Executive Summary

• Telehealth widens access but can widen the language gap for LEP patients.
• The SPEAK Act (signed February 3, 2026) requires HHS telehealth guidance by February 2027.
• 29.6 million people in the US are limited English proficient (GSA, 2025).
• The failure mode is fragmentation, not modality.

‍

Most telehealth platforms are designed around an English-language patient journey. For patients with limited English proficiency (LEP), meaning those who speak English less than very well, that can create barriers before the clinical conversation even begins. Research in JAMA Network Open (Rodriguez and colleagues, 2024) found that patients with LEP use telehealth less often and report worse video-visit experiences than English-proficient patients. Contributing factors identified across this research include difficulty navigating video platforms, English-only instructions, unavailable translated portals and challenges integrating interpreters into video visits.

‍

What does the SPEAK Act require for language access in telehealth?

The Supporting Patient Education and Knowledge (SPEAK) Act, signed into law on February 3, 2026 as part of the Consolidated Appropriations Act, 2026, directs the Department of Health and Human Services (HHS) to develop and disseminate best-practice guidance on language access in telemedicine for LEP patients. Per the statute, that guidance must cover facilitating qualified interpreter access during telemedicine appointments, multi-person video calls that put clinician, patient and interpreter on one connection, accessible instructions for joining a visit and multilingual patient portals and materials.

This is the first time telehealth language access has been defined at the national level. It converts a set of good-practice recommendations into a federal reference point that federally funded providers will be measured against.

‍

When does the SPEAK Act guidance take effect?

HHS has one year from enactment, so the guidance is expected by February 2027. As of mid-2026 the required task force had not yet been convened and public consultation had not begun, which compresses the runway. Health systems that wait for the final language before evaluating their telehealth workflow will be reacting to a deadline rather than preparing for one.

‍

Which providers should prepare now?

Any provider that receives federal funding and delivers care virtually falls within scope. That includes hospitals, health systems, Community Health Centers (CHCs), FQHCs. The organizations in the strongest position are the ones already auditing how an interpreter joins a virtual visit today, before the standard tells them how it should work tomorrow.

‍

‍

Why does telehealth widen the language gap for LEP patients?

Telehealth widens the gap when interpreter access is a separate service the clinician reaches for, rather than a function of the visit itself. The patients that hit hardest are the ones with the least room to absorb friction: older adults, people on unreliable connections and LEP patients who have no English to fall back on. A virtual model that expects those patients to source their own interpreting effectively narrows who telehealth serves. With 29.6 million LEP individuals in the United States (GSA Translation and Interpretation Services Ordering Guide, 2025), this is not an edge case. It is a design decision every telehealth program is making, whether deliberately or by default.

‍

The failure is rarely the absence of any interpreting option. It is that the option lives outside the encounter. Common breakdowns include:

• Dialing a phone interpreter on a separate line: this adds latency and forces the provider to run two audio streams at once
• Leaning on a bilingual relative: the US Commission on Civil Rights advises against untrained interpreters because they introduce error and omission
• Launching a standalone video interpreting tool: this pulls the clinician out of the telehealth window into a second application
• Skipping interpreting altogether: common whenever language need is never captured at scheduling

‍

The California Health Care Foundation documented these access patterns in its March 2026 report on AI and language access, which frames the telehealth divide as an equity problem for Medi-Cal and safety-net populations rather than a convenience feature. Each of the breakdowns above shares one root cause: the interpreting layer and the care layer are two different systems.

‍

‍

What does language access look like when it lives inside the telehealth visit?

It looks like one system, not two.

The interpreting layer and the clinical encounter run in the same place, so the clinician is never bridging a separate tool under time pressure. In practice that means AI interpreting and human interpreters share a single interface, language support is arranged as part of setting up the visit and every utterance is captured for oversight and audit. 

‍

A telehealth program does not fail on whether an interpreter exists somewhere in the vendor stack. It fails on whether that interpreter is reachable inside the visit, in the seconds when the patient looks confused and the clinician has to decide whether to push forward or pause.

‍

Should AI interpreting be limited to routine visits?

Not as a fixed rule. Treating it that way misreads the evidence. A 2026 NEJM Catalyst study by Montoya Rubiano et al. of 23 Spanish-speaking surgical patients at Brigham and Women's Hospital found that patients did not rank AI and remote video interpretation (RVI), the standard of a live interpreter joining by video, as a cheap tier and a premium tier. They viewed them as complementary, each preferred in a specific context. AI was valued for speed and privacy in time-sensitive moments. Human video interpreting was preferred for emotionally complex conversations and cultural nuance. Capping AI at appointment confirmations and intake and routing everything clinical to a separate human channel imposes a hierarchy the patients in the study did not share.

‍

What accuracy and audit safeguards matter?

For clinical use, the safeguards that matter are utterance-level audit logs, human oversight and the ability to layer in the protocols a health system already runs. Compliance-literate leaders know HIPAA and SOC 2 Type II are necessary but not sufficient for medical interpreting. The audit trail is what lets a quality team review what was actually said. Human oversight is what catches the cases where nuance carries clinical weight. Those controls should be part of the interpreting platform, not reconstructed across three vendors after the fact.

‍

‍

How does interpreting pricing shape telehealth language access?

Pricing shapes it more than most programs admit, because per-minute billing puts a meter on every interpreted second. When interpreting is billed by the minute, a longer visit costs more. That cost sits in the back of a provider's mind during exactly the encounters that need unhurried communication. Growing telehealth volume multiplies that pressure across every virtual LEP visit.

‍

A flat monthly subscription removes the meter. It makes language access a predictable line item rather than an expense that grows with patient volume. In field deployments such as the 27-site Community Clinic NWA case, that model has delivered up 63% lower cost. On the Care Culture Talks podcast, emergency physicians have described how competing for scarce interpreter resources shapes triage in ways no vendor deck captures. Predictable pricing does not solve that alone. It does stop the budget model from discouraging the very use it is meant to enable.

‍

‍

The bottom line: language access into telehealth

Most health systems already have skilled human interpreters on site, moving between rooms for the complex emotional conversations that need them. The telehealth built around that in-person care needs AI interpreting to carry language access across every virtual visit and to scale what those interpreters already do. Done well, it runs as one integrated infrastructure: no separate login, no extra tool, no break in the flow of the visit. With HHS guidance due in February 2027, now is the time to design it deliberately. Reach out to plan language access across your telehealth. 

‍