award, receives research support (to the institution), and serves as a consultant (personal fees) for AbbVie. K.B. and A.B. are employed by Pack Health, A Quest Diagnostics Company, which provided the intervention for this study. All other authors have no disclosures or potential COIs to report. This study was presented in part as a poster at the ASCO Quality Care Symposium in 2023.

Funding Sources

This research was supported by the National Institutes of Health through M.D. Anderson’s Cancer Center Support Grant CA016672 as well as an NCI core grant from AbbVie.

Author Contributions

C.D. designed the study. J.M.-P. wrote the manuscript. V.S., K.B., and A.B. collected the data. B.F. did the statistical analysis. K.S.C., D.H., M.S., N.J., W.W., and A.F. treated patients on the study. All authors reviewed and approved the final version of the manuscript.

Funding Statement

This research was supported by the National Institutes of Health through M.D. Anderson’s Cancer Center Support Grant CA016672 as well as an NCI core grant from AbbVie.

Data Availability Statement

The study details are available online on ClinicalTrials.gov NCT04774744. The study data are not publicly available to respect participant confidentiality. Requests for sharing of deidentified data should be directed to the corresponding author.