Digital health tools as conditional information scaffolds: Patient portals, wearables, communication, and trust in healthcare systems
Article information
Abstract
This study uses the Extended Mind Thesis and Distributed Cognition as sensitizing frameworks for understanding digital health tools as conditional information scaffolds rather than as directly observable cognitive extensions. Analyzing nationally representative HINTS 2025 data (N = 7,278), we distinguish two confirmatory hypotheses from a set of exploratory research questions. Final-person-weighted least squares models show that patient portal use is negatively associated with patient-centered communication (PCC) and positively associated with institutional trust. Wearable use demonstrates no direct effects on either outcome. Moderation results indicate that digital literacy and health data sharing condition some associations, suggesting possible sociotechnical coordination around digital health information rather than evidence of continuous somatic extension or full cognitive extension.
Introduction
The management of personal health information has shifted markedly from physician-centered control toward patient-facing engagement through digital technologies. Patient portals, online medical records, mobile health applications, and wearable health-tracking devices are now common parts of the health information environment. They provide patients with access to medical records, test results, clinical notes, health-related biometric data, and communication channels that may supplement encounters with clinicians (Dullabh et al., 2024; Moen & Brennan, 2005). These developments have changed how many people monitor health, seek information, coordinate care, and evaluate healthcare institutions.
Much of the digital health literature has conceptualized these technologies primarily as functional instruments: communication channels, repositories of medical information, self-monitoring tools, or behavior-change aids (Kruse et al., 2015; Patel et al., 2015). The literature however leaves a theoretical question unresolved. Digital health tools may not only provide information; they may also redistribute the informational and interpretive work that patients must perform. Patients must remember symptoms, interpret medical terminology, track results, coordinate appointments, compare recommendations, and decide what to ask clinicians. In this sense, the digitalization of healthcare is not merely a matter of access or convenience. It reorganizes where, when, and by whom health-related information work is performed.
The present study uses the Extended Mind Thesis (EMT) and Distributed Cognition (DCog) as sensitizing frameworks for examining this redistribution of health information work. EMT argues that external artifacts can support cognitive processes when they are reliably available, routinely used, and trusted as part of ongoing activity (Clark, 2010; Clark & Chalmers, 1998). DCog similarly emphasizes that cognitive work is often accomplished across systems of people, tools, and environments rather than within isolated individuals (Hutchins, 1995). Together, these frameworks invite attention to external resources, informational scaffolds, and coordinated systems of action.
The present study deliberately narrows the theoretical scope of this framing. HINTS 7 is a large cross-sectional survey, and its measures do not directly observe cognitive extension, distributed cognition, embodied integration, or the moment-to-moment coupling between patients and technologies. The data instead measure behaviors and perceptions: portal access frequency, wearable tracking intensity, digital literacy, health-data sharing with health professionals, patient-centered communication, and trust in the healthcare system. Accordingly, EMT and DCog are not treated here as theories that can be directly tested with HINTS measures. They are used as theoretical lenses that motivate more modest and observable constructs, including cognitive offloading, information scaffolding, sociotechnical coordination, and distributed information work. The present study therefore uses language of scaffolding and information work rather than claiming that portals or wearables are genuine cognitive extensions.
Patient portals are examined as institutionally linked information scaffolds. They may allow patients to access medical records outside clinical encounters, review test results, formulate questions, and communicate asynchronously with providers. Wearables are examined as potential sources of patient-generated health data (PGHD). They may expose users to health-related information streams, but the HINTS measure captures only tracking frequency, not interpretation, bodily perception, provider use, or clinical integration. This distinction is essential for rigorous interpretation.
The present study focuses on two relational outcomes: patient-centered communication (PCC) and trust in the healthcare system. PCC reflects whether healthcare professionals give patients opportunities to ask questions, respond to emotions, involve patients in decisions, explain information clearly, spend enough time, and help manage uncertainty. Trust in the healthcare system captures confidence in healthcare organizations such as hospitals, pharmacies, and other institutions. These outcomes are theoretically important because digital health tools are often justified as ways to empower patients and strengthen relationships with healthcare systems. Using nationally representative HINTS 2025 data (N = 7,278), this study asks whether population-level patterns are consistent with the idea that digital health technologies function as conditional information scaffolds. The analysis does not attempt to prove cognitive extension. Instead, it evaluates whether portal access and wearable tracking are associated with communication and trust, and whether those associations vary by digital literacy and health-data sharing.
Literature Review
Extended Mind Thesis and Distributed Cognition in Digital Health
The rapid integration of digital technologies into everyday life has prompted renewed interest in theoretical perspectives that challenge traditional assumptions about the boundaries of cognition. Two influential frameworks—the Extended Mind Thesis (EMT) and Distributed Cognition (DCog)—provide useful conceptual tools for understanding how individuals increasingly rely on external artifacts and technological systems to support information processing, decision-making, and problem solving (Clark & Chalmers, 1998; Hutchins, 1995). Although these perspectives emerged outside healthcare contexts, they offer valuable insights into how digital health technologies may reshape information management and patient engagement.
EMT proposes that cognitive processes are not necessarily confined to the biological brain. Under certain conditions, external artifacts may support cognitive activities in ways functionally equivalent to internal mental processes (Clark & Chalmers, 1998). Clark (2010) later emphasized that external resources become especially important when they are reliably available, routinely used, and trusted by the user. Research inspired by EMT has examined a wide range of technologies, including smartphones, navigation systems, digital calendars, and online information repositories, demonstrating how external resources can reduce memory burdens and support complex cognitive tasks (Heersmink, 2015; Smart, 2017).
