On 27 October 2026, an ABIM research announcement highlighted new findings about endocrinologists’ knowledge and their adoption of diabetes prescribing guidance. It offers a useful starting point for professional learning: can you explain how a recommendation changes an actual treatment decision?
The study does not show that completing a question bank improves patient outcomes. It does make a specific knowledge-to-practice question worth examining. This iatroX article explains the finding and proposes a learning activity for clinicians who want to investigate their own uncertainty.
What the researchers measured
The JAMA Network Open serial cross-sectional analysis included 1,061 endocrinologists and 177,284 Medicare beneficiaries with type 2 diabetes. It examined 2017–2022 prescribing around the 2019 American Diabetes Association guideline update against subsequent first-year ABIM Longitudinal Knowledge Assessment scores from 2022–2023.
The adjusted increase among guideline-targeted versus comparison patients was 3.9 percentage points greater for endocrinologists in the highest versus lowest LKA score quartile. The corresponding association for GLP-1 receptor agonists was not statistically significant. These are associations in a particular specialist population, rather than evidence about every physician or every diabetes treatment.
Why the timing changes the interpretation
The scores followed much of the prescribing being examined. They cannot establish that LKA participation caused the earlier changes. A stronger score might signal knowledge that was already current; other clinician, patient and practice characteristics may contribute. This is a reason to investigate the relationship, rather than a trial of an educational product.
ABIM’s accompanying announcement presents the work as part of its research on continuing knowledge and care. Read the original methods before turning the headline into a claim about education. In particular, the study provides no estimate of the effect of using iatroX, and it did not measure an iatroX learning intervention.
Choose one prescribing uncertainty you can actually resolve
A practical response is to select one narrow decision for review. “Revise diabetes” is too broad to tell you when the work is finished. A better learning question might be: “Which patient characteristics would change the treatment recommendation I initially had in mind, and which source establishes that distinction?”
Write your provisional answer before opening the reference. Include the information you would still need from a clinical encounter. This gives the review a purpose: you can compare what you assumed with what the source actually requires. Use an original fictional scenario, or a fully de-identified learning question, rather than putting identifiable clinical information into a study tool.
The historical recommendation examined in the paper is not a complete prescribing guide for 2026. For US learning, begin with the current ADA Standards of Care. Clinicians elsewhere should select the guidance and formulary relevant to their jurisdiction. Record the source and version so that a later review can identify what has changed.
Build a short question session around the decision
Use questions to expose the boundary of your understanding. After committing to an answer, explain which finding determined your choice and why the most plausible alternative was less appropriate. Then check whether the explanation supports that reasoning. A correct selection accompanied by a weak explanation still gives you something to work on.
For example, create a learning note with three columns: the patient feature you noticed, the consequence you thought it had, and the evidence needed to confirm that consequence. If your mistake was overlooking the feature, practise identifying it in unfamiliar cases. If you recognised it but drew the wrong conclusion, spend the next session on that specific connection.
Keep an assistance record simple. Mark whether you answered independently, consulted a reference, or worked through an explanation with help. Those activities can all be useful, but they answer different questions about what you currently understand. When you return later, try an unfamiliar application before rereading the note.
Make the reflection useful to your next consultation
A reflection should describe an actual change in your understanding. “I completed ten questions” records activity. A more informative note explains the assumption you revised, the source that prompted the revision, and what you intend to check next time the issue arises.
You can also identify a practical barrier that knowledge alone would not solve. Perhaps your next action is to discuss a local pathway, confirm where a recommendation sits in the formulary, or clarify how the team communicates a follow-up plan. Keep that action distinct from your educational finding. Recognising a recommendation and implementing it within a service are related tasks with different requirements.
At the later review, ask whether you can now explain the decision without the original prompt. If you cannot, revise the note around the remaining difficulty. If you can, choose a different presentation that challenges the same principle. This is a proposed learning approach, not a protocol evaluated in the JAMA study.
A worked learning question: when A1C is not the whole decision
Consider a fictional adult with type 2 diabetes whose A1C is at their individual goal but who also has heart failure. A revision approach centred only on glucose could miss the reason to revisit treatment. ADA’s 2026 recommendation 9.8 recommends an SGLT2 inhibitor for adults with type 2 diabetes and heart failure, irrespective of A1C. The learning question is which treatment objective the comorbidity introduces.
Before reading an explanation, write two short answers: why might the existing glucose result leave an important treatment question unresolved, and what further information would you need before considering an individual medicine? Check kidney function, existing treatment, contraindications, adverse-effect risks, access and patient preferences. Consult the relevant agent’s prescribing information rather than treating every medicine within a class as interchangeable.
Next, change the case to chronic kidney disease. Identify the renal information you need and locate the applicable recommendation in the current guideline. Then compare it with a case involving established cardiovascular disease. Keep a small comparison note showing the comorbidity, intended benefit and information that still needs checking. The task is to connect the clinical feature with the purpose of treatment, rather than simply recall a drug-class name.
In a paired exercise, one learner can explain the reasoning while the other asks which detail would change the answer. Swap roles for the next case. Check the clinical conclusions against the source after each independent attempt, and retain any unresolved question for appropriate teaching or supervision.
Your reflection can now name a specific distinction: whether you previously treated the glycaemic target as the whole therapeutic goal, or failed to connect a comorbidity with a different objective. Revisit that distinction in an unfamiliar question later. Keep this current-guideline learning exercise separate from the historical prescribing association measured in the ABIM study.
Where iatroX fits
The iatroX ABIM question bank provides a place to practise internal medicine decisions, while Socratic Tutor can help you examine the reasoning behind an attempted question. The CPD workflow connects focused questions, Tutor-supported learning and your reviewed reflection into a professional learning record.
That record documents your learning; this article claims no automatic ABIM MOC points or accredited CME credit. The question sample offers up to ten new questions daily shared across exams, subject to availability. It provides a small opportunity to inspect the questions and explanations before deciding how they fit your learning.
Frequently asked questions
Does the study show that LKA participation improved prescribing?
No. It examined an association with subsequently measured knowledge scores. The design cannot establish that answering the assessment questions caused earlier prescribing changes.
Can the finding be applied directly to every specialty?
The study population was endocrinologists caring for Medicare beneficiaries with type 2 diabetes. Applying its estimate to other specialties, patient groups or countries would require further evidence.
Will an iatroX CPD record automatically earn ABIM credit?
No automatic MOC or CME credit is claimed here. Verify the status of the particular activity and the requirements relevant to your professional programme.
Sources and review
Reviewed 29 October 2026. Primary sources: Gray and colleagues, JAMA Network Open, 27 October 2026; ABIM research announcement; ADA Standards of Care. The suggested learning activity is editorial guidance, separate from the study’s findings.
Put one learning question into practice: start a focused iatroX CPD session and build a reviewed reflection around the uncertainty you want to resolve.