Skip to Main Content

“Insurers, drug companies and large HMOs are mounting an extensive effort to use medical data to decide which treatments are best, which doctors are best and which plans keep people the healthiest.”

That quote, seemingly ripped from today’s headlines, was published in the New York Times 25 years ago this month. I know because I was featured in the story, which began my (mostly) friendly jousts with regulators, administrators, and even some physicians fighting externally driven change — a group I refer to as the High Priests of Status Quo.

Since that article appeared, health care has seriously lagged behind in its use of analytics, which has transformed every other major industry, even as health care spending grew from 10% of gross domestic product to 20%, adding a $1.7 trillion burden on our nation.


Today, at long last, we are on the verge of realizing the article’s promise. I believe that wider use of medical scribes will get us there.

In 1994, I was part of a forward-looking team building new generations of health data in the commercial sector. Policy experts, understanding that analytics were the key to improving health care, sought our real-world knowledge. I still have a cool bag with a hand-embroidered Secret Service logo, my reward after a particularly long and painful session explaining what does and doesn’t work in the world of collecting data. The Health Insurance Portability and Accountability Act of 1996, about which I testified as an expert witness, was as much about permitting data collection as it was protecting privacy.


During the 1990s, my colleagues and I, along with many others, built amazing detailed databases on prescriptions and medical claims. While those databases have many important uses, they lack the clinical data essential to understanding patient outcomes.

After a lull during the administration of President George W. Bush, policy experts turbocharged clinical data collection via the Affordable Care Act’s sister, the Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009. It provided a brilliant carrot-and-stick incentive that migrated almost all physicians to electronic medical records that capture clinical data digitally, providing the foundation to enable analytics.

Today, large repositories of clinical data are being assembled in many places by many different types of companies and organizations. Largely unseen is innovation at medical specialty societies, where members contribute data from electronic medical records into quality registries. Dozens of registries exist, some for major diseases like addiction, Alzheimer’s, cancer, and heart disease, that will soon contain clinical data on more than half of all patients nationwide, almost in real time. This electronic information provides robust fuel to power artificial intelligence engines for health care.

Many large tech-savvy firms have jumped into health care only to quickly discover, painfully, that health care is unique. First, it’s not just about efficiency — the priority is better patient outcomes. Second, as Uwe Reinhardt, the legendary Princeton economist, famously observed, “Every dollar of health care spending = Someone’s health-care income,” and the High Priests of Status Quo protect their position.

But in the end it’s mostly about the data. And the quality of clinical data is abysmal. Hardly a week passes without a tech company or private equity firm calling me, desperate to find better data for their cool new AI engines.

This could have been, and should have been, foreseen: Physicians are not data-entry clerks. Electronic medical records require more entry time than physicians have, so important clinical information is often entered as typed or dictated notes. This unstructured data isn’t usable analytically. There also is a cultural disincentive: Many physicians regard electronic medical records only as a necessary evil to get paid, and see little clinical value for them in a work flow that’s closer to Marcus Welby than a 21st-century operation.

We will never achieve much-needed health care efficiencies until we get serious about fixing the data. Technology is helping, but not fast enough. Voice recognition is getting good, but it doesn’t yet produce structured data. Natural language processing and machine learning are reducing the time it takes to abstract medical records, but are still too imprecise and expensive. Besides, any quality expert would say fixing the data is inefficient; it’s more important to get it right from the start.

Fortunately, there is an answer we can implement now: partner physicians with medical scribes who enter information as structured data into electronic medical records.

I can hear the High Priests screaming, “That would be a huge new cost.” So let’s do the math: There are effectively 400,000 physicians who handle more than 85% of all patient visits. At $50,000 per scribe that’s $20 billion annually, which represents only 0.6% of health care spending. The savings in physician time alone is justification for the addition of scribes, especially with growing concerns about burnout. Better data quickly will pay back in a system in which 200 million insurance claims, about 1 in 7, are initially rejected every year, according to the Department of Labor.

Then there is the benefit of powering transformative analytics to deliver better, more personalized diagnostics and care, predictive analytics to prevent problems, and Six Sigma improved work lows to promote best practices and avoid waste.

What are we waiting for? Fix the data now!

Scribes are the last piece in the 25-year puzzle to remake health care. The data will power analytics to drive the final step necessary in any transformation: overthrowing the High Priests of Status Quo.

Physicians have more power over what they do than factory workers, supply-chain managers, or cashiers. But they are no less immunized against the use of data-driven outcomes and best-practice algorithms. Some denigrate this as “cookbook medicine,” but Betty Crocker gives me great cakes every time.

