The Doctor Who Cannot Help You

Catherine Nguyen is 61, an internist in Akron, Ohio, and she has 1,640 patients. She has been practicing for 29 years. She is good at her job in the ways that matter: she listens, she remembers, she catches things. Last year she caught a drug interaction between a new cardiologist’s prescription and a medication her patient had been taking for six years, the kind of catch that requires knowing the patient and not just the chart. She went to medical school to do this work. She is still doing it. She is also drowning.

On a Wednesday morning, before her first patient arrives at 8:15, Catherine spends forty minutes on prior authorizations. She completes three. She has eleven more in the queue. Her staff handles most of the volume. She has two full-time employees whose entire job is arguing with insurance companies about whether the care she has already determined her patients need will be covered. Their combined salary is $94,000 a year. This is the cost of getting permission to practice medicine.

Catherine’s practice completes approximately 39 prior authorization requests per week. Her staff spends 13 hours on them. The 2024 AMA physician survey found these numbers are average. Average means that half of all practices handle more. Ninety-five percent of physicians in that survey said prior authorization contributes to burnout. Nearly one in four reported that a prior authorization delay led to a serious adverse event for a patient, including hospitalization, permanent impairment, or death.

Catherine is 61. She is tired. She has not updated her own Medicare comparison since she turned 60. She has been meaning to review her retirement plan. She goes home at night to the same kitchen table as her patients, with the same stack of bills, the same auto-renewing contracts, the same insurance decisions she has not revisited. She needs the AI too.

What Catherine Sees That You Do Not

#

When Loretta Simmons sits across from Catherine with five medications and a question about whether she is paying the right price, Catherine knows the answer is probably no. She knows patient assistance programs exist for Loretta’s Januvia. She knows Cost Plus carries the rosuvastatin at a fraction of the pharmacy price. She knows these things the way she knows many things about her patients’ financial realities: in passing, incompletely, without the time or the infrastructure to act on the knowledge.

Catherine’s electronic health record does not surface patient assistance programs. Her practice management software does not compare pharmacy prices. Her 15-minute appointment slot does not accommodate a conversation about how to navigate manufacturer enrollment forms. She is not reimbursed for the time it would take to research Loretta’s options. Medicare does not have a billing code for “helped patient find a cheaper source for her medication.” The work is real. The payment for the work does not exist.

So Catherine does what most physicians do. She prescribes the medication. She hands Loretta the prescription. She hopes the pharmacist will mention something. The pharmacist operates inside a dispensing system controlled by a pharmacy benefit manager whose pricing structure is opaque to both the pharmacist and the physician. Nobody in the chain is positioned to represent Loretta’s financial interest, and Catherine, who would if she could, cannot because the system she practices inside was not built for that function.

The Referral She Does Not Want to Make

#

When Catherine orders an MRI for Raymond Kozlowski’s knee, the referral defaults to the hospital system that employs her. Catherine became an employed physician seven years ago when her independent practice became financially unsustainable. The overhead of running a small practice, the billing staff, the compliance requirements, the malpractice insurance, the EHR system, the rent, exceeded what her patient revenue could support. She sold the practice to Akron Regional Health System. She kept her patients. She lost her independence.

Over 75% of U.S. physicians now work for hospitals, health systems, or corporate entities. The shift accelerated during the pandemic and has not reversed. For Catherine, employment means a stable salary, malpractice coverage, and an EHR she did not have to purchase. It also means that her referral patterns are shaped, formally or informally, by the system that employs her.

Catherine knows that the independent imaging center nine miles from the hospital offers the same MRI on the same class of machine for a fraction of the hospital’s price. She knows this because a patient told her last year, and she looked it up, and the patient was right. She does not refer patients there. She does not refer patients there because the hospital system’s referral workflow routes patients to affiliated facilities by default, because leaving the system creates administrative friction she does not have time for, and because she is aware, without anyone having said it directly, that a physician whose referral patterns consistently route revenue away from the system that employs her is a physician whose employment relationship may eventually be questioned.

This is not coercion. Nobody has threatened Catherine. The incentive structure is quieter than that and more effective. She refers to the hospital’s facility because the system makes it easy to refer to the hospital’s facility and difficult to refer elsewhere, and because her 22 patients today do not leave her the bandwidth to fight the default for each one.

The Documentation That Produces the Error

#

When Clarence Watkins receives a $14,000 bill with four coding errors, the errors trace back to Catherine’s documentation. Not her documentation specifically, but the documentation process she shares with every physician in the country. She writes clinical notes during and after the patient encounter. The notes describe what she observed, what she assessed, what she decided, and what she did. The notes are written for clinical purposes, in clinical shorthand, under time pressure.

