Clinical guidelines tell physicians when to reassess a chronic disease. They say far less about how to reason at each reassessment, which is where treatment escalation decisions are made. An illustrative rheumatoid arthritis case shows where that gap opens, and what the published evidence says about teaching and assessing that reasoning.
A woman in her early 40s with seropositive rheumatoid arthritis (RA) returns six months after her diagnosis. She is on the highest dose of methotrexate she can tolerate. Her hands are better, but not good. She still can’t open a jar in the morning, and her disease activity has sat in the moderate range for two visits in a row. The question seems simple: what do you do next?
She isn’t a real patient; she is an illustrative case written for this article. But that simple question is where the most interesting reasoning happens. The tempting move is to name the next drug, yet the moment isn’t really a choice of drug at all.
Escalation is taught as a single answer
As physicians, we tend to teach treatment escalation as a single answer. Board-style questions reward naming the correct next agent. Lectures present a ladder: conventional synthetic disease-modifying antirheumatic drug (DMARD), then a biologic or targeted synthetic DMARD, then a switch. Learners come away thinking the skill is knowing which rung comes next.
In practice, treatment escalation isn’t one choice. It is a pathway of smaller decisions, each made at a defined point and each informed by what the last one revealed. When we treat it as a single verdict, we skip the reasoning that makes the verdict sound.
Rheumatology already has a structure for timing. The latest European Alliance of Associations for Rheumatology (EULAR) recommendations call for monitoring every one to three months in active disease, and for adjusting therapy if there is no improvement by three months or the target has not been reached by six [1]. What a guideline can’t supply is the reasoning at each checkpoint. That is where clinicians most need practice.
Four questions before naming a new drug
When I break escalation into the decisions it actually contains, four questions come up at every checkpoint, before any new drug is named.
- Is the remaining activity truly inflammatory? Fibromyalgia, osteoarthritis and established joint damage can all keep disease activity scores high. Escalating immunosuppression won’t fix any of them. EULAR guidance on difficult-to-treat RA asks clinicians to confirm inflammatory activity before changing therapy for exactly this reason [2].
- Has the current therapy had a fair trial? That means an adequate dose, a route that works for the patient, real adherence and enough time [2]. Sometimes the right next step is switching methotrexate from oral to subcutaneous, not adding a second agent. In a randomised trial, subcutaneous methotrexate was more effective than the same dose taken orally [3].
- What does this patient bring to the decision? Pregnancy plans, infection history, and cardiovascular and cancer risk all change which branches are reasonable [1]. For a woman in her early 40s, pregnancy plans matter directly: methotrexate is not among the DMARDs considered compatible with pregnancy [4]. So do preferences about injections, infusions or pills, and how often someone can come to clinic. EULAR describes RA treatment as a shared decision between patient and rheumatologist [1].
- When is the next checkpoint, and what will we do if we miss it? A plan without that answer only postpones the next decision.

Confirming that remaining disease activity is truly inflammatory comes before any decision to escalate treatment. Image: AI generated (ChatGPT)
What the evidence says about teaching clinical reasoning
Consider what happens if she returns at three months still short of target. A clinician who planned the checkpoint and the next branch can move forward with confidence. One who simply added a biologic has to start reasoning from scratch, often without the information the first visit should have gathered.
Knowing the treat-to-target rule is not the same as acting on it. In a US registry of more than 27,000 patients with RA, one- to two-thirds of those with moderate or high disease activity had no change in treatment over the following year [5]. A test that only asks for the correct next drug cannot show whether a clinician would act at the right checkpoint.
Frameworks for evaluating continuing medical education (CME) make a similar distinction. A widely used outcomes model separates what clinicians know, what they can demonstrate in a controlled or simulated setting, and what they actually do in practice [6]. A question that asks only for the right drug name tests the first of these. The escalation pathway described here plays out in the later ones.
Reviews of the evidence point the same way. A synthesis of systematic reviews found that CME is more likely to improve physician performance and patient outcomes when it is interactive, uses several methods, involves repeated exposure over a longer period, and focuses on outcomes physicians consider important [7]. A scoping review of clinical reasoning assessment found that methods differ in which parts of reasoning they capture, and described how several can be combined into an assessment programme [8].
For anyone who designs, accredits or funds education, that suggests asking two questions of a programme rather than one. Did learners choose the right drug? And did they set a target, plan the checkpoint and decide in advance what would happen if it was missed?
Putting the pathway into practice
The practical lesson doesn’t require a course. At the escalation visit, ask the four questions out loud with the patient. Then document the target, the checkpoint date and the planned next branch in the chart. A documented checkpoint turns staying put into a choice to justify rather than a default. It also gives covering colleagues and primary care clinicians a map instead of a mystery.
For those who teach, the lesson is to give the follow-up visit as much attention as the first decision, because that is where escalation actually happens.
For the patient described here, a strong answer doesn’t start with a drug name. It starts with a question back to her: what does she want to be able to do again? Then it sets a target around that jar, confirms that her symptoms are inflammatory and her methotrexate has had a fair trial, books the three-month check, and decides in advance what happens if she misses it. The drug comes last.
Treatment escalation will always involve hard choices. But it is easier on everyone when we treat it as a road with signposts we agree on in advance, not a single fork with everything riding on it.
Author Bio