Robert Merold is managing director at Execullence, a health care consulting firm.

  • For scribes to be most effective they would have to be paid well and not just be pre-health students. Most scribes now are pre-PA and pre-med students looking for clinical exposure and experience hours for their applications, they aren’t staying in the field, nor are they paid well enough to make a career out of it even if they wanted to. Making scribing a longer term career would require better pay which would add expense to the cost of health care that is not addressed in this article. Would the data generated help cuts costs more than the increase in cost from paying for career scribes as an allied health profession?

  • Who are the scribes? Tell me if I’m wrong, but I thought they were students taking on the job temporarily before entering what they consider their true fields. You seem to be positing that “scribe” will become a lifelong career, so that the staff will learn all the different, complex systems they deal with at a deep level.

  • Just as much as the question : “How would you describe your pain on a scale of 1 to 10” gets subjective answers, the drive to gather “uniformized” data on a pre-set schematic may not result in relevant application to all humans. Health Care data garnered from in-person doctor-patient visits can often not be reduced to a “standard format” record keeping. Medical scribes do allow for more doctor-patient contact, but if forced to cookie-cutter recording (“check this box”, and “If A is applicable then complete 3-to-5 before continuing to B”) then those records become digitized, and do not actually represent the analog consult content. And how many more forms do we need in Health Care??

  • I started with a new PCP last year and was a little surprised to learn that she had an assistant that keeps her records and correspondence up-to-date. My former PCP was wonderful, she made sure I got the best treatment possible. But while talking to me she was always typing up her notes. My new doctor was able to leave all that up to her assistant and noticeably spent much more of my visit looking me in the eye as we talked.
    Another plus with this new doctor is that when I use their Patient Portal to contact her about a new prescription, request a referral, or a health concern, her assistant makes sure she sees my request quickly and gets back to me as soon as possible. My last doctor was teamed with a nurse who also filled this purpose, but great as she is, she had to fit this work in between her re
    gular nursing duties.
    I understand that these “scribes” need specialized training and bring up privacy concerns. But I have seen my medical records and due to time constraints, and at times the doctor’s handwriting, they were frequently missing details in my information. Plus, you can’t underestimate the impact on a patient when a doctor can concentrate on a face to face conversation with them.

  • At IT-S (Information Transformation Services), we are committed to providing quality, affordable, professional back office data entry and data processing services. We are the preeminent supplier of back office administrative services, including online and offline data entry services, data processing and data conversion. Data entry tasks are perfectly suited for offshore outsourcing, and our services can be customized to suit your project and support needs. For more information please visit

  • I can appreciate the enthusiasm of this article, but don’t share it. The EHR’s were designed for billing, at which they excel. They were not intended to advance medical knowledge, but only to process patients more efficiently from admission to discharge.

    Don’t look to EHR’s for insight or wisdom. Assigning some item a number does not make it a truth. Numbers by themselves, out of context, have little meaning. That’s why we end up with studies that reveal what’s been accepted clinical practice for decades. Anything more precise is unlikely because we are working from a limited grasp of what we should be measuring. The boundaries of medical knowledge does not stem from insufficient data, but from a flawed model.
    For example, the current medical model and EHR drop-downs would be a good match for a fleet of Toyotas. Even the comparison of data after repair might be able to detect who are the best mechanics. Why not? We built the machine and wrote the manual.
    We can’t say that for people.

    Peggy Finston MD

    • Thank you for validating the line in my column saying physicians see EMR as a necessary evil for billing but providing little clinical value for them. But that is mistaken: there are a wide range of use cases emerging from analysis of EMR data, including better diagnostics and treatment practices that will improve outcomes while lowering costs. Its why over 16 medical specialty societies are building Registries – so quality can improve and its members can be fairly reimbursed for the value they provide. Its why FDA now has a whole division for promoting use of ‘real world clinical data’. Better data capture will power all of these solutions – since doctors don’t have the time to enter its comparatively a minor expense and a huge ROI to employ scribes to do it

    • Again I appreciate your enthusiasm, but to expand on my last sentence:
      Humans are not a fleet of Toyotas. If you change the transmission in the Chevy and Toyota with exactly the same part, you won’t be successful. Chevy’s are not Toyotas. Again, we should know that. We have a manual that we wrote and can consult.
      Humans are different from another. Studies need to assume a group of people have the same constitution, psycho-social support, chronicity of illness, etc to test either a medication or a physician’s expertise. This can’t lead to meaningful data because it is based on an untruth, aka a lie. What this approach will accomplish is to encourage physicians to refer complicated patients.
      Peggy Finston MD

Comments are closed.