With 29.6 million people in the United States classified as having limited English proficiency in the 2023 American Community Survey, providing equitable care to non-English speaking patients is not a peripheral concern. It is a structural requirement of running a modern health system. Yet most institutions still treat language access as a single transaction, the interpreter call during the physician encounter, when the reality of a patient visit involves a sequence of distinct conversations across staff, departments, and time windows.
This is the touchpoints problem. And it is the reason that institutions with formally funded interpreter programs still produce inconsistent, fragmented experiences for LEP patients.
1. What are the language access touchpoints in a typical patient visit?
When a non-English speaking patient enters a healthcare facility, they do not have one conversation that needs interpretation. They have many. A typical outpatient visit can include up to eight distinct touchpoints, each requiring its own communication moment:
- Reception and front desk check-in
- Triage nurse intake
- Medical assistant rooming and vitals
- Nursing assessment
- Physician encounter
- Specialist consultation, if referred
- Diagnostic technicians, for imaging or labs
- Pharmacy pickup, billing, and discharge instructions
Each of these is a separate communication event, often with a different staff member, often in a different physical location, sometimes hours apart. Treating only the physician encounter as the moment that "needs an interpreter" leaves seven other conversations to be handled informally, in broken English, with a family member translating, or not at all. None of those alternatives meet the standard set by Section 1557 of the Affordable Care Act, which requires meaningful access at every step of care for individuals with limited English proficiency.
This is the structural mismatch between how interpretation is funded and how care is delivered.
"When we audit a non-English-speaking patient's visit, we don't find one interpretation gap. We find seven. The physician encounter usually gets covered. Everything around it doesn't." Eyal Heldenberg, Co-founder and CEO, No Barrier
2. Why does the on-site interpreter model break down across multiple touchpoints?
Having a dedicated on-site interpreter follow a patient through the entire visit is, on paper, the gold standard. Patients consistently rate it as the most desired option, and the continuity argument is real: one interpreter, one voice, one running context across the visit.
The research supports this in principle. A 2024 study published in the Journal of Primary Care and Community Health documented the actual time investment for on-site interpreters at a network of community health centers. The average clinical encounter with an interpreter was 47.7 minutes. The average total interpreter time, including the auxiliary support before and after the encounter (navigation, scheduling, follow-up calls, documentation), was 91.1 minutes per patient. The researchers concluded that longer interpreter time, particularly the auxiliary time, was associated with better patient-reported communication experiences.
That is the case for on-site interpretation. Here is the case against, in practice:
Operationally, on-site interpreters are expensive, scarce, and time-bound. A 2017 Mayo Clinic study published in BMC Health Services Research measured the mean wait time for an in-person interpreter in a busy surgical and procedural practice at 19 minutes. That is the mean. The variation was high, and 19 minutes of waiting compounds across multiple stops in a patient journey.
Off-hours coverage is worse. Most institutions cannot guarantee in-person interpreters for evening shifts, weekends, or low-volume languages. The result is a two-tier system inside the same hospital: scheduled daytime visits in Spanish get continuity, while a Sunday evening Haitian Creole speaker in the ED gets a phone call.
For the operational dynamics of this problem, see our deeper analysis in the waiting time aspect of deploying medical interpreters.
3. Why does the remote interpreter model break down across multiple touchpoints?
Remote interpretation services (Language Line, Propio, Martti, and similar) solve some of the on-site problems. Access is faster. Language coverage is broader. Off-hours availability is usually built in. For a single encounter, remote interpretation often works.
For a patient journey, it does not. Three specific failure modes show up at scale:
Cumulative wait time. Each touchpoint usually requires a new call: dial in, route the request, wait for an interpreter, hear the same disclaimer about confidentiality and the interpreter's role, then begin. The disclaimer alone runs 30 to 60 seconds. Multiply that across eight touchpoints, and you have added 4 to 8 minutes of pure overhead to the visit, before any clinical conversation happens.
