Compliant Physician Query Examples: Templates CDI Teams Can Use

Compliant physician query examples share three traits. They present clinical indicators from the health record, ask a clear question without suggesting a preferred answer, and give the provider every clinically reasonable response, including "other" and "unable to determine." Open-ended, multiple-choice and, in limited cases, yes/no queries can all meet that standard.

This guide walks through each format with annotated before-and-after templates you can adapt for FY2027 workflows. It is educational content, not legal or compliance advice. Always follow your organization's query policy, the AHIMA/ACDIS query guidance, the ICD-10-CM Official Guidelines and payer requirements.

What makes a physician query compliant?

A query is a communication tool. Its job is to clarify documentation, not to obtain a specific diagnosis. The industry reference point is the AHIMA/ACDIS Guidelines for Achieving a Compliant Query Practice, available through AHIMA. Most facility policies build on it.

Across formats, a compliant query typically:

• Is supported by clinical indicators found in the health record
• Asks a question rather than directing an answer
• Offers clinically reasonable options, plus "other" and "unable to determine" (for multiple-choice and yes/no)
• Leaves out reimbursement, DRG impact, quality scores and audit risk
• Does not introduce information that is not in the record
• Is retained according to policy so it can be audited

Why does this matter for coders? Section I.A.19 of the ICD-10-CM Official Guidelines states that code assignment is based on the provider's diagnostic statement that a condition exists. Coders cannot assign a diagnosis from lab values alone. The query is how the provider's own judgment reaches the record.

Which query formats are acceptable, and when should each be used?

Three formats dominate CDI work. Each fits different situations.

Verbal queries follow the same rules. Document them per policy, including the indicators discussed and the provider's response.

How do you write an open-ended query that is not leading?

An open-ended provider query presents facts and asks the provider to interpret them. It never names the diagnosis you hope to see.

Before: a leading open-ended query

Dr. Lee, the patient's hemoglobin dropped after surgery. Please document acute blood loss anemia so we can capture it.

Problems with this query:

• It names one diagnosis and directs the provider to document it.
• "So we can capture it" implies a financial or data motive.
• It gives no values, dates or treatment.

After: a compliant open-ended query

Dr. Lee, please review the following from this admission:

• Pre-op hemoglobin 13.1 g/dL (3/2); post-op day 1 hemoglobin 8.4 g/dL (3/4)
• Operative note: estimated blood loss 900 mL
• 2 units PRBCs transfused 3/4 per the MAR

Based on your clinical judgment, please clarify the clinical significance of these findings in the progress notes.

Why it works:

• Each indicator cites a source and date.
• "Clinical significance" leaves room for any answer, including none.
• If the provider documents acute posthemorrhagic anemia, the coder can assign D62. If not, nothing is coded from the labs.

What is the difference between a leading vs non-leading query?

A leading query steers the provider toward one answer. Warning signs include naming only one diagnosis, offering unsupported options, opening with "Please document," or mentioning payment.

A non-leading query presents the evidence, asks an open question or offers balanced options, and accepts any clinically supported response. A simple test: if a reviewer could predict the "right" answer from the wording alone, rewrite it.

What should a multiple-choice query include?

The multiple choice query format works well for specificity gaps. Every option must be clinically reasonable given the indicators. Add "other" with a free-text line and "unable to determine" (some policies say "clinically undetermined").

Before: a multiple-choice query with one real option

Patient has CHF. Please specify:

• Acute on chronic diastolic heart failure
• Other

Only one specific choice is offered. That makes it functionally a leading query.

After: a compliant multiple-choice query

Dr. Patel, the H&P documents "CHF exacerbation." The record also shows:

• Echo 3/1: LVEF 60%, grade II diastolic dysfunction
• Home furosemide 20 mg daily; IV furosemide 40 mg BID started 3/3
• BNP 1,240 pg/mL on admission (prior outpatient value 310 pg/mL on 1/15)
• Bilateral lower extremity edema and crackles on exam

Please clarify the type and acuity of the heart failure being treated:

• Acute on chronic diastolic (HFpEF) heart failure
• Chronic diastolic (HFpEF) heart failure
• Acute diastolic (HFpEF) heart failure
• Other: __
• Unable to determine

Annotations:

• Several reasonable acuity options appear, each supported by indicators.
• No code numbers or payment language appear in the query.
• After the response, the coder maps the documentation. "Acute on chronic diastolic heart failure" supports I50.33. Documentation of "CHF" alone stays at I50.9 (Heart failure, unspecified).

Always check the code page for instructional notes. For example, hypertension with heart failure triggers the combination-code rules in Section I.C.9.a.

When is a yes/no query allowed?

Yes/no queries are the most restricted format. Under the AHIMA/ACDIS guidance, they should not be used when only clinical indicators exist and the condition has not been documented. Asking "Does the patient have sepsis? Yes/No" in that situation is leading.

