This study used a cross-sectional observational design using secondary analysis of nationally representative survey data linked to area-level contextual measures. The primary aim was to identify individual and contextual determinants of telehealth use and modality choice among US adults in 2022. We report this cross-sectional observational study in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for cross-sectional studies [29].

This study analyzed the 2022 HINTS 6 restricted-use deidentified dataset. The University of Arkansas for Medical Sciences Institutional Review Board determined this study did not constitute human participants research (protocol 297801). The original HINTS data collection was approved by the Westat Institutional Review Board, and all survey respondents provided implied informed consent by completing and returning the questionnaire. The restricted-use dataset contained no direct identifiers; geographic information was limited to county-level codes. Respondents received no compensation for the secondary analysis; compensation for the original HINTS survey included a US $2 preincentive and an opportunity for a US $25 postsurvey incentive. No images containing identifiable individuals were included in this study.

Table 1 summarizes characteristics by telehealth modality. Among 6252 respondents in HINTS 6, 1641 reported any video telehealth, 876 reported exclusively audio-only telehealth, and 3529 reported no telehealth in the past 12 months. These correspond to survey-weighted prevalence of 27.2% (95% CI 25.5%‐29.1%), 12.1% (95% CI 10.9%‐13.4%), and 60.7% (95% CI 58.6%‐62.7%), respectively. Video users were younger on average (52.8, SD 16.2 y) than those with no telehealth (56.3, SD 17.1 y) and audio-only users (58.6, SD 16.8 y). Women comprised a larger share of video (1071/1641, 65.3%) and audio-only (566/876, 64.6%) users than nonusers (2012/3529, 57%). Compared with nonusers, video users more frequently reported higher income (788/1641, 48% vs 1292/3529, 36.6% at ≥US $75,000), current employment (912/1641, 55.6% vs 1807/3529, 51.2%), and insurance coverage (1559/1641, 95.1% vs 3165/3529, 90%). Digital readiness was notably higher among video users: any smart device ownership (1582/1641, 96.4% vs 3091/3529, 87.6%), health or wellness app use (1103/1641, 67.2% vs 1441/3529, 40.8%), and online health information seeking (1367/1641, 83.3% vs 2294/3529, 65%). Satisfaction with home internet for health needs was also higher in the video group (1075/1641, 65.5% extremely or very satisfied vs 1789/3529, 50.7% among nonusers). Geographic composition differed across modalities, with video visits more common among residents of large central metros (684/1641, 41.7%) and large fringe metro or suburban areas (393/1641, 23.9%) compared with nonusers (1220/3529, 34.6% and 769/3529, 21.8%, respectively). White respondents made up a similar share across video, audio-only, and no telehealth groups, while Hispanic respondents were somewhat more represented among audio-only users than among video users or those who did not use telehealth.

In a nationally representative survey of US adults in 2022, we found that overall telehealth use (any vs none) showed large regression effect sizes for health needs (eg, disability, +22.5 pp) and insurance coverage (+11.5 pp), whereas video vs audio-only modality choice among telehealth users was most strongly differentiated by digital readiness (device access and satisfaction with home internet) and eHealth behaviors. Shorrocks-Shapley decomposition revealed distinct variance patterns: for any telehealth vs none, geography (19.7%), digital factors (25.7%), and health status (15.5%) all contributed substantially; digital domains explained the largest share of variation in video vs audio choice (40.4%), while geography and area context contributed most to explaining audio-only uptake (53.2% combined for audio vs none). Men were less likely than women to use telehealth, consistent with gender differences in health care utilization [51], and telehealth users with limited device capability were more likely to rely on audio-only visits. Hypertension was positively associated with telehealth uptake but not with video vs audio choice, potentially reflecting that patients with chronic conditions requiring frequent visits benefit from any telehealth modality without a strong preference for video.

This study advances prior work in 3 ways. First, we distinguish modality choice (video vs audio) from uptake (any telehealth vs none) in a unified, survey-weighted framework, showing that digital access+health literacy and eHealth domains together explain over 40% of the variance in video vs audio choice (40.4%) and one-third in video vs none (33.4%). Second, we explicitly separate individual digital behaviors (app use and health information seeking) from device ownership and satisfaction, demonstrating that behavioral engagement is strongly linked to uptake above and beyond device availability. Third, we link person-level data with area context (division, urbanicity, and county characteristics) and apportion explained variance across domains using a Shorrocks-Shapley decomposition adapted to complex surveys—offering decision makers or researchers a transparent view of the relative leverage points for potential interventions.