DCog offers a complementary perspective by focusing less on where cognition resides and more on how cognitive work is distributed across individuals, technologies, and environments (Hutchins, 1995). Rather than treating cognition as an exclusively individual phenomenon, DCog emphasizes systems of coordination in which information is stored, transmitted, interpreted, and acted upon across multiple actors and artifacts. Cognitive performance emerges through interactions among people, technologies, and institutional structures rather than from isolated individual capacities.
Both frameworks have important implications for healthcare. Modern healthcare increasingly depends on complex information infrastructures that connect patients, providers, electronic records, mobile applications, wearable devices, and communication systems. Patients must manage appointments, monitor symptoms, interpret laboratory results, evaluate treatment options, and coordinate care across multiple settings. These activities involve substantial informational and interpretive work that extends beyond traditional clinical encounters.
At the same time, caution is warranted when applying EMT and DCog to survey-based research. Large-scale survey data cannot directly observe cognitive extension, distributed cognition, or embodied integration. Therefore, the present study does not treat EMT or DCog as theories that can be directly tested using HINTS data. Instead, these frameworks function as sensitizing perspectives that direct attention toward observable mechanisms such as cognitive offloading, information scaffolding, sociotechnical coordination, and distributed information work. These mechanisms provide a more empirically tractable basis for understanding how digital health technologies may influence communication and trust within healthcare systems.
Patient Portals as Information Scaffolds for Healthcare Communication
Patient portals represent one of the most widely adopted forms of digital health technology. Most contemporary healthcare systems provide portal platforms that allow patients to access medical records, laboratory results, medication lists, appointment information, and secure messaging functions (Kruse et al., 2015; Otte-Trojel et al., 2014). These systems are frequently promoted as tools that increase patient engagement, facilitate communication, and improve healthcare quality.
Communication plays a central role in healthcare. Effective communication contributes to treatment adherence, satisfaction, trust, health literacy, and health outcomes (Street et al., 2009). Patient-centered communication (PCC) specifically emphasizes responsiveness to patient concerns, shared decision-making, emotional support, clear explanations, and opportunities for active participation in care. Because communication influences nearly every aspect of healthcare experience, identifying factors that enhance or inhibit PCC remains an important research priority.
Patient portals may support communication by reducing informational asymmetries between patients and providers. Traditionally, healthcare information has been concentrated within clinical settings and controlled primarily by healthcare professionals (Lucivero, 2017). Portals alter this arrangement by providing patients with direct access to information that was previously difficult to obtain. Laboratory results, clinical notes, treatment recommendations, and medication records can be reviewed outside clinical encounters and revisited as needed.
From an information-scaffolding perspective, portals may reduce cognitive demands associated with remembering, organizing, and interpreting health information (Moen & Brennan, 2005). Patients no longer need to rely exclusively on memory or handwritten notes to recall treatment instructions. Instead, information can be stored externally and revisited repeatedly. Such processes resemble forms of cognitive offloading in which informational tasks are redistributed across human and technological resources.
Research generally suggests that portal users report greater engagement with healthcare information and increased participation in care processes (Otte-Trojel et al., 2014; Zaidi et al., 2022). Patients frequently use portals to prepare for appointments, review treatment information, and communicate with providers between visits. These activities may improve communication quality by facilitating question generation, increasing informational preparedness, and supporting more active participation during clinical encounters.
However, portal use may also produce more complex effects. Increased access to information may heighten awareness of communication deficiencies, reveal inconsistencies in care, or increase expectations regarding provider responsiveness. Thus, it is possible that the relationship between portal use and communication may not always be uniformly positive. Given the substantial body of literature linking portal engagement to patient participation and communication, the following hypothesis is proposed:
H1: Greater patient-portal engagement will be associated with patient-centered communication.
Digital Literacy as a Boundary Condition for Portal-Based Communication
Although patient portals provide access to information, meaningful use of that information depends on users’ abilities to navigate digital environments and evaluate health information effectively. Digital literacy refers to the knowledge and skills required to locate, interpret, evaluate, and apply information obtained through digital technologies (Norman & Skinner, 2006).
Individuals differ substantially in their ability to use digital health technologies. Research demonstrates that digital literacy influences portal adoption, engagement with online health information, and participation in digital-health activities (Deshpande et al., 2023; Mackert et al., 2016). Individuals with higher digital literacy may be better positioned to interpret laboratory results, understand clinical documentation, and communicate effectively using portal messaging systems (Deshpande et al., 2023). In contrast, individuals with lower digital literacy may experience greater difficulty navigating interfaces or understanding available information. From a distributed-information-work perspective, digital literacy may influence the extent to which individuals can successfully coordinate informational activities across clinical and technological settings. The benefits associated with portal use may depend not only on access to information but also on the skills required to interpret and use that information effectively.
Because theoretical expectations regarding the direction of moderation remain mixed, digital literacy is examined as an exploratory moderator.
RQ1: How does digital literacy affect the relationship between patient-portal engagement and patient-centered communication?
Health-Data Sharing and Portal-Based Communication
Patient portals do more than provide access to information; they also facilitate the movement of information across social and clinical networks (Salmi et al., 2024). Increasingly, patients share health information obtained through portals with healthcare providers, family members, caregivers, and other members of their social support systems (Bratches et al., 2023). Such sharing practices may influence how effectively information is interpreted, contextualized, and incorporated into healthcare decision-making.
DCog emphasizes that information often becomes meaningful through collective processes rather than through isolated individual interpretation (Hutchins, 1995). Healthcare decisions frequently involve multiple actors who contribute distinct forms of expertise, experience, and support. Family members may help interpret information, remind patients about treatment recommendations, or assist with healthcare navigation. Similarly, healthcare providers may use patient-generated information to clarify symptoms, monitor progress, or tailor treatment recommendations. Information-sharing practices may therefore strengthen the communicative benefits associated with portal use. Individuals who actively share health information may have more opportunities to discuss health concerns, seek clarification, and engage in collaborative problem-solving. Such interactions could reinforce patient-centered communication by expanding opportunities for dialogue and mutual understanding. Alternatively, information sharing may increase exposure to conflicting interpretations, create informational overload, or complicate healthcare communication. The direction of moderation therefore remains uncertain and warrants empirical examination.
