Is Anybody Out There? Understanding the Reach of Lung Cancer Screening Health Communication Through EHR Patient Portal Authors: ANJALI VASAVADA, MATTHEW TRIPLETTE, LORELLA PALAZZO, JAMES RALSTON, LISA CARTER-BAWA, CASEY LUCE, BEVERLY B. GREEN, HONGYUAN GAO, MELISSA L. ANDERSON, YU-RU SU, KRISTINE ROGERS, KAREN J. WERNLI Published: 2026 ABSTRACT Health communication addressing individuals eligible for lung cancer screening (LCS) can increase screening uptake and adherence, which remain low in the U.S. Digital Communication via EHR patient portals offers a scalable solution, but face barriers related to portal access and use. We conducted a secondary analysis of a pragmatic clinical trial (Kaiser Permanente Washington, Nov 2022–Apr 2024) to identify gaps in access and engagement with an LCS health communication intervention delivered via portal message. We assessed factors associated with 1) patient portal access (all participants, n = 1837); and 2) opening the intervention message (intervention arm only, n = 820) using multivariate log binomial regression to calculate prevalence ratios (PRs) and confidence intervals (CIs). Most participants (89.6%) had access to the portal. No access was associated with current tobacco use (PR 0.4), neighborhood income <$50K (PR 0.3), and <5 years of insurance (PR 0.5). Among the intervention group, 81.2% opened the message. Not opening was associated with older age (PR 0.7), current tobacco use (PR 0.5), >1 well-care visit in the past year (PR 0.6), and 5–10 years of insurance (PR 0.5). While patient portals can broadly deliver LCS interventions, disparities in access and engagement suggest the need for alternative communication strategies. Digital health and electronic health record (EHR)-linked patient portals have become central to health care in the United States (US), offering an efficient and accessible way for many patients to communicate with their providers and improve engagement with their care (Assiri, 2022; Tapuria et al., 2021; Vest, 2012). Broadly, digital health is defined as the use of information and communication technologies in healthcare to promote wellness, including, but not limited to, the use of wearable devices, mobile health, and telehealth (Ronquillo et al., 2025). Integrating digital health into patient care has the potential to improve patient-centeredness along the health care continuum, ensuring that care is responsive to individual patient preferences, needs and values (Edgman-Levitan & Schoenbaum, 2021). Digital health approaches, such as bidirectional communication through the patient portal, increase patient reach, accessibility, and education by supplementing the limited time patients have during face-to-face visits with clinicians (Charalambous, 2019). Health communication interventions that utilize digital channels, such as text messaging, mobile applications, or patient portals have demonstrated effectiveness to improve patient-centered outcomes (e.g., knowledge) and increase cancer screening participation in populations eligible for breast, cervical and colorectal cancer screening (Champion et al., 2020; Hirst et al., 2017; Lee et al., 2014, 2017). However, few studies employing digital health communication interventions have focused on adults eligible for lung cancer screening (LCS) (Blake et al., 2020). Despite high mortality rates of lung cancer in the U.S. (Cancer Stat Facts, n.d.), adherence to guideline-recommended annual LCS is low, even among those who have previously engaged in screening, ranging from 22% to 55% nationally (Kim et al., 2025; Lin et al., 2022; Lopez-Olivo et al., 2020; Sakoda et al., 2021). While there is potential for digital health interventions to increase LCS adherence through patient education and communication, barriers to portal access and use can undermine the effectiveness and reach of interventions embedded in the patient portal. Disparities in EHR use and patient portal enrollment among patients in oncology settings have been explored in the past, with sociodemographic patient-level factors such as rurality, age, and racial/ethnic background most frequently reported as associated with lower odds of portal access (Griffin et al., 2023; Luoh et al., 2021; Sinha et al., 2021). This study aimed to identify gaps in access to the patient portal and engagement with an LCS health communication intervention delivered via the patient portal by factors at the patient-, clinician- and system-level. The overarching goal of this study was to understand which LCS-eligible patients are potentially missed by digital health communications, to improve the reach of future health communication interventions for this population. Materials and Methods Study Setting and Population The study population for this analysis included participants in the Larch Trial, a pragmatic clinical trial testing two patient-centered interventions to improve annual repeat LCS. The details and protocol for the trial are described elsewhere (Luce et al., 2024). Trial participants were members with a paneled primary care physician from Kaiser Permanente Washington (KPWA), an integrated delivery system that provides insurance and serves ~600,000 members across Washington State (About Kaiser Permanente Washington, 2024). Trial