This study used an explanatory sequential mixed methods design [25] consisting of surveys and semistructured interviews with a geographically defined sample recruited from both community and clinic settings. This study adhered to the SRQR (Standards for Reporting Qualitative Research) guidelines (the SRQR checklist is provided in Checklist 1).

This study was conducted with English- and Spanish-speaking individuals (age ≥18 years) and residence within the RedBird neighborhood of Dallas. We defined the “RedBird area” as comprising 15 ZIP Codes surrounding a community-based clinic affiliated with UT Southwestern Medical Center and Simmons Comprehensive Cancer Center, hereafter referred to as “RedBird Clinic.” The RedBird area in the Southern Sector of Dallas is known to be adversely affected by the digital divide, illustrated in Figure 1 [14]. Gallardo [14] calculated the Digital Divide Index by adding a composite socioeconomic score (age, poverty, lower than high school education, disability status, and internet-to-income ratio) with an infrastructure and adoption score (broadband access, computer devices, and download and/or upload speed).

We recruited survey participants through the RedBird Clinic, the Simmons Comprehensive Cancer Center (hereafter “Cancer Center”), and through local community partners including faith-based organizations, nonprofit organizations, and support groups. All organization partners were located in the RedBird neighborhood from which we recruited study participants, and all were dedicated to advancing the social, economic, and/or health needs of RedBird area residents. Cancer Center participants were sampled through the patient registry, limited by those with residential addresses within the 15 Zip Code area and having completed an appointment within the past 5 years.

To capture data from participants of varying levels of digital literacy, we deployed multiple survey recruitment and data collection methods in accordance with each community partner’s preference. Ultimately, most surveys were completed either on paper or through REDCap (Research Electronic Data Capture) [26], a web-based data collection and management tool. Recruitment strategies included mailing paper surveys, sending a link for distribution in the organization’s newsletter, distributing flyers with a QR code and a phone number, and in-person recruitment. Patients of the Cancer Center who had portal accounts received a survey invitation link through the portal, while nonportal account holders received a postcard with multiple options for survey administration. In all formats, the first question of the survey explained the voluntary nature of participation, and that submission of survey responses indicated consent to participate.

The survey included questions about technology access, health literacy, information seeking and literacy, and portal and telehealth exposure. Descriptive statistics were used to summarize participant characteristics and survey responses. Next, to identify factors associated with portal and telehealth use, we compared their use by demographic characteristics and by health promotion behaviors using bivariate Fisher exact tests.

The survey also contained 6 open-ended questions focusing on portal and telehealth use, preferences, and experiences. We included questions related to both the portal and telehealth, going beyond the study by O’Connor et al [24] that focused on telehealth alone. Portal use is critical to study because portals are used by a greater share of patients than telehealth, and increasingly, health care systems use the portal as the gateway to their telehealth application [27,28]. Open-ended survey text data were analyzed using axial coding techniques, and thematic findings were used to refine the semistructured interview guide [29].

Based on survey findings, we created a semistructured guide to probe interview participants’ perceptions and experiences with the portal and telehealth, including factors that motivated their enrollment and use. We also added questions about the role of care partner support and experiences related to accessing test results, given emergent findings from the open-ended survey questions that these factors were associated with technology uptake and use (Multimedia Appendix 1).

We recruited interview participants from a subsample of survey completers who expressed a willingness to be contacted for follow-up interviews. We stratified interview sampling to recruit half of the participants from the survey cohort recruited through the Cancer Center and half through various RedBird area community organizations. Within each group, we prioritized recruitment of participants whose responses to the open-ended survey questions were more robust, presuming that this might correlate with greater engagement in the interviews. As we accrued participants, we monitored the demographics of our sample to ensure that we achieved diversity by gender and race or ethnicity that reflected the RedBird area population. We made up to 3 attempts to reach each participant via their preferred mode of contact. During recruitment, we also verified whether the individual had “consistent” or “mostly reliable” access to the web and/or access to a smartphone to support our focus on individuals with potential access to the portal and telehealth. That is, we chose to focus on those without device or Wi-Fi access barriers because the goal of our inquiry was to identify potential opportunities for educational interventions, such as barriers due to misperceptions, attitudes, or limited skills. Furthermore, device affordability and Wi-Fi infrastructure were improving under the Affordable Connectivity Act and broadband expansion by leading telecommunications companies.