Loss of continuity. A new interpreter joins at each touchpoint. None of them have the context from the previous one. The patient repeats their symptoms, their medication list, their concerns. The repetition is not just inefficient. It is a source of medical error, because key information sometimes does not get repeated, or gets repeated inconsistently. The 2024 Torresdey study found that auxiliary interpreter time (the work before and after the encounter) is what drives patient-reported communication quality. Remote interpretation systematically eliminates that auxiliary layer.
Inconsistent use under time pressure. Staff under workflow pressure (the front desk during a rush, a nurse doing a quick medication reconciliation, a tech running an imaging study) sometimes skip the interpreter call entirely. Not because they are negligent, but because the call itself is the friction. "Get by" interpretation, gestures, broken English, a passing bilingual colleague, becomes the default at exactly the touchpoints where the volume of LEP patients is highest.
Audio and connectivity issues compound these problems in clinical settings where remote interpretation depends on imperfect hardware. We covered this separately in audio and connectivity challenges with remote interpreters.
4. What does the touchpoints problem actually cost?
The downstream costs of unmanaged language access across multiple touchpoints are concrete and measurable:
- Extended visit duration. Cumulative interpreter wait time and repeated context-setting add meaningful minutes to every LEP visit.
- Provider and staff stress. The friction of arranging interpretation for every brief touchpoint contributes to what some researchers and clinicians have started calling LEP burnout, a pattern of cognitive load and workaround fatigue around language access.
- Medical errors from miscommunication. The conversations that happen without interpretation (medication explanations, pre-procedure instructions, discharge information) are exactly the conversations where errors carry the highest clinical consequence.
- Inconsistent quality of care. Two patients with the same condition but different English proficiency get materially different visits.
- Direct interpretation cost. Per-minute remote interpretation pricing means that institutions pay more, often substantially more, for the patients they serve worst.
For health systems building or evaluating a language access program, this is the cost equation that needs to be on the table. Not the per-minute interpreter rate. The full journey cost.
5. What does a touchpoint-aware interpretation model need to deliver?
A language access model that actually fits the patient journey needs to meet five operational requirements, not just one:
- Available at every touchpoint, not just the physician encounter. Reception, triage, nursing, diagnostics, pharmacy, and billing all require the same baseline access.
- Sub-minute time to interpreter. Anything above that creates the conditions for staff to skip the step.
- Continuity of context across the visit. The interpretation system needs to carry the patient's clinical context from one touchpoint to the next, so the patient does not become the repository of their own medical record.
- Coverage of off-hours and low-resource languages. Quality of care should not depend on what day of the week the patient arrived.
- A frictionless path to a qualified human interpreter when the encounter requires it (sensitive disclosures, complex consent, end-of-life conversations, mental health). This is what we call human escalation as a right, not an escalation pathway.
Neither traditional on-site nor traditional remote interpretation can hit all five at the volume and price point most health systems require. Which is why hybrid models, AI medical interpretation at the point of care, with human escalation built in, have become a serious operational option rather than a theoretical one.
This is the architecture No Barrier's AI medical interpretation platform was built around. Available on any connected device at every touchpoint. Available in 40+ languages with dialect depth (see our analysis on why three languages and many dialects matters more than total language count). Sub-minute connection time. Human escalation available at any moment of the encounter, by either the patient or the provider.
6. How should health systems evaluate a touchpoint-aware language access solution?
The right evaluation starts by mapping the actual touchpoints in the institution's patient journey, not the touchpoints the current interpreter contract is written around. A solution that handles the physician encounter well but cannot scale to reception, pharmacy, and diagnostics is still leaving the same gaps the existing model leaves.
We built a more detailed evaluation framework for healthcare leaders working through this decision in our executive checklist for choosing your next AI interpreter vendor.
If you would like to walk through your institution's specific touchpoint map with our team, we are happy to do that as a working session, not a sales pitch. You can book a 30 minute conversation here.
.svg){kind=small}