Turn a clinical guideline into measurable cohort criteria
Ask a guideline question, get an answer with cited evidence, and turn the recommendation into a written cohort definition that a clinician and an analyst can both read. This page ends where the criteria are agreed and the cohort is saved. Running the measure over that cohort is covered in Find and evidence care gaps in a patient population.
Guideline-Based Care Gaps for a Specific Patient Cohort
Who this is for
Clinical quality leaders, the CMIO office, guideline implementation committees, and the analysts who serve them. The reader here owns the question "what does this recommendation mean for our patients," not the outreach that follows.
Why it matters
A specialty society publishes an updated guideline. The surveillance imaging interval for a post-treatment population changes, or a treatment threshold moves. The quality committee needs an answer at its next meeting: how many of our patients now fall outside the recommendation, and who are they.
Getting that answer usually means a clinician reading the guideline, translating it into criteria, handing those to an analyst, waiting for a query, discovering the criteria were interpreted differently than intended, and repeating. The loop runs on calendar weeks while the guidance sits unimplemented.
Translation is where the loop breaks, and it breaks quietly. A recommendation written for clinicians mixes clauses a system can evaluate with clauses that need judgment, and nothing in the sentence marks which is which.
What you gain
A written cohort definition, with the guideline citation attached to it, that both halves of the committee can read. When the definition is explicit, a disagreement about the cohort becomes a disagreement about a stated criterion rather than an argument about what a query did. When the guideline is revised, you edit the criteria you can see instead of reconstructing what someone meant last year.
The guideline states the recommendation. Patient Journey Intelligence identifies which patients fall outside it and shows the evidence. Care decisions remain with the treating clinician.
Before you start
- Access to the Clinical Guidelines module and to Cohort Builder
- Patient-level data ingested for the population the guideline addresses, including the note types carrying the relevant findings
- The guideline your committee has adopted, in the version you intend to measure against
- A clinical owner who will approve the translated criteria before anyone acts on the output
Step 1: Ask the guideline question and review the cited evidence
- In the left navigation, go to Agents and Tools > Clinical Guidelines.
- Ask a question about the target condition or care topic, for example: "What monitoring is recommended for patients with diabetes?"
- Read the answer together with its cited sources, and open the citations that carry the recommendation you intend to measure.
- Confirm the population, the clinical action, and the timing interval the recommendation specifies.
- Record the guideline name, version, and citation. You will attach these to the cohort in Step 3.
Step 2: Separate the measurable clauses from the ones needing judgment
Read the recommendation one clause at a time and sort each clause into something the platform can evaluate or something a clinician has to define first. This is the step that determines whether the resulting number means anything.
The example below is illustrative. Use the wording from the guideline your committee has adopted.
| Clause in the recommendation | Becomes | Type |
|---|---|---|
| "patients with type 2 diabetes" | Condition criterion, with the concept set your team already uses | Measurable |
| "assess glycemic status at least twice per year" | Fewer than 2 HbA1c results in the trailing 12 months | Measurable |
| "at least quarterly when therapy has changed" | Interval greater than 92 days, applied only after a medication change event | Measurable, if medication data is complete |
| "in patients meeting treatment goals with stable control" | Needs a local definition of "stable" before it can be evaluated | Clinical judgment |
- Write each clause down and mark it measurable or requiring judgment.
- For measurable clauses, name the data element: diagnosis, medication, measurement, procedure, visit, age, or date.
- For judgment clauses, get a written local definition from the clinical owner, or exclude the clause and state that the cohort is broader than the recommendation.
- Note any clause that depends on data your deployment may not hold completely, and say so where the result is reported.
- Have the clinical owner review the sorted list before you build anything.
Step 3: Build the cohort and attach the guideline citation
- In the left navigation, go to Cohort Builder.
- Create a new cohort from the measurable criteria agreed in Step 2, or open an existing cohort that should be refined against them.
- Review the returned population and confirm it matches the population the recommendation addresses.
- Put the guideline name, version, and citation in the cohort description, along with any clause you excluded and why. This is what makes the cohort auditable a year from now.
- Save the cohort.
Step 4: Hand the cohort to care gap analysis
The saved cohort is now an ordinary Patient Journey Intelligence cohort and runs through the standard measure workflow.
- Continue in Find and evidence care gaps in a patient population, selecting the cohort you just saved.
- Choose the care topic or measure that matches the recommendation you translated.
- When you report the results, cite the guideline version the criteria came from. A gap count is only interpretable against a stated definition.
Recipe reference
Each stage of this scenario is also a reusable building block.
Ask guideline questions and apply findings to a cohort workflow
When to use on its own: When a clinician or analyst needs evidence-based context before defining criteria, or when a guideline revision has to be assessed against the current population.
Features involved: Clinical Guidelines, Assistant, Cohort Builder.
Edge cases / limitations: Guideline answers support decision-making but do not replace clinical review. Not every recommendation maps cleanly to available patient data, and a clause that cannot be evaluated should be stated rather than dropped silently.
Value: Connects published clinical evidence to a cohort definition someone can audit.
For the care gap analysis, review, and export steps that follow, see the recipe reference on Find and evidence care gaps in a patient population.