participants were aged 50–78 years with English or Spanish as a primary language. Participants were identified for the trial with a LCS low-dose computed tomography (LDCT) scan with normal or benign findings (LungRADS 1 or 2), hereafter known as the “index LDCT.” Participants were excluded if they had previous lung cancer diagnosis, missing tobacco history, needed a language interpreter (except Spanish), or if they expressed not wanting to be contacted for research. The study was reviewed and approved by the Kaiser Permanente Interregional Institutional Review Board (IRB) (#1912783). The IRB waived informed consent of participants in the pragmatic trial, and the trial was registered at ClinicalTrials.gov prior to any patient enrollment (ClinicalTrials NCT05747443). Health Communication Intervention Participants were randomized to the LCS health communication intervention or usual care. The Patient Voices Video is a health communication intervention that emphasizes the importance of returning for annual LCS despite normal findings. The development and design of the Patient Voices Video has been described previously (Luce et al., 2024). The intervention included print and video content with key messages reinforced in both modes. Three key messages were: a) Participating in LCS is an important step in taking care of your health by finding lung cancer early (i.e., to normalize LCS); b) You are due for your next lung scan in 12 months (i.e., information when due); and c) Talking with friends and family about LCS might be helpful in feeling supported that you are doing something positive for your health (i.e., social support). About 3 weeks after the index LDCT, the health communication intervention was delivered through secure patient portal message (if enrolled) or by USPS (if not enrolled on portal). One week after the first communication, all participants were mailed a reminder USPS letter with same key messages and video link and QR code. Key Variables Patient-level variables obtained through the EHR at time of trial identification at baseline, including: age (binary; 50–64, 65+), race (categorical; White, Black or African American, Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, more than one race, other, prefer not to answer), ethnicity (categorical; Hispanic/Latino, Non-Hispanic/Latino, Prefer Not to Answer), sex (binary; male, female), insurance type (categorical; Medicare, commercial, state subsidized, or private pay), tobacco use (binary; current user, former user), tobacco history (continuous; pack-years), coronary artery disease (CAD) diagnosis (binary; yes, no), chronic obstructive pulmonary disease (COPD) diagnosis (binary; yes, no), and Elixhauser comorbidity score, a measure that quantifies the presence of specific comorbid conditions based on hospital administrative data (Yurkovich et al., 2015) (categorical; 0, 1–4, 5+). Variables estimated from census-tract data were residence (binary; urban, rural), area deprivation index (ADI) (binary; less than state median, greater than state median), and median family income (categorical; 0–49,999, 50,000–74,999, 75,000–99,999, 100,000+). Clinician-level variables obtained from the EHR were number of well-care visits with a primary care physician (PCP) in the last 12 months (binary; 0, 1+), LCS prior to the index LCS LDCT (binary; 0, 1+), department of LCS ordering clinician (categorical; family medicine, pulmonology, internal medicine), PCP clinic location (binary; Eastern WA, Western WA), and PCP practice co-located with radiology for imaging (binary; yes, no). System-level variables obtained from the EHR were length of enrollment in the KPWA system (categorical; <5 years, 5–10 years, 10+ years). Variables estimated from census-tract data were time to travel to facility (binary; <15 min, ≥15 min). Outcomes We measured two portal outcomes associated with access and reach to LCS interventions: 1) patient portal access, defined as having an active account in the KPWA Patient Portal (MyChart) system and a log-in attempt in the last 365 days, measured in all trial participants (n = 1,837); and 2) portal engagement with the health communication intervention delivered through the patient portal, defined as opening the intervention message within 180 days, restricted to those with portal access (n = 820) (Figure 1). Statistical Analysis For both outcomes, we first ran descriptive analyses and conducted univariate log binomial regression for each variable. For selection of variables as covariates in the minimally adjusted model, we initially generated a fully adjusted model with all of the previously mentioned covariates. Guided by a conceptual model and results from the fully adjusted model, we used a stepwise approach to determine the minimally adjusted multivariate model. We employed one-way ANOVA to assess the significance of each predictor, adding them to the model only if they significantly improved the fit (p < .05). The final multivariate models for both outcomes adjusted for age, sex, insurance, tobacco use, tobacco history, Elixhauser comorbidity score, ADI, median family income, well-care visits, previous LCS, clinic location, time to travel to facility, and length of enrollment. Adjusted multivariate estimates and standard errors