Approximately one-half of patients with a cancer diagnosis experience financial toxicity (FT), a term used to describe the negative impact that medical expenses can have on patients and their families [1-4]. Out-of-pocket (OOP) expenses, such as medical copays, travel and lodging costs, and patient and caregiver lost wages, contribute to the high FT experienced by patients with cancer. These costs also pose a significant barrier to participating in cancer clinical trials (CCTs). In 1 study, 48% of patients in phase 1 trials reported monthly OOP costs of at least US $1000 [5]. Patients with cancer experience higher OOP costs compared to other chronic diseases [6] because of a higher frequency of visits related to a trial’s protocol and need for lodging for observation [7], which poses a barrier to CCT enrollment [8].

To address the high OOP costs in CCT and attempt to mitigate the FT associated with CCTs, the Lazarex Cancer Foundation created the Improving Patient Access to Cancer Clinical Trials (IMPACT) Program in 2018. The IMPACT Program is a financial reimbursement program (FRP) for OOP travel and lodging expenses associated with CCT participation. Prior studies have demonstrated that IMPACT sites saw higher rates of patient enrollment in CCTs and increased participation among underrepresented groups, including racial and ethnic minorities and patients from lower socioeconomic backgrounds [9]. However, despite initial positive results, a subsequent study assessing the effectiveness of the program revealed that more than 40% of CCT participants offered the IMPACT program did not enroll, suggesting the presence of persistent barriers [10].

As a result, IMPACT 1.0, a precursor to this study, was launched to better understand the barriers to and facilitators of enrolling in the Lazarex IMPACT program at the Abramson Cancer Center. IMPACT 1.0 involved enrollment in the FRP via a telephone call with an IMPACT coordinator. The initial study uncovered several patient barriers to enrolling in the IMPACT program: the CCT team did not mention the IMPACT FRP to patients, leading to a lack of awareness, and when they did, it was often mid or late in the study, missing opportunities for reimbursement of expenses and difficulties connecting with patients via phone [11]. In addition, the program poses a large administrative burden on the IMPACT team and clinical trial (CT) coordinators. This was particularly due to repeated unsuccessful attempts to contact patients via telephone and email to offer the program, which was not only inefficient but also delayed the enrollment process and potentially discouraged program and CCT participation. These findings highlight the need for a more streamlined, accessible, and timely approach to offering and enrolling patients in the IMPACT FRP.

This study leveraged novel digital technology to address these barriers and evaluate the feasibility and acceptability of using an automated text-based system to systematically offer, screen, and enroll patients in the IMPACT program via Way to Health (W2H), a web-based platform created by the Penn Medicine Center for Health Care Innovation and Penn Center for Health Incentives and Behavioral Economics. Text messaging is increasingly being used for communication between health care sites and patients to coordinate care as well as in medical research where it has been used to remind patients about study tasks [12]. The automated text-messaging system was designed to provide a systematic way of offering patients the IMPACT program. It allowed patients to self-screen for the program’s household income–based eligibility criteria and provide flexibility for enrollment based on patient preference and need by allowing patients to complete the application independently via a text or email link to the application or via an IMPACT coordinator telephone call.

Our objective was to determine the feasibility, patient acceptability, and satisfaction of automated text messaging to offer, screen, and enroll patients in the IMPACT FRP. In addition, we hypothesized that direct texts to patients to offer, screen, and enroll in the FRP would reduce administrative burden on the IMPACT team and Lazarex Cancer Foundation and ensure that all eligible patients, regardless of perceived need, are offered the opportunity to participate in the IMPACT program. A secondary objective was to identify the barriers to and facilitators of using the text-based system and subsequently enrolling in the IMPACT program.

This study used a mixed methods approach using an explanatory sequential design. Quantitative data were collected through engagement metrics, including text response rates and enrollment rates, as well as patient-reported satisfaction scores. Program enrollment rates were used to determine feasibility, whereas the engagement metrics were used to measure the acceptability of the program. We then conducted semistructured interviews with a subsample of patients who responded to at least one of the FRP texts and agreed to be interviewed to determine the barriers to and facilitators of enrolling in the IMPACT program via text, perceived advantages and disadvantages of the text messaging program compared to a phone call, and overall feedback on the acceptability of the automated text messaging program. The qualitative data were used to better understand the quantitative data and inform implementation.

Patients eligible to receive texts regarding the FRP consisted of a subgroup of patients enrolled in cancer treatment CTs, specifically breast, leukemia, and chimeric antigen receptor T-cell (CAR-T)–treated neoplasms (lymphoma, glioblastoma, prostate, and pancreatic adenocarcinoma), at the Abramson Cancer Center of the University of Pennsylvania, Philadelphia, between March 26, 2024, and July 31, 2024. The 3 CT teams from which we received patient referrals were chosen by the IMPACT study team based on IMPACT 1.0 referral and enrollment trends and diversity of enrolled patients. Trial teams’ willingness to participate was also taken into account.

