Summary

The published ECAT QDM specification is sensitive to documentation timing and local workflow configuration because it uses a union of alternative timestamps rather than a single canonical boarding start. As a result, hospitals with similar patient flow experiences may generate materially different measured boarding performance, depending on how admission decision, bed-assignment, and observation-status events are recorded locally. Stakeholders reflected that the general logic is sound but recommended CMS provide clarification that the timestamp is intended to represent the earliest clinically operative boarding signal and consider publishing pathway-level information and collecting pathway-resolved timestamps to improve auditability and comparability.

Vector 1: Delayed Admit Order Signing

Risk: In the published QDM logic, one pathway is triggered when an Encounter, Order coded to Decision to Admit to Hospital Inpatient has an authorDatetime at least 241 minutes before ED departure. In many implementations, authorDatetime may reflect the time the electronic order is finalized or co-signed by an attending physician, rather than the earliest clinically operative time when the admission decision was made or communicated. If organizations complete or finalize the order closer to physical departure (or after), the measured interval can be shortened without any change in operational boarding.

Why this matters: This creates sensitivity to documentation workflow rather than to the underlying patient experience.

Suggested revision: Clarify in the specification or implementation guidance that the measure should use the earliest clinically operative admission-decision timestamp available, rather than a delayed signature or administrative completion timestamp. If the intent is to use authorDatetime, the guidance should explicitly state that it must represent the time the order was first communicated or placed for care-delivery purposes, not a later time of an author's electronic completion or co-signature.

Vector 2: Cancel-and-Reorder Cycling

Risk: The function AdmitInpatientOrBedAssignmentEncounterOrder() uses Last(... sort ascending), returning the latest qualifying authorDatetime among bed-related Encounter, Order records. Earlier orders with statuses outside the allowed set do not contribute. As a result, an early bed request can be superseded by a later replacement order, causing the boarding pathway to reflect the later timestamp, rather than the first operational boarding signal. A hospital can place an early bed request, functionally board the patient, and then cancel the order and replace it closer to departure or upon physical arrival to a hospital floor. The boarding clock for this pathway resets to the re-order timestamp, potentially dropping below 241 minutes, even when the patient's operational boarding began much earlier.

Why this matters: This makes the measure vulnerable to timestamp reset behavior when orders are canceled and re-entered for operational or workflow reasons.

Suggested revision: Consider using the earliest qualifying bed-related order timestamp associated with the encounter, rather than the latest. At minimum, assess whether the current use of Last() is intentional because it systematically biases this pathway toward shorter measured boarding times when multiple qualifying orders are present. Consider monitoring cancel/re-order rates in submitted data as an outlier signal.

Vector 3: Observation Status Assignment to Suppress Boarding Numerator

Risk: The logic excludes encounters from the boarding and extended length-of-stay components when an Observation Services encounter exists during the relevant ED encounter period. Because the exclusion is based on any overlap, even short or workflow-driven observation assignments may remove long-stay encounters from the relevant numerators. Strategically assigning patients to observation status during extended ED stays-even briefly-removes them from both the Boarded Time >240 min and ED LOS >480 min numerator components.

Suggested revision: Consider adding a minimum duration of observation threshold, rather than treating any overlap as sufficient. Could also evaluate observation-exclusion rates as a monitoring signal for unusual pathway use.

Cross-Cutting: Publish Pathway-Level Data

Because the boarding construct is a union of four pathways, the composite result alone does not show which source event is driving numerator inclusion. Publishing pathway-level capture rates would improve interpretability and help identify institutions with anomalous reliance on, or absence of, particular pathways. This requires no structural CQL change and provides a low-cost detectability enhancement.

Cross-Cutting: Collect Raw Event Timestamps as Supplemental Data Elements

The logic already resolves the date-time values underlying each pathway, including admission-decision order, admission-decision assessment, inpatient or bed-assignment order, inpatient-admission start, and ED departure. Requiring submission of timestamps as supplemental data would materially improve auditability and permit downstream validation without redesigning the measure. Currently, these are consumed internally and only the binary ≥241-minute result surfaces in the numerator. Stakeholders suggest requiring that each resolved timestamp be submitted as a named supplemental reporting field. Potential benefits include:

• reconstructing the boarding timeline for denominator encounters;
• identifying whether later timestamps are systematically replacing earlier operational signals;
• supporting reconciliation against other administrative or operational data sources; and
• enabling longitudinal monitoring for documentation drift that is not apparent in the binary numerator.

This would likely require modest expansion of supplemental reporting, but the relevant DateTime values are already computed within the logic. Requiring submission of the resolved decision-to-admit, bed-assignment, inpatient-admission-start, and ED departure timestamps would materially improve detectability without changing the core numerator architecture.

Conclusion

These concerns do not depend on overt record falsification. They arise from permissible documentation and workflow choices interacting with a union-based logic structure and timestamp selection rules that may not align with the earliest clinically meaningful boarding event. The result is vulnerability in cross-hospital comparability and public-reporting validity. These suggestions, synthesized from various stakeholders, could be implemented as targeted specification refinements to improve robustness without requiring measure redesign.