Primary Care Practice Factors Associated With Telehealth Adoption in the United States: Cross-Sectional Survey Analysis

Introduction

In 2021, the National Academy of Medicine called for further study of telehealth to help strengthen primary care as part of their “Implementing High-Quality Primary Care in the United States” report [National Academies of Sciences, Engineering, and Medicine. Implementing High-Quality Primary Care: Rebuilding the Foundation of Health Care. Washington, DC. The National Academies Press; 2021. 1]. Past studies highlight a range of patterns in telehealth use by patient-level factors [Colburn D. The impact of telehealth expansion on health care utilization, access, and outcomes during the pandemic: a systematic review. Telemed J E Health. May 2024;30(5):1401-1410. [CrossRef] [Medline]2], but primary care practice–level capabilities, such as resources to enable telehealth, have not been systematically assessed. Organizational and contextual factors, such as a federally qualified health center (FQHC) designation and a practice’s neighborhood characteristics, may also be associated with telehealth use. We analyzed a national survey of US adult primary care practices to examine relationships between practice-level factors and telehealth adoption.

Methods

Overview

We analyzed cross-sectional data from the 2022-2023 National Survey of Healthcare Organizations and Systems, which surveyed practice leaders about telehealth use in 2021, practice payment models, and care delivery structures and processes (1245/3499 people responded to the survey, for a 36% response rate). Survey administration information and data sources are described in

Multimedia Appendix 1

Further information on survey administration, analytic datasets, weights, and missing data.

The primary outcomes are the proportion of all outpatient visits done via telehealth (real-time audiovisual or audio-only, eg, telephone) and the proportion of telehealth visits done audio-only. We categorized practices by quartile of telehealth use to compare practice-level differences by characteristic. We estimated multivariable linear regression models for all cases (N=1071) and used average marginal effects to generate estimates and 95% CIs for the primary outcome measures. Model covariates included practice ownership, clinician staffing, FQHC status, and telehealth-enabling resources. All analyses were completed in Stata (version 17; StataCorp), and all models used robust estimates of variance and 2-tailed P values, with P<.05 set as the threshold for significance. Weights were used in all analyses to account for nonresponse probabilities, detailed in

Multimedia Appendix 1

Further information on survey administration, analytic datasets, weights, and missing data.

Ethical Considerations

This study was deemed exempt by the Dartmouth College Committee for the Protection of Human Subjects (00032337). As practices were the unit of analysis, participants received an information sheet in lieu of informed consent. Practices were deidentified prior to analysis.

Results

The median use of telehealth for practice encounters was 20% (IQR 10%-35%), and the median proportion of audio-only telehealth visits was 29% (IQR 17%-50%). High-telehealth-use practices were more likely to care for a high proportion of uninsured patients, to have expanded broadband access for patients, and to have trained patients to use telehealth (Table 1).

In multivariable regression analyses (Figure 1), a high payer mix of uninsured patients and high broadband expansion corresponded to higher telehealth use (P=.02 and P=.008, respectively), while rurality corresponded to lower telehealth use (P=.008). Audio-only telehealth use was higher among FQHCs compared to other practices (P=.02), and assisting patients with using telehealth was associated with both higher telehealth use and a smaller proportion of audio-only telehealth use (P=.002 and P=.001, respectively).

‎

Discussion

In this national study of primary care practices in 2022, we found that respondent practices serving the uninsured reported greater use of telehealth for patient encounters, and FQHCs used more audio-only telehealth. This suggests that low socioeconomic status populations had a higher need for telehealth and that cuts to audio-only reimbursement would disproportionately impact care for such patients. The “digital divide,” systematic barriers for accessing and using technology and telehealth among various populations [Lythreatis S, Singh SK, El-Kassar A. The digital divide: A review and future research agenda. Technol Forecast Soc Change. Feb 2022;175:121359. [CrossRef]3], is a known issue for FQHC and rural populations [Uscher-Pines L, Sousa J, McCullough C, Dong S, Kapinos K. Telehealth visits in health centers serving low-income patients in California: final results from the Connected Care Accelerator Initiative (2022-2024). RAND Corporation. 2024. URL: https://www.rand.org/pubs/research_reports/RRA3468-1.html [accessed 2025-03-13] 4,Ko JS, El-Toukhy S, Quintero SM, Wilkerson MJ, Nápoles AM, Stewart AL, et al. Disparities in telehealth access, not willingness to use services, likely explain rural telehealth disparities. J Rural Health. Jun 2023;39(3):617-624. [FREE Full text] [CrossRef] [Medline]5], and our study provides national evidence that these practice-level factors are significantly associated with lower adoption of video-based telehealth.

Importantly, we found that the telehealth-enabling practices of training and assisting patients with using telehealth and broadband expansion were associated with higher telehealth adoption. In contrast, electronic health record integration for video visits, participation in alternative payment models that incentivize care quality, and practice ownership were not significant practice-level factors, contrasting with a prior study suggesting health system integration was linked to higher telehealth use at the physician level [Cuellar A, Jena AB. Volume of care for primary care physicians in integrated vs independent practices through the COVID-19 pandemic. JAMA Health Forum. Sep 01, 2023;4(9):e232883. [FREE Full text] [CrossRef] [Medline]6].

Limitations include potential nonresponse bias due to the modest response rate, though we used weights to account for this (sensitivity analyses of missing data are in

Multimedia Appendix 1

Further information on survey administration, analytic datasets, weights, and missing data.

These findings provide important national data for the design of policy and practice interventions to expand telehealth use. Practices focused on enabling telehealth appear able to meaningfully increase its uptake [Anaya YB, Bañuelos Mota A, Hernandez GD, Osorio A, Hayes-Bautista DE. Post-pandemic telehealth policy for primary care: an equity perspective. J Am Board Fam Med. 2022;35(3):588-592. [FREE Full text] [CrossRef] [Medline]7]. Federally, renewing the lapsed support for broadband accessibility is an important means to address the digital divide [Boucher-Robinson S, Varn J. States reckon with lapse of the broadband affordable connectivity program. Pew Charitable Trusts. URL: https://pew.org/3MTXLw9 [accessed 2024-12-20] 8,Frieden R. Best Practices in Promoting Widespread and Affordable Broadband Service After the Covid-19 Pandemic. Social Science Research Network. URL: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4628562 [accessed 2025-03-13] 9].