A trained qualitative interviewer (ER) conducted interviews by phone using the semistructured interview guide. Interviews lasted an average of 30 minutes (range 16‐61 minutes). Interviewers conducted rapid data analysis [30,31] during interview data collection to allow for immediate exploration of emerging patterns and insights. Rapid data analysis also serves as a tool to enhance communication among members of the interview team, facilitate monitoring effectiveness of the interview guide and determination of thematic saturation, and promote consistency of data capture.

Concurrently, we also conducted an in-depth qualitative analysis of interview transcripts. All transcript data were thematically coded by trained qualitative researchers using NVivo (Lumivero). The lead qualitative investigator (RTH) drafted an initial codebook deductively driven by concepts in the interview guide and inclusive of emergent findings extracted from the rapid analysis. Next, 2 qualitatively trained staff (ER and AG) engaged in open coding of 15% of the dataset (n=3 transcripts), iteratively refining the codes and definitions to match themes and patterns. They tested the codebook by independently coding an additional 10% of the dataset (n=2 transcripts) and then met to compare codes and calculate an interrater reliability score using Cohen κ. Because the rate of concordance exceeded 0.75, indicating substantial agreement, the remaining transcripts (n=15) were divided between the coders and analyzed independently. The research team met weekly to discuss and interpret findings. None of the qualitatively trained researchers engaged in data collection or analysis were physicians or had clinical relationships with participants.

All study-affiliated personnel have current certification in the ethical conduct of human subjects research on file with the University of Texas Southwestern Medical Center (UTSW)’s Human Research Protection Program. We obtained informed verbal consent for participation and audio recording in accordance with the approved study protocol from the institutional review board at UTSW (STU-2023‐0469). All audio recordings were transcribed verbatim and deidentified by a professional transcription vendor contracted with UTSW under a Health Insurance Portability and Accountability Act waiver. Participants were mailed a $30 gift card in appreciation for their participation.

We distributed 1262 total surveys via (1) digital methods: 646 were distributed and 64 completed, and (2) paper: 616 were distributed and 118 completed. Characteristics of the 182 total participants (182/1262, 14.4% overall response rate) did not differ by survey modality (data not shown, chi-square P value >.05) (Table 1). Of 182 survey participants, most were older (n=109, 66%; age ≥60 years), African American or Black (n=120, 66%), and women (n=142, 79%). Survey participants reported experience with health technology: a little more than half (97/182, 54%) had completed a telehealth appointment, and a majority (131/182, 72%) had used a patient portal.

For interviews, we contacted 59 individuals, 3 declined to participate (reason cited was “not interested”), 7 were ineligible due to lack of a Wi-Fi connection or a smartphone, and 29 did not respond to ≤3 recruitment attempts. Of the 20 semistructured interview participants, most were older (n=14, 70%; age ≥60 years) and 40% (n=8) were men. We had attempted to diversify our interview sample with individuals of varying levels of portal and telehealth experience; however, due to lack of recruitment response among portal and/or telehealth nonusers, our interview sample included a greater proportion of those with greater technology experience (Multimedia Appendix 2). Nearly all interview participants had used the portal at least once (18/20, 90%), and a majority completed at least 1 telehealth appointment (12/20, 60%).

As shown in Table 2, more than half of our sample used both the portal and telehealth (91/180, 51%), and nearly a quarter (43/180, 24%) reported not using either technology; the remainder used 1 technology or the other. We combined those using only 1 technology or the other due to a very small number of patients reporting use of telehealth but no portal use (n=6). Compared with technology users, those reporting no use of portal or telehealth had a lower level of education (P<.001) or were more likely to be Spanish speaking (P<.011).

Table 3 illustrates frequency of endorsement of each statement about using the web for health promotion activities. The most commonly endorsed statement was “I can use my computer or mobile phone to communicate about my health needs” (81.5%). The percent agreeing to the statements varied statistically significantly by portal and/or telehealth (P<.001), with greater agreement among those using portal and telehealth use.

We found evidence supporting 4 key factors by O’Connor et al [24]: improving individuals’ health autonomy, integrating use of digital health technology into individuals’ daily lives, obtaining recommendations from trusted individuals, and appreciating the value of digital health technology.

