I trained as a physician and spent the first part of my career moving between clinical practice and research, much of it in hematology. The cases that stayed with me were rarely the ones where knowledge was missing. They were the ones where reasoning had to hold up across a sequence of decisions made under uncertainty: what to ask first, what to test, what to rule out, and when to change course.
Today I work on tools that measure how physicians make those decisions. That work has changed how I read the current debate about continuing medical education (CME). In my view, it is being conducted almost entirely on the wrong axis.
Price is not the variable that matters
Free CME is now abundant. Federal agencies, specialty societies, industry-funded providers, reference tools, and media platforms all supply it. Much of this material is good, and none of it should be dismissed simply because it costs nothing.
Real physician voices, twice a week
Free, and one click to unsubscribe.
But price is the least interesting property of an educational activity. The expensive input is physician time, and an hour costs the same whether the activity was free or carried a fee. The real question is what that hour produces. For most CME today, the honest answer is a record that the hour took place.
A credit is a unit of elapsed time
This is not a criticism of accreditation. It is a description of the unit. One hour of participation yields one credit. The system was designed to document participation, and it does that well. In the United States, physicians still have to verify the administrative basics, and those checks matter:
- Confirm the activity is designated AMA PRA Category 1 by a provider accredited through the Accreditation Council for Continuing Medical Education (ACCME), a recognized state medical society, or Joint Accreditation.
- Confirm Maintenance of Certification (MOC) separately, because Category 1 credit carries no board points unless the activity was registered with the board.
- Confirm the topic satisfies current state mandates and, where relevant, the federal MATE Act: a one-time, 8-hour training on opioid or other substance use disorders that Drug Enforcement Administration (DEA) registrants must have completed before their first registration or renewal submitted on or after June 27, 2023.
- Confirm the expiration date.
- Read the disclosures.
Complete all five, and one fact has been established: The credit will be accepted. Nothing in that process tells us whether anyone’s reasoning improved.
Passive environments cannot reveal decision behavior
Most CME is passive by design. It rewards exposure, completion, and recall. Those things are easy to measure, which is precisely why they became the measures.
The evidence points the same way. A 2015 synthesis of systematic reviews by Cervero and Gaines found that CME does improve physician performance and patient outcomes, but that the effect is strongest for activities that are interactive, use multiple methods, and involve repeated exposure.
Clinical work, however, is not recall. A physician builds a differential, decides what to ask first and what to test, holds or narrows uncertainty, and revises when new signals arrive. Outcomes often depend on that pathway rather than on the final label. An environment that never forces a choice cannot show how choices are made.
This is why a physician can know a guideline thoroughly and still struggle when a patient does not present the way the guideline assumes. It is also why a well-produced module can be fully understood and still change nothing when the condition appears with competing signals, an incomplete history, and time pressure attached.
The gap is not a knowledge gap. It is a gap between what is taught and what is measured.
Format selection follows the incentive
Consider what the incentives produce on both sides. If credit tracks time, the rational choice for a physician behind on a cycle is the fastest available credit. The rational choice for a provider is an activity that is easy to finish. Both behave sensibly, and the system reliably documents attendance while telling no one whether care changed.
Topic selection follows the same logic. Physicians tend to choose activities in areas where they are already fluent, because those move quickly and confirm what they know. The genuinely uncomfortable topic, the one managed from partial memory a few times a year, is the one most often deferred.
That is exactly backwards. The deferred topics are where the distance between confidence and competence is widest, and physicians are poorly placed to judge that distance themselves. A systematic review in JAMA found that physicians’ self-assessments agree poorly with external measures of their competence, and that the least skilled and the most confident were among the least accurate.
In the United States, board certification now probes that distance more often than the traditional 10-year exam, for example through the American Board of Internal Medicine’s quarterly Longitudinal Knowledge Assessment, but still in a format that rewards test preparation. Outside those assessments, physicians receive very little information after residency about where their reasoning stands relative to their peers.
What an activity should be asked to produce
A more useful way to evaluate any CME activity, free or paid, is to ask whether it generates decisions and whether it returns information about them. I would suggest four questions.
- Did the problem arrive before the answer? A case that names the diagnosis in its title, or offers four options with one plausible distractor, has already done the difficult work. Patients do not arrive with a differential attached.
- Was anything compared? That could be against current evidence, against the physician’s own earlier reasoning, or against what peers in the same specialty did with identical information. A verdict of “correct” carries far less information than the specific point where a pathway diverged.
- Did the activity produce a moment of disagreement? An hour completed without a single instance of “I would have gone the other way” was probably chosen for comfort.
- Can the physician name, a week later, one thing they now do differently? That is the outcome that matters clinically. Accreditation asks providers to analyze change at the program level, but the system rarely shows an individual physician where their own reasoning diverged.
Activities that meet all four conditions exist. They are slower per credit, and they tend to make performance look worse before it improves. Every incentive in the current system, including the schedule of the person completing them, pushes the other way.
The standard should move
None of this is an argument against free CME. Wider access to accredited education is not a problem that needs solving, and some of the strongest material available costs nothing at all.
It is an argument for changing what the field claims to measure. Completion is a proxy that has been treated as an outcome for a long time. Decision behavior, meaning the observable pattern of how a physician reasons through cases over time, is much closer to what drives results for patients. And it is measurable, provided the environment is built to generate decisions rather than to deliver content.
Physicians are already careful about which credits count. The more useful discipline, and the harder one, is being equally careful about which ones teach.
Kseniia Zarubina is a physician who builds digital health products and a product and project manager at Eximion, an interactive, simulation-based continuing medical education (CME) web platform for specialized physicians that replaces passive lectures and webinars with realistic clinical case scenarios.
For more than eleven years, Zarubina has worked where medicine meets technology: AI, digital health, and clinical science. Zarubina leads cross-functional initiatives, translating complex scientific and business challenges into scalable digital solutions and driving product strategy, market intelligence, and executive decision-making across engineering, business, and clinical teams.
Zarubina is the author of fifteen peer-reviewed publications, has coauthored more than fifty scientific papers and conference abstracts, and teaches at the Faculty of Fundamental Medicine at Lomonosov Moscow State University. Eximion shares updates on Instagram.