A billing coder in a different department translates those clinical notes into CPT codes and ICD-10 diagnosis codes. The coder was not in the room. The coder is working from notes that may be incomplete because Catherine was running behind and documented the encounter in the parking lot before driving to her next appointment. The coder is processing dozens of encounters per day under productivity standards that reward speed. The translation from clinical narrative to billing code is where most billing errors originate, and the errors are not produced by incompetence on either side. They are produced by a communication system that was never designed for accuracy. It was designed for throughput.

Catherine does not see the billing codes that are generated from her notes. She does not review the charges that appear on her patients’ bills. She does not know that Clarence was charged for a six-hour recovery room stay when her surgical notes documented three hours. The billing department does not consult her. The appeals process, when Clarence’s daughter initiates it, does not involve Catherine either. She is the origin point of the clinical information that becomes the bill, and she has no visibility into or control over what happens to that information after she closes the chart.

Thirteen Hours a Week

#

The prior authorization burden is the most documented administrative cost in American medicine, and the numbers have gotten worse every year the AMA has measured them. Practices complete an average of 39 prior authorization requests per physician per week. The staff time consumed is 13 hours. Forty percent of physicians have employees who work exclusively on prior authorization. Eighty-nine percent say the process increases burnout.

Catherine’s two prior auth employees cost her practice $94,000 annually. That is the salary of a nurse practitioner who could be seeing patients. It is the salary of a care coordinator who could be calling Loretta about her Januvia, finding Raymond a cheaper MRI, and reviewing Clarence’s billing codes before they become charges. Instead, the $94,000 pays for two people whose job is to call insurance companies and wait on hold and submit forms and resubmit forms and appeal denials and reappeal denials, all to obtain permission for care that Catherine has already determined is medically necessary based on 29 years of training and clinical judgment.

The prior authorization system was designed to control costs. The AMA survey found that 87% of physicians report it leads to higher overall healthcare utilization, not lower. Patients whose treatment is delayed seek emergency care. Patients whose medications are denied try alternatives that do not work and return for additional visits. Patients who abandon recommended treatment because the authorization process exhausted them show up later, sicker, more expensive to treat. The system controls costs the way a dam controls water: it works until the pressure behind it exceeds the structure’s capacity, and then the flood costs more than the flow would have.

Catherine at Her Kitchen Table

#

Catherine goes home at 6:40 most evenings. She has a mortgage, two retirement accounts she has not consolidated, an auto insurance policy she has not compared in four years, and a Medicare future she has not planned for because planning for her own retirement feels like a project for someone with hours she does not have.

She is 61. The mean age of physicians in the United States is 51.8 years. Nearly a third of all licensed physicians are 60 or older. Forty-three percent are 55 or older. The workforce that cares for the nation’s aging population is aging alongside it, and the administrative burden that consumes Catherine’s professional hours also consumes the personal hours she might otherwise use to manage her own financial and healthcare decisions.

Catherine’s situation is her patients’ situation turned inside out. Her patients need agent technology to navigate the systems that extract from them. Catherine needs agent technology to navigate the same systems, plus the additional systems that consume her professional capacity. The prior authorization agent that files and tracks and appeals on behalf of her practice returns clinical hours to patient care. The buying agent that finds her patients’ medication alternatives returns trust to the physician-patient relationship, because the physician can say “your agent found a cheaper option” instead of “I wish I could help you with the cost but I don’t have time.” The billing accuracy agent that catches coding errors before they become charges protects both the patient and the physician from a system that neither of them designed and neither of them controls.

The Same Table

#

The series that follows this article describes twelve categories of financial transaction where agent technology can represent the reader’s interests against institutional systems optimized for institutional outcomes. Every one of those transactions has a physician on the other side who is also being consumed by the same institutional systems.

The physician who cannot mention the patient assistance program is not withholding information. She is operating inside a 15-minute appointment funded by a reimbursement model that does not pay for the conversation. The physician whose referral goes to the expensive facility is not profiteering. She is following the default workflow of the system that employs her because she does not have the time or the institutional latitude to fight the default for every patient. The physician whose documentation produces a billing error is not careless. She is documenting under time pressure for a clinical purpose, and the billing translation happens downstream in a system she does not see.

The reader who understands this will read the rest of this series differently. Not as a story about patients versus institutions, with the physician somewhere in the middle holding a clipboard and shrugging. As a story about two people at the same table, facing the same systems, needing the same tools, for different reasons that converge on the same point: nobody in this landscape has representation. The institutions automated their interests decades ago. The physician and the patient are both still navigating by hand.

Catherine is 61. She has 1,640 patients. She catches drug interactions. She listens. She is drowning in paperwork that was designed to control costs and instead controls her. She goes home to a kitchen table with its own pile of unreviewed contracts and uncompared insurance plans and unresolved financial questions. She needs the AI too. The series that follows is for her patients. This article is for her, and for the reader who will sit across from her next week and understand, for the first time, that the doctor is not on the other side of the problem. She is in it.