were exponentiated to obtain prevalence ratios (PR) and 95% confidence intervals (CI). Statistical analyses were performed on R Statistical Version 4.4.1. Reported p-values are 2-sided, with 0.05 as the significance threshold. Results Access to the Patient Portal Among all trial participants (n = 1,837), 89% were enrolled and active on the patient portal, with a median age of 67 years. There were no notable differences between those enrolled and those not enrolled on patient portal by sex, race, ethnicity and residence (Table 1a). Tobacco use, family income and enrollment in KPWA were significantly associated with portal access in this population (Table 2a). Individuals who were current tobacco users were 56% less likely to be enrolled on the patient portal compared to former users (aPR 0.44, 95% CI 0.30–0.63, p < .001). Median neighborhood family income (in USD) was also associated with access to the portal, with households earning $0–49,999 70% less likely (aPR 0.30, 95% CI 0.14–0.63, p < .01) and $50,000–74,999 62% less likely (aPR 0.38, 95% CI 0.20–0.70, p < .01) to be enrolled in the patient portal compared to households with a median neighborhood family income of $100,000+. Lastly, we found that enrollment in the KPWA system was associated with portal access, with those enrolled in the system for fewer years less likely to be on the portal compared to those enrolled for 10+ years (<5 years: aPR 0.53, 95% CI 0.35–0.81, p < .01, 5–10 years: aPR 0.54, 95% CI 0.34–0.90, p < .05). Engagement with Digital Health Communication Among participants randomized to the intervention arm of the parent study and enrolled on the patient portal (n = 820), 81% (n = 666) opened the health communication intervention delivered through the patient portal secure message. Individuals who opened the health communication interventions were younger (median 67 years) compared to those who did not (median 71 years). There were no other notable demographic differences observed between those who opened the message versus those who did not (Table 1b). Similar to the portal access analysis, tobacco use and length of enrollment were also significantly associated with opening the health communication intervention (Table 2b). Current tobacco users were 47% less likely to open the health communication intervention compared to former users (aPR 0.53, 95% CI 0.35–0.80, p < .01). Length of enrollment at KPWA was also associated with message engagement; participants enrolled 5–10 years were 42% less likely to open the message compared to those enrolled 10+ years (aPR 0.58, 95% CI 0.34–0.99, p < .05). In addition, we found variation in message engagement by age and well-care visits. Adults >65 years were 34% less likely (aPR 0.66, 95% CI 0.51–0.88, p < .01) to open the message compared to those aged 50–64 years. Participants with >1 well-care visit in the last 12 months were also significantly less likely to open the portal intervention message (aPR 0.57, 95% CI 0.37–0.86, p < .01) compared to patients who had no well-care visits in the last year. Discussion We determined that multilevel factors influence who is actively using patient portals in an LCS population and the limit of intervention delivery through the portal. We determined that despite overall high engagement in our study sample, patient disparities remained in access to and reach of LCS interventions embedded in the patient portal, potentially perpetuating downstream unmet health care needs. While the use of patient portal functions (e.g. communication, reminders, scheduling) may improve care for patients across the cancer care continuum by increasing engagement with their health care (Coughlin et al., 2018; McCleary et al., 2018; Pho et al., 2019; Schliemann et al., 2022; Taplin et al., 2012), health systems play a crucial role in onboarding new patients to their EHR-linked portal to improve access and use of the patient portal as a communication tool. Our results also highlight the importance of including other approaches in outreach communication for LCS (e.g. US mail and phone) that may better align with the needs and preferences of some unreached populations. In our trial population, current tobacco use was associated with both lower access to the patient portal and lower engagement with LCS health communication interventions. These findings are consistent with prior research showing that adults currently using tobacco are less likely to engage in routine healthcare and preventative cancer screenings, compared with adults with no tobacco history or former use (Kahende et al., 2009; Royce et al., 2014; Sanford et al., 2019). These findings may also be indicative of a broader phenomenon of tobacco-related stigma as a deterrent to engage in health screening behaviors. In the US healthcare context, adults with a significant tobacco history report experiencing pervasive stigma in various encounters with the health system (Cataldo et al., 2011; Evans-Polce et al., 2015; Hamann et al., 2014; Lozano et al., 2020). Various studies have illustrated the consequences of tobacco-related stigma on engagement with care, specifically in the context of uptake of lung cancer screening (Carter-Harris, 2015; Lei & Lee, 2019; Prout et al., 2018; Quaife et al., 2017). Given this, when