CCT participants were referred by their CT coordinators to the IMPACT study team once weekly, typically at the end of the week. To establish a consistent protocol, we asked CT teams to refer all patients to IMPACT, regardless of perceived or known financial need. Additionally, we asked teams to notify patients about the IMPACT program and upcoming texts as well as provide our study flyer at the time of their CCT informed consent. As early as possible during the following week, the IMPACT coordinator would add patients to a list within the electronic medical record, linked to the W2H platform by a query. The query automatically enrolled patients in the W2H automated text messages and sent them the first text.

The texts (Multimedia Appendix 1) required participants to verify (1) their identity, (2) enrollment or consideration of enrollment in a cancer treatment CT, and (3) that they met FRP income eligibility by texting their household size and reviewing the upper limit of income for a household of that size. If they met the eligibility criteria, then patients were offered enrollment via an online application link sent via text or email or via a telephone call with the IMPACT coordinator (Multimedia Appendix 2).

Eligible participants were invited to apply to the Lazarex IMPACT FRP. Those who completed the text conversation subsequently received a Net Promoter Score (NPS) survey through text as the last question. Thus, we collected satisfaction data from this smaller subgroup of participants who completed the entire text-based conversation. Those who did not complete the text-based conversation and who participated in an interview completed the NPS survey administered as the last question of the interview process.

Patients who engaged with the automated text messages were invited to participate in an approximately 10-minute semistructured interview via telephone regarding their experience in exchange for a US $25 Consumer Value Store (CVS) Gift Card. Patients were only invited to interview once they met the following conditions: (1) disqualified for the FRP via text and (2) opted out of text conversation (responded “BYE”), (3) successfully received and submitted an application for the FRP or (4) after the conversation closed, following nonresponse.

All patients who were interested in completing the semistructured interview were read a statement of research before completion of the interview. Interviews were conducted by the IMPACT coordinator (AS) between April 2024 and October 2024.

Acceptability was defined as the extent to which users considered the automated text appropriate, as measured by engagement rates and satisfaction items, two constructs of the Theoretical Framework for Acceptability [13]. Acceptability of FRP automated texts was assessed by measuring engagement rates with the texts at each point in the text conversation (offering, screening, and enrollment), whereas feasibility was evaluated based on FRP program enrollment rates. Feasibility was defined as the proportion of eligible patients completing enrollment in the IMPACT FRP via the text-based workflow, reflecting the intervention’s practicality and adoptability under real-world constraints.

The NPS was used to measure patient satisfaction with the automated text messages. The NPS survey allowed respondents to rate their likelihood of recommending the text conversation on a scale of 1 to 10, with zero being completely unlikely and 10 being extremely likely [14]. To calculate the NPS, the percentage of detractors is subtracted from the percentage of promoters, where scores of 0 to 6 are detractors, 7 to 8 are passives, and 9 to 10 are promoters. The NPS can fall between −100 and +100 [14]. An NPS of >0 is considered “good,” >50 is considered “excellent,” and >70 is considered “best in class” [15]. Our version of the NPS was as follows: “Using a 0‐10 scale: How likely is it that you would recommend using this text-based platform to a friend or colleague?”

Interviews were recorded and transcribed verbatim and subsequently imported into NVivo 12.0 (QSR International). They were conducted, read, and coded independently by 2 investigators (AS and HP) and then coded jointly using consensus conferences to resolve coding disagreements. Thematic analysis was used to identify themes and subthemes. The interviews were performed and analyzed until thematic saturation—the point at which additional interviews did not yield new themes—was reached [16].

This study was approved by the University of Pennsylvania’s Institutional Review Board (protocol 854446). All participants who received a text were provided the opportunity to opt out of the text conversation. Interview participants gave informed consent prior to the recorded interview. No waiver of consent was requested or granted for primary data collection or analyses. Secondary analyses performed were conducted under original institutional review board protocol, which allows for these analyses without additional participant consent. All study data were treated as confidential and stored in secure systems only accessible to authorized research team members. Identifying information was removed, and data presented in this manuscript were deidentified, ensuring individual participant identities were protected throughout all phases of the research process. Qualitative interview participants were compensated for their time and insights with a US $25 CVS gift card, as disclosed in both the informed consent process and study protocol.

As shown in Figure 1, the study sample consisted of 51 CT participants who successfully received a text message regarding the IMPACT Program. Interviewed patients consisted of a convenience subsample of 28 CT participants who successfully received at least one of the IMPACT program study enrollment text messages and consented to the interview.

Table 1 presents the demographic and trial participation data for the sample who received text messages and the subsample of patients who completed interviews. When comparing the 2 samples, the median and average age and representation by sex, race, and ethnicity distribution were similar. The interview sample had slightly higher breast and CAR-T trial participants. The overall sample of patients who received program texts (n=51) had slightly higher leukemia trial participants than the subsample who were interviewed (n=28).

A total of 28 interviews were conducted, lasting between 3 and 20 minutes. On average, patients completed the interview within approximately a month (range 0‐95 days) after the close of their text conversation.

