I am in the exam room, leaning forward hearing my elderly patient describe weeks of worsening fatigue. Her daughter fills the gaps as her mother, hard of hearing and anxious, struggles to find the right words. Her recent lab work shows a steep drop in her blood count. We are working together to understand what is going on. As I prepare to enter orders in the electronic health record (EHR), my screen suddenly fills with an alert: eight hierarchical condition category (HCC) diagnoses that have nothing to do with today’s visit, and I am supposed to acknowledge whether or not they still describe my patient’s health status. The interruption breaks my concentration and pulls me away from a frail patient who needs my full attention. We are already 25 minutes into a 15-minute follow-up appointment.

Administrative tasks are a growing time drain in primary care. While some of this tedious work still serves patient care, diagnosis coding, choosing descriptive phrases from an enormous list that describe patients’ illnesses to an extreme level of specificity, does not. It is clerical work masquerading as clinical responsibility, adding nothing meaningful to medical decision-making. Worse, it competes directly with a physician’s attention, posing a real risk to patient safety. Insurers administering Medicare Advantage plans have been especially interested in HCC coding. These codes are used to measure illness severity, and higher scores translate to larger payments from Medicare. This has led to several high-profile cases of fraudulent upcoding by insurers, with United Health Care among the most widely discussed. The susceptibility of HCC codes to manipulation has long been criticized. Far less attention has been paid to its impact on the physicians carrying out this work and our patients.

Health systems have sought to optimize HCC coding since accurate coding means more equitable payment from insurance companies. This is especially so in risk-based payment contracts such as accountable care organizations (ACOs). Because physicians know patients best, we are well positioned to identify and capture these codes. As a result, HCC optimization efforts increasingly target the physician’s workflow, like chart prompts, coding “nudges,” and even financial incentives to encourage participation. Diagnosis codes are legion; there are over 70,000 of them in the current U.S. ICD-10 coding system. Of these, around 7,700 codes map to 115 HCC categories. The more specificity in the coding, the greater chance that the diagnosis will be deemed HCC worthy. For instance, type 2 diabetes mellitus without complications is a non-HCC code, whereas type 2 diabetes mellitus with obesity makes the HCC list. Most patients in the Medicare age group live with many interrelated chronic conditions, each with its own expansive list of qualifying descriptors. Selecting the most accurate code often means sifting through a dizzying set of options. Risk adjustment itself makes sense, as sicker patients require more resources. But the current system encourages coding intensity over clinical relevance and shifts attention away from active listening and thoughtful problem-solving in the exam room.

There are some creative alternatives to risk adjustment’s current state. Growing efforts aim to incorporate claims data, information that does not rely on physicians, as well as EHR-derived data into risk adjustment models. These approaches would improve the integrity of disease severity assessment while reducing opportunities for manipulation. Equally important, they would begin to relieve physicians of a labor-intensive task that drains time, erodes focus, and diminishes the joy and purpose of patient care. The intrusion I experienced in that exam room was not an isolated annoyance. It was a symptom of a system that routinely pulls physicians away from patients to satisfy documentation requirements that serve the revenue cycle more than care. Fixing this must rise on the CMS priority list. Modernizing risk adjustment and decoupling it from direct physician coding would not only strengthen program integrity, but it would also preserve something far more valuable: the ability of primary care physicians to remain fully present with the people who rely on us.

Jeffrey H. Millstein is an internal medicine physician.