22.2 Selecting Cases for Second-Level Review

Selecting Cases for Second-Level Review

A second-level review brings additional expertise to a documentation, coding or clinical-support question that needs closer examination. It can occur during the stay, before billing or retrospectively. The organization should define its referral criteria, reviewer qualifications, priorities and escalation process. There is no national requirement that every death, rejected query or single-MCC record receive the same review before discharge.

The purpose is accurate representation of the encounter. A review may support the existing documentation, identify a needed clarification, reveal an inappropriate query or identify a reporting error. It should remain open to each outcome. AHIMA's clinical-validation brief describes a second-level process as a possible organizational approach, particularly when ambiguity or recurring denial risks need deeper review. AHIMA Clinical Validation, 2023 Update

Define a referral by the question it must resolve

Targeted referral and sampling serve different purposes. A high-risk workqueue can prioritize known concerns; an appropriately designed sample can reveal issues that the trigger logic misses. Choosing one does not eliminate the value of the other. Set priorities using the question, consequence, timeliness and available expertise, with a route for manual referrals when software does not recognize the issue.

Review a single-CC or single-MCC case without assuming the result

“Fragile DRG” is an informal operational term for a case whose grouping depends on one qualifying secondary diagnosis. The actual effect of changing that diagnosis depends on the DRG family, remaining diagnoses, exclusions, procedures and other grouper inputs.

In the FY2027 v44 heart-failure-and-shock family, MS-DRG 291 is with MCC, 292 is with CC and 293 is without CC/MCC. Consider an illustrative case already appropriately assigned to this family: its sole qualifying MCC is removed after proper clarification, but another qualifying CC remains. With other grouping factors unchanged, the CC tier remains relevant; the case does not automatically fall to 293. Re-run the applicable grouper using the final supported information. CMS FY2027 v44 definitions, Appendices A and C

The review should ask why the diagnosis is documented, what supports it, whether any conflict needs clarification and whether reporting rules are met. It should not attempt to create “comorbidity redundancy.” Each additional diagnosis must independently satisfy the applicable requirements. National approximate dollar amounts or relative weights cannot establish the actual payment loss for an individual hospital claim. FY2027 ICD-10-CM guidelines, Sections I.A.19 and III

Treat a quality flag as a request to verify the measure

For each flag, establish the measure, reporting program, version, population and data period. Then review numerator and denominator logic, exclusions, codes, POA indicators and procedure dates as applicable. Keep the clinical assessment, coding decision and measure calculation distinct.

For example, PSI 04 is not a component of AHRQ PSI 90 v2025. A referral can still concern PSI 04, but it should not call that review a PSI 90 component review. PSI 11 likewise has several diagnosis/procedure pathways and timing rules; it is not simply “postoperative ventilation of at least 96 hours.” Review the actual specification rather than a shortened workqueue label. AHRQ PSI 90 v2025 composite guide; PSI 11 v2025 specification, pages 1–3

POA is determined relative to the inpatient admission order, with the applicable rules for conditions developing during outpatient services before that order. It is not determined simply by arrival at the hospital or the date a copied sentence was entered. U means insufficient documentation; it does not prove a condition was acquired in the hospital. A Y indicator is not a universal statutory exclusion from every PSI. Apply the particular measure's rules and clarify unresolved documentation neutrally. FY2027 POA guidelines

CMS's HAC Reduction Program uses CMS PSI 90 alongside specified infection measures. Its 1% payment reduction concerns applicable hospitals whose Total HAC Score falls in the worst-performing quartile; a single flagged case does not automatically impose that reduction. Do not transfer a measure or penalty rule to the separate Hospital VBP Program without checking that program and year. CMS HAC Reduction Program

Review mortality and complex procedures for accuracy

An expired patient's record may warrant closer review of the course, diagnosis evolution, admission timing and completeness. The outcome does not authorize a search for diagnoses designed to raise expected mortality. Solventum APR-DRG SOI/ROM subclasses are a particular methodology; they are not interchangeable with all CMS mortality measures, rating systems or registries. CMS publicly reports condition- and procedure-specific 30-day risk-standardized mortality measures. Determine the actual model before asserting that a diagnosis, DNR status or palliative-care code changes its result. A low-risk classification followed by death is a review signal, not proof of poor care. Solventum APR-DRGs; CMS outcome measures

For a procedure referral, identify the exact uncertainty: objective, body part, approach, device, qualifier or duration. Read the operative and relevant respiratory records and apply the current PCS definitions and guidelines. The ventilation category for greater than 96 consecutive hours differs from 24–96 hours; exactly 96 is not greater than 96. A workqueue flag does not establish ventilation hours or an MS-DRG, and an ECMO label alone does not establish every grouping input. FY2027 PCS tables

Refer clinical-support concerns without converting them into automatic denials

Malnutrition, sepsis, respiratory failure, kidney injury and encephalopathy can raise difficult questions, but the concern must be explained in the actual case. Use the relevant clinical lesson and source rather than applying a universal list of mandatory diagnostic cutoffs.

For example, albumin and prealbumin alone do not diagnose malnutrition. Their presence can prompt review of the provider's assessment and the nutrition record; their absence or an incomplete short vignette does not itself rule malnutrition out. Similarly, a treatment response, oxygen flow rate or one laboratory value should not replace the complete clinical picture. The reviewer identifies the concern and supports appropriate clarification; the responsible provider exercises diagnostic judgment. ASPEN visceral-protein position paper; 2026 query guidance

Use the actual review authority when analyzing denials. CMS's April 2025 Regions 3–5 RAC statement of work distinguishes coding validation and medical-necessity review from clinical validation and prohibits clinical validation in those RAC reviews. Do not treat every commercial clinical-validation denial as an example of the same RAC authority. CMS RAC Regions 3–5 statement of work, pages 13 and 24

Finally, evaluate the original query before escalating a provider interaction. A correct response may be that the diagnosis cannot be determined or that the existing record already answers the question. Use the organization's compliant follow-up process for a genuine unresolved issue. A second-level reviewer or physician advisor can help assess the query and communication; the role does not authorize pressure for a preferred diagnosis. 2026 query guidance, provider collaboration and multiple-query sections