developing digital interventions for LCS, particularly for adult populations with significant tobacco histories, it is important to consider communication strategies that do not perpetuate stigmatizing language. Within the population eligible for lung cancer screening, adults aged 65+ were as likely to be on the patient portal as adults aged 50–65, but less likely to open intervention portal message. Previous research indicates that adults aged 65+ are interested in using patient portals to manage their health (Mace et al., 2022; Rodler et al., 2021), but barriers in use may persist by comfort level with technology (Chen et al., 2018), health literacy (Irizarry et al., 2017), and lower levels of education (Nahm et al., 2016). In qualitative studies, older adults report concerns about lack of training and support in portal use, leading to varied levels of understanding and use of patient portal functions (Irizarry et al., 2017; Nahm et al., 2016). Given that the use of telemedicine and digital health interventions has increased significantly since the COVID-19 pandemic (Getachew et al., 2023; Meskó, 2022; Shaver, 2022), health care providers should ensure that older adults can meaningfully use the patient portal to manage their care and access information before relying on this health communication intervention for LCS care. Older adults may also benefit from comprehensive instructions or tutorial sessions for using the patient portal. In contexts where older adults are not able to use the portal, digital communications should be supplemented by mailed communication or phone calls to ensure reach of health information. While we observed an 89% access rate in the patient portal in our study population, health systems across the U.S. have lower rates of uptake, with an estimated 57% patient portal access rate nationally (Strawley & Richwine, 2023). Rates are substantially lower in nonprofit health centers, such as federally qualified health centers (FQHCs) and community health clinics, ranging from 34% to 42% (Bush et al., 2018; Matthews et al., 2024). The underlying healthcare system for the trial has a robust EHR and linked patient portal infrastructure that has been integrated for almost 20 years (Ross et al., 2014). This has led to high uptake and use of the patient portal by patients, families, and caregivers (Chen et al., 2009). Despite this integration, our study uncovered persistent gaps in access to and use of the KPWA patient portal, consistent with other studies, which have identified age (Wang et al., 2023; Yoon et al., 2024), insurance type (El-Toukhy et al., 2020; Nishii et al., 2022), and socioeconomic status (Wang et al., 2023) as factors that influence use of the patient portal. Unique to our study population, we also found that a shorter length of enrollment in KPWA was associated with lower use of the portal, something that could be mitigated by ensuring new enrollees have access to and know how to enroll in the patient portal during their first visit with a health system provider. Study strengths include strong internal validity of findings due to adequate sample size, data leveraged from a pragmatic trial within a defined setting for LCS, and validated exposure and outcome measures obtained from the Kaiser Permanente Virtual Data Warehouse (Ross et al., 2014). This study also has limitations. Generalizability of findings is limited by use of a single study population in a setting of insured adults eligible for LCS in the Pacific Northwest, which may not reflect the diversity of U.S. health care settings, particularly those with high proportion of people who are uninsured. Further, KPWA has a higher adoption of the patient portal compared to other settings across the United States (Strawley & Richwine, 2023). As of 2022, members of KPWA resemble the overall population of Washington in various sociodemographic factors, with the exception that KPWA membership includes a higher proportion of White members compared to the overall population in WA (Davis et al., 2023). Similar studies may be warranted in different health care settings, specifically health systems that have lower rates of patient portal use, or have populations that are more diverse by race/ethnicity and insurance type. Lastly, our study population had already participated in at least one LCS scan as an eligibility criterion for the parent study. LCS screening interventions focused on uptake of initial screening might identify different patterns of disparities, due to differences in populations eligible for screening versus those who have already screened. Conclusions Our study provides preliminary findings on which factors are associated with access to the portal and engagement with portal messaging. Our results show that in a population eligible for LCS, there are disparities in use and access of patient portal by patient characteristics including tobacco use, income, older age, and length of enrollment within insurance plan. Future research can prioritize qualitative work with key subpopulations of interest to provide more context and in-depth understanding of the quantitative associations that were seen in this study, as well as explore disparities in other healthcare systems with more demographic diversity.