Doctors performing surgery.

Doctors performing surgery in Brussels, Belgium. Photo by Piron Guillaume on Unsplash.

Dr. Robot

Conrad Amenta

New software is industrializing medicine by turning doctors into data entry clerks—and making them suicidally depressed in the process.

In more ways than one, medicine is dying.

A 2015 article in JAMA: The Journal of the American Medical Association suggests that almost a third of medical school graduates become clinically depressed upon beginning their residency training. That rate increases to almost half by the end of their first year.

Between 300 and 400 medical residents commit suicide annually, one of the highest rates of any profession, the equivalent of two average-sized medical school classes. Survey the programs of almost any medical conference and you’ll find sessions dedicated to contending with physician depression, burnout, higher-than-average divorce rates, bankruptcy, and substance abuse.

At the risk of sounding unsympathetic, medicine should be difficult. No other profession requires such rigorous and lengthy training, such onerous and ongoing scrutiny, and the continuous self-interrogation that accompanies saving or failing to save lives.

But today’s crisis of physician burnout is the outcome of more than just a job that’s exceptionally difficult. Medicine is undergoing an agonizing transformation that’s both fundamental and unprecedented in its 2500-year history. What’s at stake is nothing less than the terms of the contract between the profession and society.

The Rise of the Electronic Medical Record

An electronic medical record, or EMR, is not all that different from any other piece of record-keeping software. A health care provider uses an EMR to collect information about their patient, to describe their treatment, and to communicate with other providers. At times, the EMR might automatically alert the provider to a potential problem, such as a complex drug interaction. In its purest form, the EMR is a digital and interconnected version of the paper charts you see lining the shelves of doctors’ offices.

And if that’s all there were to it, a doctor using an EMR would be no more worrisome than an accountant switching out her paper ledger for Microsoft Excel. But underlying EMRs is an approach to organizing knowledge that is deeply antithetical to how doctors are trained to practice and to see themselves. When an EMR implementation team walks into a clinical environment, the result is roughly that of two alien races attempting to communicate across a cultural and linguistic divide.

When building a tool, a natural starting point for software developers is to identify the scope, parameters, and flow of information among its potential users. What kind of conversation will the software facilitate? What sort of work will be carried out?

This approach tends to standardize individual behavior. Software may enable the exchange of information, but it can only do so within the scope of predetermined words and actions. To accommodate the greatest number of people, software defines the range of possible choices and organizes them into decision trees.

Yet medicine is uniquely allergic to software’s push towards standards. Healthcare terminology standards, such as the Systematized Nomenclature of Medicine (SNOMED), have been around since 1965. But the professional consensus required to determine how those terms should be used has been elusive.

This is partly because not all clinical concepts lend themselves to being measured objectively. For example, a patient’s pulse can be counted, but “pain” cannot. Qualitative descriptions can be useful for their flexibility, but this same flexibility prevents individual decisions from being captured by even the best designed EMRs.

More acutely, medicine avoids settling on a shared language because of the degree to which it privileges intuition and autonomy as the best answer to navigating immense complexity. One estimate finds that a primary care doctor juggles 550 independent thoughts related to clinical decision-making on a given day. Though there are vast libraries of guidelines and research to draw on, medical education and regulations resist the urge to dictate behavior for fear of the many exceptions to the rule.

Over the last several years, governments, insurance companies, health plans, and patient groups have begun to push for greater transparency and accountability in healthcare. They see EMRs as the best way to track a doctor’s decision-making and control for quality. But the EMR and the physician are so at odds that rather than increase efficiency—typically the appeal of digital tools—the EMR often decreases it, introducing reams of new administrative tasks and crowding out care. The result is a bureaucracy that puts controlling costs above quality and undervalues the clinical intuition around which medicine’s professional identity has been constructed.

Inputting information in the EMR can take up as much as two-thirds of a physician’s workday. Physicians have a term for this: “work after clinic,” referring to the countless hours they spend entering data into their EMR after seeing patients. The term is illuminating not only because it implies an increased workload, but also because it suggests that seeing patients doesn’t feel like work in the way that data entry feels like work.

The EMR causes an excruciating disconnect: from other physicians, from patients, from one’s clinical intuition, and possibly even from one’s ability to adhere faithfully to the Hippocratic oath. And, if the link between using a computer and physician suicide seems like a stretch, consider a recent paper by the American Medical Association and the RAND Corporation, which places the blame for declining physician health squarely at the feet of the EMR.

Drop-down menus and checkboxes not only turn doctors into well-paid data entry clerks. They also offend medical sensibility to its core by making the doctor aware of her place in an industrialized arrangement.