The practice brief describes narrower uses, such as:

1. Determining present-on-admission (POA) status
2. Substantiating or further specifying a diagnosis already present in the record, such as findings in a pathology or radiology report
3. Establishing a cause-and-effect relationship between documented conditions
4. Resolving conflicting documentation

Confirm the current list in the practice brief and your facility policy. Some organizations restrict yes/no queries further.

Before: yes/no used to introduce a diagnosis

Creatinine rose from 0.9 to 2.1. Does the patient have acute kidney injury? Yes / No

No provider has documented AKI. This query introduces a new diagnosis through a closed question.

After: yes/no used for POA status

Dr. Gomez, your progress note on 3/4 (hospital day 2) documents "acute kidney injury." ED labs on 3/3 show creatinine 2.0 mg/dL; the outpatient baseline on 1/10 was 0.9 mg/dL. Was the acute kidney injury present at the time of the inpatient admission order?

• Yes
• No
• Clinically undetermined
• Other: __

The diagnosis is already documented. The query only clarifies timing. The coder assigns N17.9 (Acute kidney failure, unspecified) with the POA indicator that matches the response.

How should clinical indicators be presented in a query?

Clinical indicators in queries are the evidence that justifies asking. Weak indicators make queries look like fishing. Strong ones make the query self-explanatory to the provider and to an auditor.

Best practices:

• Cite the source, date and time (ED note, MAR, lab, imaging report)
• Show trends and baselines, not single values
• Include treatment and monitoring, such as medications, fluids, oxygen or consults
• Present findings objectively, without bolding or commentary
• Use prior-encounter information only when relevant and allowed by policy

This connects to Section III.B of the Official Guidelines. Abnormal findings are not coded unless the provider indicates their clinical significance. When findings are outside the normal range and the provider has ordered further testing or treatment, it may be appropriate to ask the provider whether a diagnosis should be added. For the respiratory example above, a documented response of acute respiratory failure with hypoxia would support J96.01.

Compliant physician query examples: a quick template library

Adapt these CDI query examples to your facility's approved forms. Replace the bracketed placeholders with record data.

1. Clinical significance (open-ended): "The following findings are noted: [INDICATOR, SOURCE, DATE]. Please document the clinical significance of these findings, if any."
2. Specificity (multiple-choice): "[DOCUMENTED DIAGNOSIS] is documented with [INDICATORS]. Please clarify the type/acuity: [OPTION A] / [OPTION B] / [OPTION C] / Other / Unable to determine."
3. Conflicting documentation: "The [NOTE, DATE] documents [X], while the [NOTE, DATE] documents [Y]. Please clarify the diagnosis that best reflects your clinical judgment."
4. POA status (yes/no): "[DOCUMENTED DIAGNOSIS] is documented on [DATE]. Was it present at the time of the inpatient admission order? Yes / No / Clinically undetermined / Other."
5. Cause and effect: "[CONDITION A] and [CONDITION B] are both documented. Please clarify whether a relationship exists: Related / Not related / Other / Unable to determine."

These compliant physician query examples are starting points. Your compliance team should approve final wording before use.

What are common query mistakes that create compliance risk?

Auditors look for patterns, not just single queries. These mistakes come up often:

• "Please document" phrasing that names one diagnosis
• Unsupported options added to a multiple-choice list to pad it
• Missing "other" or "unable to determine" choices
• Financial or quality language, such as DRG, CC/MCC, SOI/ROM or denial risk
• Consequence statements, such as "Without this, the claim may be denied"
• Pre-checked boxes or highlighted "preferred" answers
• Information not in the record introduced by the query author
• Repeat queries sent to get a different answer after "unable to determine"
• Lost queries that are not retained per policy

Review a sample of queries regularly against the compliant physician query examples in this guide. Track response and agreement patterns, and retrain when one format or author shows outliers.

How can AI query builders help without replacing CDI judgment?

AI can speed up drafting. ICD Code Finder's AI CDI query builder drafts open-ended and multiple-choice queries from the indicators you supply. It uses your own OpenAI, Anthropic, Gemini or OpenRouter key, and you pay the provider directly. Follow your organization's policies on PHI and AI vendor use before entering any patient data.

Every AI draft needs review by a qualified CDI specialist or coder. Check it against the same list: supported options, neutral wording, "other" and "unable to determine," and no payment language. AI output is a draft, not a compliant query by default.

Other tools that support the workflow:

• Chat with the ICD-10-CM Official Guidelines to find the section behind a query rationale
• Search ICD-10-CM codes to confirm code titles and instructional notes before mapping a response; see Excludes1 vs Excludes2 exclusion notes
• Build and share code sets for common query targets, such as heart failure specificity codes; Pro and Team plans add exports and FY-update email alerts
• Review FY2027 new, revised and deleted codes so templates reflect current specificity, effective October 1, 2026

For the official code set files, see the CDC NCHS ICD-10-CM page.

Next steps

Start by auditing five recent queries against the before-and-after compliant physician query examples above. Rewrite any that lead, then save approved templates for your team. When you are ready to map provider responses, look up codes free or draft a query with the AI query builder. For more practical guidance, browse more CDI & Queries articles.