RQ2: How does health-data sharing affect the relationship between patient-portal engagement and patient-centered communication?
Patient Portals, Transparency, and Institutional Trust
Trust is a foundational element of effective healthcare systems. Patients who trust healthcare institutions are more likely to seek care, follow medical recommendations, disclose relevant information, and maintain long-term relationships with providers (Hesse et al., 2005). Trust also contributes to perceptions of legitimacy, competence, and fairness within healthcare organizations.
Digital technologies increasingly shape how trust is formed and maintained. Contemporary patients often encounter health information through multiple channels, including websites, social media, online communities, healthcare organizations, and digital platforms. Although greater access to information can support informed decision-making, it can also expose individuals to misinformation, contradictory claims, and uncertainty regarding information quality (Tan & Goonawardene, 2017).
Patient portals differ from many other digital information sources because they are directly connected to healthcare organizations and clinical records. Information available through portals is institutionally linked, personalized, and typically generated within formal healthcare systems. As a result, portals may provide a relatively stable informational environment compared with broader online information ecosystems.
Transparency has long been recognized as a key component of institutional trust. When patients can access medical records, review clinical notes, and monitor treatment information, healthcare processes may appear more understandable and accountable. Such access may reduce uncertainty and reinforce perceptions that healthcare organizations are acting openly and responsibly.
Prior research suggests that portal users often report greater confidence in healthcare organizations and greater satisfaction with healthcare information (Carini et al., 2021; Moqbel et al., 2022). Accordingly, patient portals may contribute to trust by increasing informational accessibility and reducing informational asymmetries.
H2: Greater patient-portal engagement will be associated with trust in the healthcare system.
Digital Literacy and Trust in Healthcare Systems
Although digital literacy may facilitate access to healthcare information, its relationship with institutional trust is likely complex. Individuals with higher digital literacy often possess greater capacity to evaluate information quality, compare sources, identify inconsistencies, and critically assess institutional claims (Neter & Brainin, 2012). Greater evaluative capacity may strengthen trust when healthcare information appears transparent, accurate, and responsive. However, it may also increase skepticism when individuals encounter discrepancies, delays, or limitations within healthcare systems. Digital literacy may therefore operate simultaneously as a resource for informed engagement and as a catalyst for critical evaluation. Research has produced mixed findings. Some studies suggest that greater informational competence promotes confidence in healthcare institutions, whereas others indicate that highly informed individuals may become more critical consumers of healthcare information (Mackert et al., 2016; Norman & Skinner, 2006). Because competing theoretical expectations exist, the moderating role of digital literacy remains an empirical question.
RQ3: How does digital literacy affect the relationship between patient-portal engagement and trust in the healthcare system?
Health-Data Sharing and Trust Formation
Trust is not formed solely through direct interactions between individuals and institutions. It is also shaped through social networks in which information is exchanged, interpreted, and discussed. Information-sharing practices may therefore influence how digital-health technologies affect perceptions of healthcare organizations. Individuals who actively share health information with providers, family members, or caregivers may experience greater informational support and more opportunities for collaborative interpretation. These interactions may reinforce confidence in healthcare systems by reducing uncertainty and increasing perceptions of informational transparency. In contrast, concerns regarding privacy, surveillance, and secondary uses of personal data may undermine trust. Digital-health technologies require patients to disclose and exchange information through technological infrastructures that are not always fully understood. Hence, health-data sharing may strengthen trust for some individuals while weakening trust for others. The extent to which information-sharing practices influence relationships between portal use and trust remains unclear.
RQ4: How does health-data sharing affect the relationship between patient-portal engagement and trust in the healthcare system?
Wearables as Sources of Patient-Generated Health Data
Wearable health technologies have become increasingly common components of contemporary digital-health ecosystems. Devices such as smartwatches, fitness trackers, and biometric monitoring systems generate continuous streams of information regarding physical activity, sleep, heart rate, and other physiological indicators (Piwek et al., 2016). Early discussions of wearable technologies sometimes framed these devices as extensions of human cognition or embodiment (Nelson et al., 2019). However, such claims require evidence regarding how information is interpreted, integrated into decision-making, and incorporated into everyday cognitive practices. The HINTS dataset does not provide measures capable of evaluating such processes directly.
Accordingly, wearable technologies are conceptualized in the present study as sources of patient-generated health data (PGHD). This framing more closely reflects the available measures and aligns with reviewer concerns regarding conceptual overreach. The wearable measure used in HINTS captures frequency of device use rather than interpretation, clinical integration, or cognitive incorporation of device-generated information.
Patient-generated health data may nevertheless influence healthcare communication. Wearables provide users with access to information that may increase health awareness, stimulate questions, or encourage discussions with healthcare providers. At the same time, information generated by wearable devices may be difficult to interpret without appropriate contextual knowledge. Because prior findings regarding wearable technologies remain mixed, their relationship with communication is treated as exploratory.
RQ5: How does wearable-use frequency relate to patient-centered communication?
Digital Literacy and Wearable Use
The potential value of wearable-generated information may depend heavily on users’ ability to understand and interpret that information. Unlike many patient-portal functions, wearable technologies often generate large volumes of continuously updated data that require interpretation. Individuals with higher digital literacy may be better equipped to understand trends, identify meaningful patterns, and determine whether observed changes warrant further attention. In contrast, individuals with lower digital literacy may struggle to interpret device-generated information or may ignore available data altogether. Thus, digital literacy may condition the relationship between wearable engagement and communication outcomes.
