Using an Interpreter Is Great, but It Also Needs to Be Documented

— Here's yet another example of how automation can help us

A photo of a female physician listening to a medical interpreter
  • author['full_name']

    Fred Pelzman is an associate professor of medicine at Weill Cornell, and has been a practicing internist for nearly 30 years. He is medical director of Weill Cornell Internal Medicine Associates.

Perhaps there's a way we can get automation to help us eliminate some of the rote things we are asked to do that have little to nothing to do with the art and science of practicing medicine.

As we head into a major site visit for credentialing our institution, it turns out that one of the areas that has been detected as a challenge has been the documentation of the use of interpreters across our system in the electronic medical record.

As everybody knows, when someone speaks a language other than English, it is critically important that they have access to medically trained interpreters, because it's hard enough getting healthcare without not being able to understand what people are telling you. Speaking louder and slowly, and enunciating, just does not cut it.

Used But Not Documented

At our institution we have access 24/7 to highly capable medical interpreter services, through telephonic interpreters primarily, but also in-person interpreters when they are needed. We also have video console tablets in our practice that can connect us to alternative interpreters, including video interpreters for American sign language.

But over and over, quality improvement reviews have shown that the use of interpreters is not regularly documented. And we all know that there are times that folks fudge this, that everyone tries to squirrel around it, that everybody tries to make do. People will ask the one question they know in another language, and hope that the short and simple answer they get will be the end of it.

In our electronic medical record, the patient's preferred language can be listed as part of their chart, and this already leads to certain useful electronic fixes to make things safer and smoother. For instance, if the patient's preferred language is listed as Spanish, their prescriptions can be sent to the pharmacy with Spanish instructions, and their aftervisit summary will be translated into Spanish for them to read on printed materials or on the portal as a summary of their visit that day.

From the time they arrive at the building, through check-in and vitals and a visit with the nurse and interactions with the physician or nurse practitioner, to the checkout staff scheduling follow-up visits -- each moment it's in everybody's interest we make sure we're all on the same page. For the most part, I think we're all pretty good at availing ourselves of these services. But, for the life of us, it seems that we're just terrible about adding it into the patient's chart to show that we used an interpreter.

The administration has helped in some ways, by creating macros that have the preferred language, but at the time of the visit with the front desk staff, or with the nurse, or with the medical technician, or with the doctor, it seems that everyone has too much going on and we forget that we need to put this text into the chart.

Make It Automatic

I think a much more streamlined and effective system would be that if the patient has a preferred language listed other than English, that no matter what the clinical interaction was that was being created in the electronic health record, a series of choices would be automatically presented to the person creating that note, which they would have to fill out before they could proceed -- because this is the proof of the matter.

The choices could be simple:

  1. As the provider for this encounter, I am fluent in the patient's preferred language.
  2. A medical interpreter trained in the patient's preferred language was used throughout this encounter.
  3. The patient reports that, despite having a language other than English listed as their preferred language, they are comfortable proceeding and report understanding all of the clinical materials that were presented to them today.
  4. Patient declined the use of the hospital interpreter today, they are aware of their right to have access to these interpreter services, and instead elected to speak for themselves or to have a family member or friend or another person speak for them.

Something along those lines.

Since we have to get this stuff into the charts, and our track record thus far clearly shows that we're not good at always doing it, maybe we can build a system that always does it. This seems to be at least one answer to how we can get a lot of the stuff done that -- despite a lot of people with vested interests expecting and demanding we do it -- just never seems to happen.

Coding for Chronic Conditions

Another example is the use of hierarchical conditions coding (HCC). Apparently, every year Medicare seems to "forget" that people have a bunch of chronic medical conditions. The Medicare system seems incapable of remembering that patients have end-stage renal disease, morbid obesity, COPD, sleep apnea, severe peripheral artery disease status post amputations, and so many more problems.

To accurately reflect the complexity of the care we provide, we are being asked through best practice advisories in the chart to "remind" Medicare at least once a year what a patient's chronic medical problems are by adding on these HCC codes, even if we are not managing those conditions.

In the old days, before this mandate came into effect, providers were loath to add on medical problems to a bill to an insurance company for a medical condition they didn't address that day, because that was the very definition of medical fraud. And in the ensuing 7 years since this expectation has come into play, we've seen very low uptake of this adding on of additional codes by medical providers.

Once again, automation could be an answer, in the form of a simple data dump -- a checklist that is presented to a provider for confirmation that these conditions still exist -- that occurs at some time other than during our incredibly short office visit that is too often taken up by managing a patient's day-to-day problems that they're seeing us for.

One could even imagine a situation where the medical providers did not need to participate in this at all, that this data could be retrieved from the chart with the use of artificial intelligence to scrub the records and ascertain the truth.

These are just a couple of examples of ways that I think we could build a better system, that takes this stuff off the doctors' plates, and turn purely administrative tasks into something that we don't need to think about. I can imagine a system where screening questionnaires get sent out to patients, the right appointment type is made, previsit planning is done and healthcare maintenance options are queued up, appropriate referrals are auto-generated, all with a view to freeing up the doctors to practice medicine.

We spend enough time in the charts clicking boxes and typing. Maybe, for once, automation can make our lives better, instead of worse.