Patient interview data revealed that only 46.4% (13/28) of patients were notified by the CT team about the IMPACT program and texts at the time of informed consent. Among the remaining patients, 14 (50%) learned about the IMPACT program at the time they received the first text, and 1 (3.5%) indicated they were unsure when they learned about the IMPACT program. As shown in Figure 1, of the 77 patients who consented to CCTs across the 3 research teams, only 51 (66%) patients were referred for the text conversation. Nearly one-third of eligible patients were not referred, which reflects a combination of system-level and staff workflow barriers, such as inconsistent protocol adherence.

Finally, we asked patients what advantages and disadvantages they saw to offering the program through texts compared to offering the program through an IMPACT coordinator over a live phone call (Table 3). Approximately one-third of interviewed patients reported the text conversation as being quicker and more efficient than enrollment in the FRP by telephone call.

This study showed that using an automated text messaging platform to offer, screen, and enroll patients in the IMPACT FRP is feasible and acceptable. To our knowledge, this is the first study to leverage text messaging to engage individuals in CCTs in an FRP. There was high engagement with texts (49/51, 96.1%), screening for the FRP (51%), FRP enrollment rate among patients deemed eligible through text (9/17, 52.9%), as well as overall enrollment rate (16/24, 66.7%), excellent degree of patient satisfaction (NPS=51), and high comfort with technology (78.6% strongly agreed to comfort with technology). In addition, one-third of all patients who received an application via text required no help from the IMPACT coordinator to complete the application. The use of the automated text messaging system significantly reduced the time spent contacting patients because initial engagement with the texts led to quicker responses from patients via telephone call. Moreover, almost one-third of interviewed patients reported the text conversation as being quicker and more efficient than enrollment in the FRP by telephone call.

We identified barriers and facilitators at the system, individual, and implementation level that could inform program designs to increase participation. The most frequently cited system facilitator was the assistance provided by the IMPACT coordinator, highlighting the importance of having a knowledgeable and responsive contact to complete enrollment for those who opt out of texts and address questions that arise. Another facilitator included communication from the CT teams regarding the texts, which led to a decrease in skepticism and an increase in the likelihood of response to the text messages. Conversely, the most cited system barrier was related to the lack of communication from the CT team. Although CT teams were asked to notify patients of the FRP texts, as well as hand out our study flyer, only 46.4% (13/28) of interview patients were notified by the CT team that they would receive FRP texts, highlighting the persistence of a barrier from IMPACT 1.0: lack of patient awareness of the FRP due to inconsistent mentions from CT team members. Moreover, approximately one-third of all patients potentially eligible for the FRP were not referred to the IMPACT coordinator for enrollment in the automated texts, and thus, this is an important barrier to the implementation of this program.

One possible way to ensure that patients are informed by their teams about the FRP texts is to include referral to the program as part of the standard operating procedures in protocols and informed consent forms to ensure all patients are notified of the program and will anticipate receiving the first text. This strategy may also reduce text authenticity concerns, which were a significant patient-reported barrier within the individual domain. In addition, inclusion of the program website link or requiring a 2-factor authentication or a unique patient identifier in the initial text could help increase the legitimacy of the text messages.

Another barrier to implementation is related to the most cited barrier: “limited information via text.” Owing to the Health Insurance Portability and Accountability Act laws, the texts could not contain personal health information and, therefore, could not mention CT information or even the word “cancer.” Omission of this information caused at least one patient to opt out of the text conversation. To address this issue, it may be helpful to create a text introduction with a layered information approach. Through this approach, basic program information would be included via text while also including one or multiple URL links, which navigate a patient to a more detailed resource or website. In addition, we plan to address language barriers in future deployments of the automated text system by integrating multilingual capabilities. This includes developing and implementing message libraries in the most common languages spoken by the target patient population, starting with Spanish but expanding to others as identified through community needs assessment.

The prior literature has demonstrated that text-based communication can be used in lieu of clinic visits for symptom monitoring, which can then reduce clinic wait time before chemotherapy [18] and increase access to supportive care (such as physical activity motivation and medication reminders to promote adherence). A systematic review demonstrated that text-based messaging improved the patient experience, but outcomes were mixed in part because of the heterogeneity in study design, measured outcomes, and threat of bias [19]. This study adds to the body of literature showing that text messaging is an acceptable way to communicate and engage patients in a variety of health-related behaviors. Text-based communication is more accessible to patients than online electronic medical record portals: 97% of Americans state they communicate via text, whereas only 65% of Americans who were offered an online patient portal accessed it [20].

The limitations of this study include a relatively small sample size, with participants from a single institution and conducted with English-speaking populations, which may reduce the generalizability. The demographic composition of the sample, while diverse, may not fully represent all patient populations with cancer. The interview process may have been subject to recall bias as well, as some participants were interviewed weeks after their text interaction. Finally, participants with low technology literacy may not have responded to the texts, introducing the possibility of selection bias.

Despite these limitations, the study is among the first to introduce and evaluate a text-messaging system to introduce, screen, and enroll patients in an FRP. The findings demonstrate digital platforms such as W2H can be leveraged to develop automated text messages that increase awareness, screen, and enroll patients in FRPs. W2H can be deployed at other institutions, and thus, this program can be scaled. Automated text-based messaging provides a potential low-cost opportunity to scale the program to other institutions. Future directions include conducting cost-effectiveness analysis to inform dissemination, implementation, and policy.