RQ6: How does digital literacy affect the relationship between wearable-use frequency and patient-centered communication?
Health-Data Sharing and Wearable Use
Patient-generated health data become clinically meaningful only when they are communicated, interpreted, and acted upon. Sharing wearable-generated information with healthcare providers may create opportunities for more informed discussions, greater patient engagement, and collaborative decision-making. Similarly, sharing information with family members or caregivers may facilitate social support and collective problem-solving. These processes are consistent with distributed-information-work perspectives that emphasize the importance of information exchange across networks of individuals (Wolff et al., 2015). However, the extent to which health-data sharing influences relationships between wearable use and communication remains uncertain.
RQ7: How does health-data sharing affect the relationship between wearable-use frequency and patient-centered communication?
Wearables and Institutional Trust
Wearable technologies may also influence trust in healthcare institutions. Individuals who actively monitor health information may become more engaged with healthcare systems and more aware of how institutions manage, interpret, and respond to health-related information. However, wearable use may also expose users to discrepancies between personal observations and institutional recommendations. In some cases, increased access to health information could strengthen confidence in healthcare systems; in others, it could encourage skepticism. Because theoretical expectations remain ambiguous and empirical findings are inconsistent, the relationship between wearable use and institutional trust is treated as exploratory.
RQ8: How does wearable-use frequency relate to trust in the healthcare system?
Digital Literacy, Wearables, and Trust
Digital literacy may influence how wearable-generated information affects trust. Highly digitally literate individuals may be better able to evaluate data quality, understand measurement limitations, and contextualize information appropriately. Such capabilities may facilitate informed engagement with healthcare systems. At the same time, greater digital competence may encourage critical scrutiny of healthcare recommendations and institutional practices. Consequently, digital literacy may strengthen or weaken relationships between wearable engagement and trust.
RQ9: How does digital literacy affect the relationship between wearable-use frequency and trust in the healthcare system?
Health-Data Sharing, Wearables, and Trust
Sharing wearable-generated information may also shape trust outcomes. Information that is discussed with healthcare providers, interpreted collaboratively, and incorporated into healthcare interactions may reinforce confidence in healthcare institutions. In contrast, negative experiences involving information sharing may undermine trust. From a distributed-information-work perspective, information derives value from its circulation across networks of actors who can interpret and act upon it. Health-data-sharing practices may therefore represent an important mechanism linking wearable engagement to institutional trust.
RQ10: How does health-data sharing affect the relationship between wearable-use frequency and trust in the healthcare system?
Complementary Digital Health Technologies
Although patient portals and wearable devices are frequently studied separately, many individuals use multiple digital-health technologies simultaneously. Patient portals and wearables provide different forms of informational support. Portals facilitate access to institutionally generated information such as clinical notes, laboratory results, and appointment records. Wearables generate patient-centered information regarding everyday health behaviors and physiological indicators. When used together, these technologies may create opportunities for integrating clinical information with patient-generated data.
From a distributed-information-work perspective, multiple technologies may complement one another by supporting different aspects of information management. (Piwek et al., 2016) However, greater technological complexity may also increase informational burdens and create new coordination challenges. Because little research has examined the combined influence of portals and wearables on communication outcomes, the interaction between these technologies is treated as exploratory.
RQ11: Do patient-portal engagement and wearable-use frequency interact such that patient-centered communication differs among individuals who engage extensively with both technologies?
Methods
Data for this study were drawn from HINTS 7, the seventh Health Information National Trends Survey (HINTS) administered by the National Cancer Institute (2025). HINTS is a nationally representative survey of civilian, noninstitutionalized adults aged 18 years or older in the United States. Data collection occurred from March 25 through September 16, 2024, using a self-administered paper survey delivered by mail with a push-to-web option. The final public-use file contained 7,278 completed surveys and a weighted response rate of 27.3% (Westat, 2025). HINTS 7 used a two-stage address-based sampling design. In the first stage, addresses were selected from a national residential address frame. The sample was stratified by high versus low racial/ethnic minority concentration and rural versus urban location, producing four strata. High-minority and rural strata were oversampled to increase precision for subpopulation estimates. In the second stage, one adult was selected within each sampled household using the Next Birthday Method. HINTS 7 also included embedded methodological experiments involving incentives and respondent commitment statements. These design features are important because they explain why survey weights are needed for population-representative estimates.
The analysis used the HINTS 7 public-use SPSS file and the public codebook. Invalid response codes, including missing, multiple responses, skipped items, refused responses, and don’t know responses, were recoded to missing before analysis. Variables were cleaned using the HINTS codebook definitions. All descriptive statistics and regression models used the final-person-level survey weight to improve population representativeness. Replicate weights were not incorporated into the variance estimation in the present analysis. As a result, point estimates are weighted, but standard errors and p values should be interpreted cautiously, particularly for marginal findings.
The full file included N = 7,278 respondents. Model-specific complete-case analysis was used because missingness differed across outcomes and predictors. The final PCC model included N = 4,823 respondents, whereas the final trust model included N = 5,397. Missingness was lowest for internet use frequency (0.4%) and highest for PCC (13.0%), income (12.4%), and wearable tracking intensity (10.4%). Comparison between retained and excluded respondents indicated modest differences in age, education, income, and internet use that did not alter substantive conclusions.
Multiple imputation was considered but was not implemented because the primary analyses involved numerous interaction terms, weighted estimation procedures, and model-specific outcome samples. Complete-case analysis was therefore used to maintain consistency across model specifications. Comparisons between retained and excluded respondents indicated only modest demographic differences, suggesting that complete-case estimation was unlikely to substantially alter substantive conclusions, although some bias due to nonrandom missingness cannot be ruled out.