From Snowflake to Cog

Physicians were once trained through an informal system of apprenticeship. They were overwhelmingly white and male, and there was little in the way of regulatory oversight or public accountability. It was a physician’s privilege to determine who received treatment, and how, and at what cost.

Supernatural justifications for treatment techniques eventually ceded to pseudoscientific ones; prayer was replaced by bloodletting and cocaine (and more prayer). Wilhelm Fliess engaged in surgical trial-and-error on his collaborator Emma Eckstein. His friend Sigmund Freud institutionalized female hysteria. Franz Joseph Gall performed backbends to legitimize racism via phrenology.

Then, in 1910, the Flexner Report caused a paradigmatic shift in medical education. Abraham Flexner was not a doctor, but a secondary school principal from Louisville, Kentucky, who later joined the Carnegie Foundation for the Advancement of Teaching. It was there that he wrote “Medical Education in the United States and Canada,” and transformed the lives of millions of people.

The Flexner Report recommended that medical education develop an evidence-based curriculum. Under its influence, medicine was subjected to the rigors of peer review and the scientific method for the first time. Residency programs were established, uniting the university and the hospital, and placing apprenticeship within the academy. Medical teachers were expected to be proponents of the latest and most credible research. State licensure was tied to education, introducing some semblance of standards.

The recommendations in the Flexner Report also formed the basis of what we today understand as the social contract between the medical profession and the people whom it serves. Patients are entitled to competence, altruism, morality, integrity, accountability, transparency, objectivity, and promotion of the public good. In return, physicians are entitled to trust, autonomy, self-regulation, a funded healthcare system, inclusion in public policy, monopoly, and prestige.

In the intervening years, the tenets of physician prestige and self-regulation have remained intact. But the introduction of computerization has begun to rewrite the social contract between doctors and society, as EMRs lay the groundwork for the industrialization of medicine.

Industrialization is the premise that people working together in a coordinated fashion will work more efficiently than one person doing everything themselves. To achieve this coordination requires standardization (the wheel goes on the car the same way every time); a technological innovation that makes work as simple as possible (an assembly line with power tools); and cheap labor (poor people).

An expert dressmaker may have once been responsible for every aspect of their craft: designing the dress, procuring the fabric, cutting and stitching, marketing and selling. Some dressmakers might be particularly good at one or more of those things. A few might even be good at all of them. But even in the best-case scenario, the quality of the dresses and the rate of their production will vary wildly.

Dressmaking is the kind of thing that’s easy to industrialize. The pieces of the process can be categorized, standardized, and delegated. The language we use to refer to the parts of the dress, and the tasks associated with the job, are clear. Reducing the qualifications for participation in dressmaking renders individuals interchangeable and disposable.

Industrialization has been applied to almost every field in which something is produced and sold. Now, EMRs are applying it to medicine. In the industrialized conception of medicine, as in the industrialized conception of all professions, more tasks become routine, and routine tasks are delegated downward. It’s no surprise that in the health policy world the introduction of EMRs often accompanies a discussion about hiring less educated professionals, like nurses and pharmacists. Meanwhile, fewer and fewer spaces are designated as safe for creativity and intuition, because these are considered unpredictable and unreliable.

Winners and Losers

One wonders if it’s possible to carve out a third way between the purely intuitive and the mechanically standardized. Atul Gawande has written extensively about this possibility, depicting a meeting of minds between autonomous doctors and health systems designers—and he manages to do so without making it seem terrifying or fantastical. In this world, technologies might seek to complement and enhance, rather than replace, the physician’s ability to incorporate research into practice.

Natural language processing and dictation will allow physicians to use any words they like while recording notes into an EMR, as opposed to drop-down menus and pick-lists. Artificial intelligences like IBM’s Watson will comb through research on behalf of the physician and aid in clinical decision-making. The doctor’s lounge, an increasingly rare phenomenon, is a basic form of technology that allows physicians to connect and share information. Not all innovations need to be bleeding edge.

But reform is big business. The “eHealth” industry, which produces the infrastructure with which the square peg of medicine will be crammed into the round hole of scalable technology, is estimated to reach $308 billion by 2022, and is a key driver of America’s $3 trillion national healthcare expenditure. The Healthcare Information and Management Systems Society (HIMSS) Annual Conference & Exhibition—the biggest eHealth conference in the world—was attended by just over 43,000 people in 2016. The allure of a disruptive solution that will tidily rationalize medicine has too many short-term winners to question—even if those winners are neither physicians nor patients.

Conrad Amenta has written about play for Kill Screen magazine and music for Cokemachineglow. He lives in San Francisco where he works as a health policy researcher.

This piece appears in Logic's issue 1, "Intelligence". To order the issue, head on over to our store. To receive future issues, subscribe.