In this iterative, mixed methods study, we describe attitudes toward and experiences with telehealth and the portal among a minoritized population in Dallas, Texas. We identify 7 factors that are key to the uptake and use of digital health technologies. Four of these factors had previously been identified by the meta-analysis of qualitative studies by O’Connor et al [24]: improving health autonomy, integrating digital health technology into individuals’ daily lives, receiving technology recommendations from trusted individuals, and appreciating the value of digital health technologies. The 3 additional factors we identified in this study—enlisting support from care partners, managing severe or chronic illness, and accessing test results rapidly—are valuable contributions because they provide insight into potential intervention opportunities among specific populations and clinical settings. Survey findings also align with the factors by O’Connor et al [24] that are key to telehealth utilization in underserved populations (improving health autonomy, integrating use of digital health technology into one’s daily life, and appreciating the value of digital health technology), as well as other factors we know from the literature to be relevant to successful engagement with the portal or telehealth (eg, health literacy, confidence using technology, and overall technology access). Our finding that language and level of education are associated with nonusers of the portal and telehealth underscores the importance of making the portal and telehealth accessible to populations that are Spanish-speaking and, among English speakers, with lower literacy levels. Notably, although a Spanish version of the patient portal is available in some health care systems, multiple challenges have dampened enthusiasm, uptake, and use of it [32].

Support from care partners has been described as a key facilitator for technology access among other populations, especially older individuals and those with chronic illness or disabilities [33,34]. To our knowledge, this is the first study to qualitatively describe care partner support as a key facilitator of digital health technology engagement among minoritized populations. Prior work suggests that care partners value having access to patient information and report benefits related to greater understanding of medications and follow-up steps and increased feelings of control [35]. Care partner assistance with patient portals is more common when patients are older, have limited English proficiency or lower overall literacy, and greater illness severity [36]. While our sample was older, our findings around care partner support as a facilitator suggest that care partners may also play a more prominent role among lower-resource, minoritized populations.

Second, our finding about the value placed on rapid access to test results has far-ranging implications across clinical settings and populations. Participants ranked rapid access to this information as a key to motivating portal use. For example, individuals recruited from the Cancer Center expressed mostly positive and some negative opinions of rapid access to test results. Specifically, a few participants felt that patients should not be given a cancer diagnosis via the portal, but that access to laboratory results and appointments was very important. At the same time, individuals recruited from the community (who presumably were less likely to be in active cancer treatment) also felt strongly that rapid access to appointments, medications, and the opportunity to communicate with their providers was a strong motivator for portal use. While we did not ask about community participants’ medical conditions, this finding may reflect higher rates of chronic illness in lower-resource populations, such as diabetes and hypertension, which also involve regular appointments and coordination across multiple providers.

The 21st Century Cures Act [37], which functionally required that institutions immediately make available to patients, on request, information in the electronic health record, has spurred debate since it was passed in 2021 [38,39]. Most health care organizations choose to meet this regulatory requirement by making results automatically viewable on the patient portal as soon as they are available in electronic health records, meaning patients are often able to see results before speaking with a clinician. Consequently, patients and families living with cancer will likely receive “nonnormal” results with some frequency as they use the portal to review data, laboratory test results, and imaging. There is controversy about providing patients with immediate access to test results in part because this information can be conveyed prior to the patient speaking with their clinician [38,40]. The potential positive and negative impacts of immediate test result access are likely to vary immensely across individuals in both patient and clinician groups [41,42]. Health care organizations may wish to provide additional education and support to patients and care partners, particularly around how to configure portal settings to either receive or mute new test result notifications. Our findings showing some complex preferences about information provision among our participants suggest that a “one-size-fits-all” approach will not be sufficient.