Patient-Centered Communication (PCC) was measured using the HINTS-derived PCC scale. The codebook identifies seven items assessing how often doctors, nurses, or other health professionals gave respondents a chance to ask questions, addressed feelings and emotions, involved them in decisions, made sure they understood what to do, explained things clearly, spent enough time, and helped them deal with uncertainty. HINTS provides a derived PCCSCALE ranging from 0 to 100, with higher values indicating more favorable communication experiences. The weighted mean was 76.16 (SD = 23.20, Cronbach’s α = .94), with observed values from 0 to 100.
Trust in the Healthcare System was measured with the HINTS item asking, “How much do you trust the health care system (for example, hospitals, pharmacies, and other organizations involved in health care)?” The original response coding was cleaned and reverse-coded so that higher values indicate greater trust. The resulting scale ranged from 1 to 4. The weighted mean was 3.10 (SD = .81).
Portal Access Frequency was measured using the HINTS item that asekd how many times respondents accessed their online medical record or patient portal in the last 12 months. The cleaned measure was coded from 0 to 4, with higher values indicating more frequent access. The weighted mean was 1.59 (SD = 1.41).
Wearable Tracking Intensity was measured using the HINTS item asking how often respondents used an electronic wearable device such as Apple Watch and Fitbit to track health in the past month. The cleaned variable ranged from 0 to 4, with higher values indicating more frequent tracking. The weighted mean was 1.28 (SD = 1.67). This item captures reported frequency of wearable use only. It does not measure what type of data were generated, whether respondents reviewed or interpreted those data, whether wearable data shaped bodily perception or self-understanding, or whether data were discussed with clinicians or incorporated into care decisions. A separate descriptive adoption indicator showed that 41.1% of weighted respondents reported any wearable health tracking in the past 12 months.
Digital Literacy was measured with three HINTS items: (1) I find learning how to use new technology frustrating (reverse-coded); (2) I can use applications/ programs (e.g., Zoom) without help; (3) I have the skills to find health information online. Responses ranged from 1 (Strongly disagree) to 4 (Strongly agree). Items were averaged (M = 3.12, SD = .76, α = .72).
Health Data Sharing was constructed from two dichotomous Yes/No items: (1) whether the respondent had shared health information from a device or app with a healthcare provider during the past 12 months, and (2) whether the respondent had shared health information with family or friends. Responses were recoded 0 = No and 1 = Yes and summed (range = 0–2). Higher values indicate broader sharing of digital health information. The weighted mean was .98 (SD = .85; Spearman-Brown = .67).
Control Variables
Age was measured continuously in years (M = 49.00, SD = 18.05). Education was measured on a 7-point scale (1 = < 8 years, 2 = 8–11 years, 3 = 12 years/High School, 4 = Post-High School/vocational, 5 = Some college, 6 = College graduate, 7 = Postgraduate) (M = 4.75; SD = 1.53). Gender was assessed via biological sex at birth and coded as a binary variable (1 = Female 50.58%; 0 = Male 48.76%; don’t know: .66%). Income was measured using a 9-point ordinal scale representing household income brackets, ranging from 1 (0 to $9,999) to 9 ($200,000 or more) (M = 5.69; SD = 2.33). Internet Use Frequency was measured on a 6-point scale by asking “About how often do you use the Internet, either on a computer, laptop, smartphone or any other device?” (M = 5.64, SD = 1.03).
Social Media Engagement was measured on a 6-point scale with the following items: (1) In the last 12 months, how often did you share personal health information on social media? (2) How often did you share general health-related information on social media (e.g., a news article)? (3) How often did you interact with people who have similar health or medical issues on social media or online forums? and (4) How often did you watch a health-related video on a social media site (for example, YouTube)? (M = 2.12, SD = .71, α = .72).
Analytic Strategies
The analysis proceeded in several stages. First, weighted descriptive statistics, missingness, and weighted correlations were computed. Second, two weighted least squares regression models were estimated, one predicting PCC and one predicting trust. Each model included demographic controls, media-use controls, digital literacy, health-data sharing, portal access frequency, wearable tracking intensity, and five exploratory interaction terms: portal x wearable, portal x digital literacy, portal x health-data sharing, wearable x digital literacy, and wearable x health-data sharing. Continuous variables involved in interaction terms were mean-centered before product terms were created. Third, simple slopes were computed at low (−1 SD), mean, and high (+1 SD) values of moderators for each focal interaction. Fourth, robustness models added healthcare-context covariates where available, including general health, chronic condition indicators, healthcare utilization, insurance status, race/ethnicity, and region. These robustness models are reported in summary form because their purpose was to assess pattern stability rather than replace the primary models.
This analytic strategy distinguishes confirmatory tests from exploratory analyses, reducing the risk of treating post hoc moderation patterns as direct evidence for EMT or DCog.
Weighted least squares regression was employed to incorporate final-person survey weights in estimating population-level associations. Replicate-weight variance estimation was not implemented in the present analyses. Because HINTS uses a complex survey design, standard errors and statistical significance tests reported here may differ from those obtained using the full set of replicate weights recommended by HINTS documentation. Accordingly, emphasis is placed on the direction, magnitude, confidence intervals, and overall consistency of observed associations rather than on individual p-values alone, particularly for interaction effects. Variance inflation factors ranged from 1.18 to 2.67, well below the commonly accepted threshold of 5.0, indicating that multicollinearity was not a significant concern. Robustness models added healthcare-context covariates where available, including general health, chronic condition indicators, healthcare utilization, insurance status, race/ethnicity, and region (Table 5).
Robustness Models Adding Health and Healthcare-Context Covariates.