A key strength of this study relative to the telehealth disparities literature is the granularity of the person-level technology data available in HINTS 6, which allows us to operationalize digital readiness beyond proxies such as binary internet access or area-level broadband coverage alone. Specifically, HINTS captures multiple, policy-relevant dimensions of digital readiness—device access and ownership, perceived adequacy of home internet for meeting health-related needs, and behavioral engagement with digital health. We leveraged this richness in a two-level way to avoid inconsistent aggregation: (1) we modeled these measures as distinct indicators in the regression analyses to preserve interpretability and map directly onto different intervention levers, and (2) we grouped them into prespecified domains in the Shorrocks-Shapley decomposition to provide a coherent summary of the relative explanatory contribution of the broader digital readiness “stack” compared with sociodemographic, clinical, and place-based domains. Future work could apply psychometric methods to develop and validate a composite digital readiness scale. In contrast, our indicator plus–domain approach prioritizes actionable interpretability for policy or practice [52,53].

The decomposition analysis shows 3 patterns. First, the digital access+health literacy and eHealth consistently explain a large fraction of variation, but its weight is contrast-specific: it is largest for video vs audio and video vs none, smaller for any vs none, and smallest for audio vs none. Second, geography+urbanicity+area context is most influential for audio vs none, moderate for any vs none and video vs none, and least for video vs audio. Third, the combined health status and needs+health system access contribution is largest when the contrast is framed as use vs nonuse (audio vs none and any vs none), and smaller when the choice is video vs audio. Finally, because the effect size for social isolation was small and imprecise across modality contrasts, we interpret this as suggestive at best. It may reflect unmeasured factors (eg, mobility or transport barriers or care preferences) correlated with isolation rather than a direct effect.

Taken together, these patterns imply that (1) modality is largely a function of digital readiness and skills, while (2) adoption (any use vs none) is a joint function of clinical need, geography or place, and digital readiness. Despite modest R²—typical for behavioral use models—the decomposition points to clear, policy-addressable levers: device access and digital engagement for video-based care; and place-based strategies (alongside coverage and need) to broaden overall telehealth uptake.

A prior nationwide study in older adults shows that living in a neighborhood at the lowest tertile of broadband internet subscription is associated with a significantly lower likelihood of using telehealth [54]. We similarly observe strong associations between digital readiness and video use. Lacking a smart device is strongly associated with choosing audio over video but was not significantly associated with any telehealth vs none in adjusted models. In uptake models, app use and online health information seeking are positive, whereas device category is not significantly associated with any telehealth vs none. However, our adjusted county-level percent broadband access shows small associations with any and audio telehealth. This likely reflects differences between availability vs quality or speed and aggregation by counties, which might dilute neighborhood heterogeneity. Together, the findings underscore that closing gaps requires more than network presence; adoption and affordability are important. Our modality-specific results align: device richness and internet satisfaction push use toward video over audio, while those with fewer devices or lower satisfaction are more likely to rely on audio. This is consistent with a study that emphasized the importance of optimal broadband connectivity for improved mental health access and raised quality considerations as visual cues may provide useful clinical information and improve patient-clinician rapport/access to video-based care [55].

Furthermore, a recent study by Tilhou et al [56] reported that telehealth expansion exposed differences in the use of telehealth services that persisted even among beneficiaries with high-speed internet and that expansion of telehealth service or access to high-speed internet might not close gaps in use. Despite strong associations for digital access and eHealth behaviors, geography and social context still explain sizable variance. Thus, high-quality connectivity is likely important but not sufficient without attention to coverage, device access, and digital skills or engagement. Our domain-level results empirically support that digital inclusion is considered a “super” social determinant of health [37] as the device access and skills or engagement carry a substantial explanatory weight, while area context signals persistent place-based differences. The decomposition attributes important shares to geography and smaller but nontrivial shares to area context even after detailed individual controls, reinforcing that place effects remain operative for digital health adoption, consistent with a study conducted in Chicago which found that segregation and concentrated poverty influence why residents do not have home internet access [57].

Our findings point to complementary levers. First, sustaining insurance coverage and targeting populations with elevated needs (eg, disability and chronic conditions) may yield large absolute gains in telehealth use. Second, programs that upgrade device access and digital engagement (eg, app enablement and coaching in online health information seeking) should accompany broadband deployment; these address the “devices” and “skills or training” pillars alongside connectivity. Third, where home connectivity is constrained, offering high-bandwidth alternatives (community clinics and libraries) and maintaining audio-only options preserves access while longer-term infrastructure and affordability gaps are addressed. Finally, because geography and area context explain substantial variance, especially for audio telehealth–place-based investments (including affordability supports) are warranted. These implications are directly supported by our Shorrocks-Shapley decomposition, which quantifies that digital readiness is the dominant explanatory domain for video vs audio modality choice; whereas place-based factors play a larger role for audio-only uptake.