Digital inclusion, or “[ensuring] that all individuals and communities, including the most disadvantaged, have access to and use of digital tools,” [43] must be emphasized in order to overcome the existing digital divide [13]. With tools such as Gallego’s Digital Divide Index and the Centers for Disease Control and Prevention’s social vulnerability index [44], among others, social epidemiologists have provided ample evidence of where behavioral interventionists should implement more intensive interventions to maximize use of resources to improve health outcomes [45]. Additionally, intervention strategies shown to be effective in addressing digital health equity have emphasized user design (eg, compatibility with existing devices and culturally appropriate content), supportive infrastructure (eg, assistance with devices and connectivity), and educational support (ie, from professionals, family, or friends) [46]. Our findings of the importance of care partners, rapid access to test results, and coordinated care for individuals with more complex disease contribute to this shared understanding of what, where, and to whom digital health equity interventions should be applied. And as a next step, using an implementation science framework such as Glasgow’s RE-AIM (1999) may help us optimize how the intervention is applied [47].

Finally, it is important to recognize that providers and health care systems contribute to the persistence of digital health disparities. Multiple studies using data from the Health Information National Trends Survey have documented bias in provider recommendations for portal and telehealth use, with greater encouragement directed toward patients who are female, White, and college-educated [48-50]. Other Health Information National Trends Survey–based studies have found that provider recommendation is a key driver of telehealth use [51] and that adults with limited English proficiency are more likely than those proficient in English to seek health information from professionals [52]. Notably, 1 study found that Spanish-preferring participants placed greater trust in information from government agencies and often turned to web-based videos for health-related information [52]. These findings illustrate the importance of educating providers and health care institutions about how provider bias can exacerbate the digital divide in health care access and disseminating reliable health information through multiple channels.

Although significant progress has been achieved in strengthening digital health access and use among underserved communities [53], several barriers remain. These include lower educational attainment [54], lack of awareness of the benefits of digital health technologies [55-57], mistrust in the health care system evidenced as concern for data privacy [58,59], and ambivalence toward technology [60]. These barriers often exacerbate existing health disparities that affect the same minoritized populations with multiple chronic conditions, who stand to benefit greatly from portal and telehealth access. Future implementation strategies to address the persistent digital divide will need to provide more personalized options, be supported by robust financial and human resources, and include trust-building with cultural minority communities [60].

With only 20 interviews, in a mostly older female Non-Hispanic Black or African American population, it is not possible to generalize these findings broadly. We also chose to focus our interviews on individuals who did not have barriers in access to Wi-Fi or smartphone technology in order to delve more deeply into potential intervention opportunities that could be addressed through education versus systemic infrastructure barriers over which we had limited control. While the infrastructure for Wi-Fi is increasing, and cellular phone ownership exceeds 90%, even among low-income populations [61], this sampling strategy resulted in the exclusion of the perspectives of a more diverse range of participants, which limits generalizability and precludes a more comprehensive understanding of the issues. Additionally, while we identified care partners as an important facilitator and explored this topic in depth in interviews, we missed an opportunity to obtain quantifiable data on care partner demographics and technology-related descriptors of individuals acting as care partners by not including such questions in the survey. This would have enabled us to contribute to the literature on the demographics, technology-related skills, and attitudes of people who use technology as care partners versus those who need care partner assistance. Understanding the relationship between care providers and receivers may facilitate conceptualization of new hypotheses and strategies for interventions.

However, our study has several strengths. Our finding that the uptake and use of technology may be related to chronicity and severity of illness merits attention. While perhaps those with a more life-threatening condition may be more acutely motivated to engage in their health care, findings from this study suggest that those with chronic conditions such as diabetes, certain cancers, and hypertension may also be persuaded to use the portal to monitor their test results. Also, our study contributes uniquely to the literature by focusing both quantitatively and qualitatively on the perspectives of a multiracial, minoritized population toward digital health technologies. Our finding that language and level of education are associated with lower portal and telehealth use is useful for informing future intervention efforts. Finally, future research may be enhanced by examining behavioral outcomes, which may reveal insights regarding the possible associations between digital health interventions, health care utilization, and patient outcomes.

We contribute evidence that seven key factors are relevant to the development of strategies to promote digital inclusion in future interventions for underserved populations: (1) improving patient autonomy, (2) integrating digital health technology into daily life, (3) receiving recommendations from trusted individuals, (4) appreciating the value of the technology [24], (5) enlisting support from care partners, (6) managing severe or chronic illness, and (7) accessing test results rapidly. Future studies should focus on making the portal and telehealth more accessible to English-speaking populations of lower educational attainment and Spanish speakers and should address how different clinical settings and health conditions affect the uptake of the patient portal and telehealth.