Results
Descriptive Statistics
Table 1 reports missingness and analytic sample sizes. Missingness varied by variable, reflecting skip patterns and item nonresponse. PCC had 948 missing cases (13.0%), wearable tracking intensity had 754 missing cases (10.4%), and income had 899 missing cases (12.4%). The final complete-case PCC model included 4,823 respondents, and the final trust model included 5,397 respondents. These sample sizes are lower than the full public-use file because all variables in the final interaction models were required to be present. Correlations of the major variables are presented in Table 2.
Missingness and Model-Specific Analytic Samples.
Weighted Correlations Among Study Variables.
Respondents excluded from the PCC model were somewhat younger (M = 48.81 vs. 49.00), reported higher educational attainment (M = 4.82 vs. 4.75), lower income (M = 5.60 vs. 5.69), and more frequent internet use (M = 5.69 vs. 5.64). Respondents excluded from the trust model were somewhat older (M = 49.12 vs. 49.00), reported higher educational attainment (M = 4.79 vs. 4.75), lower income (M = 5.62 vs. 5.69), and less frequent internet use (M = 5.60 vs. 5.64). Although statistically detectable, these differences were modest and did not alter substantive conclusions.
Inferential Statistics
H1 predicted that greater use of patient portals would be positively associated with patient-centered communication (PCC). Contrary to expectations, portal use was negatively associated with PCC (b = −.114, p < .001, Table 3). Thus, H1 was not supported; the association was significant in the opposite direction. Notably, the multivariable coefficient differs from the small positive bivariate correlation reported in Table 2. This pattern suggests that the portal–communication relationship is sensitive to adjustment for demographic, socioeconomic, media-use, and digital-engagement factors. Accordingly, the negative coefficient should be interpreted as a conditional association estimated after accounting for these covariates rather than as a simple population-level relationship. However, this coefficient should not be interpreted as evidence that portals worsen communication. Given the cross-sectional design, the association is consistent with several possibilities: patients experiencing lower-quality communication may use portals in a compensatory manner; portal users may evaluate communication more critically because they have greater access to records and follow-up information; or the association may reflect confounding by health status, healthcare utilization, or complexity of care.
Weighted Least Squares Regression Predicting Patient-Centered Communication and Trust in Healthcare System.
RQ1 asked how digital literacy affects the relationship between portal use and PCC. The interaction between portal use and digital literacy was not significant in predicting PCC (b = .047, p = .223). Simple slopes were nonsignificant at low, mean, and high levels of digital literacy. This suggests that the association between portal use and PCC did not vary substantially across levels of digital literacy.
RQ2 asked whether health-data sharing would moderate the portal-PCC relationship. This interaction was also nonsignificant (b = .005, p = .996). This finding indicates that the portal-PCC association did not vary meaningfully by health-data sharing in these data. Health data sharing, however, was independently and positively associated with PCC (b = .134, p = .012), indicating that respondents who shared more health information tended to report better communication with providers.
Supporting H2, patient portal use was positively associated with institutional trust (b = .313, p = .008). This indicates that portal use may help patients feel more confident in the healthcare system, even though it does not appear to improve patient-centered communication.
RQ3 asked how digital literacy would affect the relationship between portal use and institutional trust. The interaction between portal use and digital literacy was significant and negative for trust (b = −.153, p = .004; Figure 3). Simple-slope analyses indicated that portal use was positively associated with trust at low (b = .046, p = .042) and average (b = .025, p = .035) levels of digital literacy, but not at high levels of digital literacy (b = .003, p = .169). These exploratory findings suggest that the trust-enhancing association of portal use may diminish as digital literacy increases, although this pattern was sensitive to additional healthcare-context controls.
Portal Access Frequency x Digital Literacy Predicting Trust in the Healthcare System.
RQ4 asked how health data sharing would affect the relationship between portal use and institutional trust. The interaction between portal use and health data sharing was significant and positive for trust (b = .116, p = .022; Figure 4). This result indicates that the trust-enhancing association of portal use was stronger among respondents who more actively shared health information. In other words, the positive association between portal use and trust was stronger among respondents reporting greater health-data-sharing activity. This exploratory pattern is consistent with the possibility that portal engagement may be more strongly associated with trust when embedded within broader practices of information exchange.
Portal Access Frequency x Health-Data Sharing Predicting Trust in the Healthcare System. Running head: Digital Health and Information Scaffolding
Turning to wearable health technologies, RQ5 asked how reported frequency of wearable health-tracking device use relates to PCC. Wearable use was not significantly associated with PCC (b = −.019, p = .634). Thus, reported wearable-use frequency did not independently improve or worsen perceived communication with providers in this exploratory analysis. Because the measure is a single frequency item, this null result cannot determine whether wearable-generated data were ignored, uninterpreted, clinically irrelevant, or simply not shared in patient-provider encounters.
RQ6 asked how digital literacy would moderate the link between reported wearable-use frequency and PCC. The interaction between wearable use and digital literacy was significant and positive for PCC (b = .137, p = .035; Figure 1). Among respondents with higher digital literacy, wearable use was weakly but positively associated with communication quality, whereas among respondents with lower literacy, no meaningful association emerged. This pattern suggests that digitally skilled users may be better positioned to incorporate device-generated information into health-related discussions, but the measure does not show whether wearable data were actually interpreted, narrated, or discussed with providers.
Wearable Tracking Intensity x Digital Literacy Predicting Patient-Centered Communication.
RQ7 asked how health data sharing would moderate the link between reported wearable-use frequency and PCC. The interaction between wearable use and health data sharing was significant and positive for PCC (b = .172, p = .020; Figure 2). Respondents who used wearables more frequently and reported more health data sharing also reported stronger communication with doctors. This pattern suggests that wearable-related information may matter more when embedded in broader sharing practices, but the analysis cannot verify that wearable data specifically were reviewed, interpreted, or used in clinical exchange.