The overarching goal of telehealth is to democratize high-value care by overcoming geographic and logistical barriers. However, our findings indicate that this promise might remain unfulfilled. The significant differences identified in our nationally representative analysis reveal that telehealth, in its current form, may risk exacerbating the very differences it is intended to solve [58]. These results carry direct implications for policymakers, health care systems, and technology developers aiming to ensure the expansion of virtual care is available to all individuals.

In our data, “geography” primarily captures regional policy environments, market or provider supply for virtual care, as well as digital infrastructure and adoption. “Underserved” in this context refers to areas with lower broadband adoption or affordability and speeds, higher area deprivation, rural or micropolitan location, and limited provider supply for video-capable care. For audio only vs no telehealth, geography explained the largest share of the model’s variance—substantially more than in other contrasts—underscoring that a one-size-fits-all strategy is insufficient. The significant reductions in the probability of telehealth use by geography, for example, for residents of the Midwest and South compared to the Northeast point to potential structural issues. This aligns with research highlighting how state-level policies, particularly those governing reimbursement and physician licensure, create a fragmented regulatory landscape that hinders uniform adoption [59]. To address this, federal and state policies should prioritize targeted investments in underserved regions [60,61]. This includes not only expanding community infrastructure like broadband internet, a known barrier [60-64], but also incentivizing provider adoption through sustainable reimbursement models and advancing interstate licensure compacts to improve the supply of virtual care in these areas.

Overall, since affordability is a binding constraint for adoption, recent federal policy changes—including 2025 adjustments to Broadband Equity, Access, and Deployment program affordability requirements and the termination of Digital Equity Act grant programs—threaten to undermine gains in telehealth use, particularly in high-need regions where subsidized connectivity and digital inclusion supports have been critical [65,66].

Second, our analysis provides robust empirical weight to the emerging concept of digital determinants of health [63,64]. We found that the combined domains of health literacy and digital access are as important as geography in explaining telehealth use. Critically, this difference is not merely about access to a device but about the capability to use it effectively for health. We did not find consistent negative associations for low health numeracy after adjustment. Instead, app use and online information seeking were the behavioral factors most strongly linked to higher telehealth uptake. One interpretation is that these behaviors proxy digital readiness—familiarity with app-based health workflows similar to those required for scheduling, check-in, and joining video visits. Patient studies of video visits report that technical and setup challenges (eg, connecting to the visit, audio or video problems, and confusing initial setup) remain common and systematic reviews identify limited technical literacy and technology barriers as recurring obstacles to telehealth use in older adults. Accordingly, this pattern is hypothesis-generating and suggests that the usability burden of telehealth platforms could disproportionately deter patients with less prior digital health experience [5,8,9,67].

Overall, initiatives should address both digital access and digital literacy [68]. This requires a 2-pronged approach: funding for community-based digital health literacy programs and a push for user-centered design principles in the development of patient-facing technologies to reduce overall burden on users.

Cross-sectional data preclude causal inference and telehealth measures were self-reported and lacked encounter frequency or quality. County-level broadband access may mis-measure adoption, affordability, and speed (and operates at a coarser geography); collinearity with urbanicity and digital behaviors may attenuate or reverse associations for the county broadband variable. Model R² values are modest, indicating unmeasured determinants (eg, provider supply). Future research should examine finer gradations of educational measure to distinguish groups that may face distinct barriers to telehealth adoption. Nevertheless, the survey-weighted national scope, explicit modality contrasts, and Shorrocks-Shapley decomposition across domains provide a comprehensive, policy-relevant portrait of telehealth determinants in 2022. Furthermore, understanding intersectional perspectives across race, income, education, and broadband availability might be an area of fruitful research.

Video telehealth uptake in the United States reflects intersecting determinants with modality choice driven most strongly by digital readiness and overall uptake shaped by digital factors, place-based characteristics, and health needs. Policies that simultaneously advance connectivity, affordability, device access, and digital skills support are likely to narrow persistent gaps in telehealth use across modalities. Methodologically, this study advances the telehealth disparities literature by applying Shorrocks-Shapley decomposition to survey-weighted national data to quantify the relative explanatory contribution of determinant domains rather than examining predictors in isolation. Our approach provides a domain-level ranking of where disparities are most strongly concentrated, distinguishing digital readiness mechanisms from place-based constraints. Leveraging HINTS’ rich person-level technology measures, we show that digital access and eHealth behaviors account for a larger share of explained variation in video-vs-audio modality choice than traditional sociodemographic characteristics. These findings support a dual strategy: maintain reimbursable audio-only care to preserve access while digital constraints persist and pair sustained coverage with targeted investments in devices, affordable high-speed connectivity, and digital skills supports to enable more equitable access to video-based telehealth.