85 Million Doctor Visits Have a Language Problem. AI Scribes Make It Worse.

Between 50 and 85 million ambulatory visits every year in the United States involve a patient who doesn't speak English fluently. The AI scribes being sold to those physicians were built for English-only visits. The math doesn't work.

The scale of the problem

The U.S. Census Bureau estimates that approximately 25.7 million Americans — around 8% of the population ages 5 and older — have Limited English Proficiency (LEP), defined as speaking English "less than very well."

But in healthcare, the effective encounter rate is significantly higher than 8%. LEP populations tend to have older age profiles, higher rates of chronic disease, and larger multi-generational households — all of which drive more frequent primary care, pediatric, and obstetric visits. The result: between 50 million and 85 million ambulatory office visits per year in the United States involve a language barrier.

Across approximately one billion total annual ambulatory care encounters in the U.S., that's 6% to 9% of all outpatient visits nationally — and it concentrates heavily in specific geographies and practice types.

Where these visits concentrate

The national average of 6–9% understates what individual practices experience. Nearly 60% of all non-English medical visits in the U.S. are concentrated in four states:

For safety-net clinics and Federally Qualified Health Centers (FQHCs), the number is even more stark: 20% to 40%+ of all ambulatory visits in community health center settings involve non-English or LEP patients. For private independent practices in urban areas, it typically runs 3–6% nationally — but that average obscures enormous variation by neighborhood.

A family medicine physician in a Los Angeles suburb, a Houston primary care clinic, a Miami pediatrics practice — they're not seeing 8% non-English visits. They're seeing 30%, 40%, or more.

What this does to a physician's day

Every visit requiring a medical interpreter — whether in-person, telephonic, or via video services like LanguageLine — takes 1.5x to 2x longer than an English-congruent visit. A standard 15-minute primary care slot becomes a 25–30 minute appointment. In a fully booked day, that compression cascades.

The documentation overhead doesn't disappear at the end of the visit either. It follows the physician home. Notes that required an interpreter to conduct still need to be charted in English. That charting happens after hours, on top of everything else.

Why AI scribes fail this problem

Most ambient AI scribes on the market were built for a single-speaker, English-language visit. The physician speaks, the AI transcribes, the note is generated. That model breaks in three distinct ways when a language barrier is present:

The languages involved

Spanish accounts for approximately 63% of all non-English medical visits nationally — roughly 35–50 million visits per year. But the remaining 37% matters too:

The compliance angle

Section 1557 of the Affordable Care Act requires healthcare providers receiving federal funds to provide qualified language assistance services. That obligation exists regardless of what AI tools are or aren't available. Physicians serving LEP patients are already legally required to bridge the language gap — the question is whether their documentation workflow reflects what actually happened in the visit, or only the English-language portions of it.

What this means for independent practice

Large hospital systems can afford to absorb the operational cost of language barriers — dedicated interpreter services, extended scheduling blocks, documentation review staff. Independent practice physicians carry this burden personally.

Every non-English visit that requires after-hours charting is time the physician didn't get back. At 6–9% of all visits nationally — and significantly higher in urban and safety-net settings — the aggregate is not small. An independent physician seeing 25 patients a day in a linguistically diverse neighborhood may be running 3–5 interpreter-assisted visits daily. Every one of those visits is a documentation problem that follows them home.

How ambient AI scribing actually works — and where the language problem gets solved

This guide explains the technology behind ambient AI scribing, what separates tools that handle multilingual visits from those that silently fail them, and what to look for before committing to a vendor.

Built for every visit. Not just the English ones.

NoPajamaTime works in 100+ languages. Every visit ends with a structured English clinical note — regardless of what language the appointment happened in. No interpreter confusion. No missing clinical content. No after-hours translation work.