Wearable Tracking Intensity x Health-Data Sharing Predicting Patient-Centered Communication.
RQ8 asked how reported wearable-use frequency would relate to institutional trust. Wearable use was not significantly associated with institutional trust (b = −.013, p = .660). This suggests that wearable-use frequency, on its own, neither strengthens nor weakens trust in the healthcare system in the present survey data.
RQ9 asked how digital literacy would moderate the link between reported wearable-use frequency and institutional trust. The interaction between wearable use and digital literacy was significant and negative for trust (b = −.144, p = .030; Figure 5). Among higher-literacy respondents, greater wearable use was associated with slightly lower institutional trust. This pattern suggests that more knowledgeable users may be more skeptical about how institutions interpret, integrate, or respond to patient-generated data, but the survey cannot establish whether such skepticism arose from actual clinical encounters involving wearable data.
Wearable Tracking Intensity x Digital Literacy Predicting Trust in the Healthcare System.
RQ10 asked how health data sharing would affect the link between reported wearable-use frequency and institutional trust. The interaction between wearable use and health data sharing was significant and positive for trust (b = .145, p = .027; Figure 6). Respondents who used wearables more frequently and reported more health data sharing also reported higher institutional trust. This association suggests that reciprocal data-flow practices may matter for trust, but it should not be interpreted as direct evidence that wearable data were clinically integrated or jointly interpreted.
Wearable Tracking Intensity x Health-Data Sharing Trust in the Healthcare System.
Note. Predicted values are based on the fully adjusted weighted least squares model with nonfocal covariates held at their model reference or centered values. Lines display low (−1 SD), mean, and high (+1 SD) values of the moderator.
Finally, RQ11 asked whether portal use and wearable use interact such that PCC is highest among individuals who use both technologies. The interaction term between portal use and wearable use was non-significant (b = .023, p = .558). This null exploratory result suggests that combining multiple digital scaffolds does not automatically yield supra-additive or synergistic relational benefits. One possible interpretation is that relational benefits may require qualitative, systematic coordination between the patient’s individual capabilities, technological interfaces, and institutional workflow structures.
Robustness analyses are reported in Table 5. After adding general health, healthcare utilization, insurance status, chronic-condition count, cancer history, race/ethnicity, and region, the direct association between portal access and PCC remained nonsignificant, wearable tracking intensity was negatively associated with PCC, and the wearable x health-data-sharing interaction remained positive for PCC. For institutional trust, the portal x wearable and portal x health-data-sharing interactions remained statistically significant, whereas the portal x digital-literacy, wearable x digital-literacy, and wearable x health-data-sharing interactions were not significant in the robustness specification. These results suggest that several exploratory moderation patterns are sensitive to additional healthcare-context controls and should be interpreted cautiously.
Simple-Slope Probes for Exploratory Interactions.
Discussion
Drawing on the Extended Mind Thesis (EMT; Clark & Chalmers, 1998) and Distributed Cognition (DCog; Hutchins, 1995), this study examined how digital health technologies are associated with communication, trust, and patterns potentially consistent with information scaffolding and distributed information work, patient-provider communication, and institutional trust. The theoretical scope is intentionally modest: the HINTS data capture portal use, wearable use, data sharing, digital literacy, PCC, and trust, but they do not directly observe cognitive extension or distributed cognition. The interpretation is also intentionally tiered. H1 and H2 provide the manuscript’s confirmatory tests of directional portal-use expectations, whereas RQ1–RQ11 provide exploratory evidence about boundary conditions and heterogeneous patterns. Instead of treating portals and wearables as genuine extensions of cognition, the present study asks whether their observable use patterns are consistent with partial offloading and sociotechnical coupling in healthcare. EMT helps explain why reliable access, usability, and trust matter for technology-supported cognition, while DCog explains why redistributed information work requires coordination among patients, tools, providers, and institutions.
Contrary to H1 and a substantial body of prior research emphasizing portals’ potential to facilitate shared decision-making and information access (Kruse et al., 2015; Otte-Trojel et al., 2014; Zaidi et al., 2022), greater patient portal use was negatively associated with patient-centered communication. Portal use was not associated with more favorable patient-centered communication in these data. Instead, they may function primarily as a compensatory mechanism adopted in response to perceived deficiencies in clinical encounters. Patients may use portals to better understand information and manage follow-up after brief or unclear clinical encounters. One possible interpretation is that portals may function as compensatory informational resources for patients who perceive gaps in communication, although the present data cannot determine whether cognitive workload actually increased or decreased. This interpretation aligns with a recent finding that lower patient-centered communication predicts both reduced portal offers by providers and lower patient engagement with portals (Anthony et al., 2024). It also suggests that patients sometimes experience portals as additional work rather than empowerment, particularly when institutional responsiveness lags. Neither digital literacy nor health data sharing significantly moderated the negative association. This suggests that individual interpretive capacities and sharing behaviors cannot fully overcome deeper structural barriers to reciprocal sense-making, such as time constraints, informational asymmetry, and limited clinician integration of portal data during encounters (Street et al., 2009). Simply giving patients access to records does not automatically improve patient-provider communication.
In contrast to the communication findings, patient portal use was positively associated with institutional trust, supporting H2. This result is consistent with the view that curated, institutionally tethered informational environments can buffer epistemic uncertainty generated by uncurated general internet health searches (Hesse et al., 2005; Tan & Goonawardene, 2017). Portal use may increase trust because it gives patients a stable information repository connected with providers and reduces their need to use less reliable online sources. The moderation patterns involving digital literacy and health data sharing, however, are exploratory and should be interpreted as theory-generating boundary conditions rather than as direct tests of EMT or DCog. Higher digital literacy weakened, rather than strengthened the positive portal-trust association, suggesting that digitally skilled users may notice delays, gaps, or inconsistencies in portal information. One speculative interpretation is that digitally skilled users engage institutional information more critically. By contrast, health data sharing strengthened the trust benefit of portal use. When patients actively shared portal-derived information with providers or family, the positive link to institutional trust intensified. This pattern is consistent with DCog’s emphasis on social embedding and reciprocal coordination, but it requires future confirmatory testing.
Wearable health-tracking devices exhibited a distinct and more muted pattern. Reported wearable-use frequency showed no significant direct associations with either patient-centered communication or institutional trust. At the population level, more frequent wearable use neither improved nor degraded these relational outcomes. This null exploratory result is consistent with scholarship positioning wearables as facilitators rather than primary drivers of health behavior change or relational dynamics (Patel et al., 2015). Importantly, the single-item wearable measure does not show whether respondents interpreted wearable outputs, incorporated them into self-understanding, discussed them with clinicians, or engaged in shared interpretation with providers. Moderation analyses revealed potential boundary conditions: digital literacy positively moderated the relationship between wearable use and patient-centered communication, while health data sharing positively moderated wearable associations with both communication and trust. These conditional effects are consistent with DCog’s premise that information becomes functionally meaningful only when embedded within shared systems of interpretation and coordinated action (Boyle et al., 2023; Hutchins, 1995), but they remain indirect. Higher digital literacy also produced a modest negative moderation on the wearable-trust relationship, suggesting a more critical orientation toward institutional information. Finally, no significant portal-wearable interaction emerged for PCC, indicating that combining multiple digital scaffolds does not automatically produce supra-additive relational benefits in real-world healthcare contexts.
Theoretically, these results refine EMT and DCog by specifying what can and cannot be inferred from large-scale survey data and by separating confirmatory from exploratory claims. The confirmatory findings show that portal use was associated with PCC in the opposite direction from expectations but was positively associated with institutional trust. The exploratory findings suggest possible boundary conditions involving partial offloading, compensatory adoption patterns, more critical engagement with institutional information, and sociotechnical coordination. For wearables specifically, the results should be understood as associations involving reported frequency of wearable use and potential exposure to patient-generated health data. Patient-centered communication and institutional trust emerge not merely from greater access or data volume but from the quality of coordination among tools, user capacities, and institutional practices. This more cautious interpretation addresses longstanding critiques of EMT regarding over-extension (Adams & Aizawa, 2008) and moves digital health scholarship beyond technosolutionist assumptions that equate adoption with improved outcomes (Blease et al., 2025). By integrating EMT’s focus on individual-artifact coupling with DCog’s emphasis on system-level coordination, the current study offers a measured account of how health information work is redistributed and sometimes strained in contemporary healthcare.
Practically, our findings caution against policy and design approaches that prioritize rapid technological adoption and access without parallel investments in communication infrastructure, interpretive support for patients, clinician training on integration of patient-generated data, and transparent data governance. Patient portals and wearables can meaningfully complement relational care when embedded in well-coordinated sociotechnical systems; they cannot substitute for high-quality interpersonal interaction or institutional responsiveness.
Several limitations warrant consideration. The cross-sectional design of HINTS data precludes causal inference, raising questions about whether portal use precedes or follows dissatisfaction with communication. In addition, the use of complete-case analysis may have introduced some degree of selection bias if missingness was systematically related to unobserved characteristics not captured in the comparison analyses. More importantly for the theoretical claims, HINTS does not include direct measures of cognitive extension, distributed cognition, cognitive load, interpretive routines, or in-situ patient-provider coordination. The wearable measure is especially limited because it is a single frequency item. It does not indicate whether wearable data were integrated into self-interpretation, clinical communication, bodily perception, or shared interpretation with providers.
Accordingly, the exploratory wearable findings should be treated as associations involving potential sources of patient-generated health data, not as evidence of continuous somatic extension or clinical integration. More generally, the exploratory RQ findings should be treated as provisional evidence for future hypothesis development rather than as confirmatory tests of the theoretical framework. Future research should employ longitudinal designs and mixed-methods approaches including ethnographic observation and qualitative interviews to trace how patients narrativize digital data over time, negotiate its meaning during clinical encounters, and integrate (or fail to integrate) these tools into everyday cognitive routines. Experimental studies could further test hypotheses generated by the exploratory patterns reported here.
Beyond these measurement limitations, certain concepts such as critical engagement with institutional information, sociotechnical coordination and distributed information work function as interpretive characterizations of observed patterns rather than directly measured constructs. Future research should develop and validate measures capable of assessing these processes directly.
Finally, because replicate-weight variance estimation was not implemented, the precision of standard errors and statistical significance tests may be imperfect relative to analyses that fully account for the HINTS complex sampling design. Therefore, moderation findings should be interpreted as provisional and in need of replication using design-based variance estimation procedures. Future analyses should employ HINTS replicate weights and design-based variance estimation procedures.
Ethical considerations
This study involved the analysis of publicly available, user-generated content sourced from the National Cancer Institute between September 16, 2024. As the research did not involve human intervention, direct interaction with users, or the collection of private, non-public information, formal ethics committee review was waived by the IRB of the University of Oklahoma . To ensure ethical compliance and protect user privacy, no personal identifiers were retained in the dataset, and any quoted material within the manuscript has been thoroughly anonymized.
Data availability statement
The data that support the findings of this study are openly available at https://hints.cancer.gov/?utm_source=chatgpt.com.
Notes
Consent for publication
Not applicable. The manuscript contains no individual user data, no direct quotations from identifiable persons, and no images or other identifiable material.
Funding Information
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.