Screening Claims for Emergency Diagnostic-Safety Review
An independent published methods example linking emergency discharges to later Medicare admissions to prioritize diagnostic-safety review.
Home / UpEvidence / Screening Claims for Emergency Diagnostic-Safety Review
Source published October 6, 2026 · Resource reviewed October 8, 2026
The research question
Can claims help prioritize conditions for emergency diagnostic-safety review?
Data files and cohort construction
National Traditional Medicare claims, 2022–2024, identified nonelective admissions among adults aged 65 or older. Master Beneficiary Summary Files supplied demographics and mortality. A clinician panel assembled 1,802 ICD-10 codes in 76 condition categories.
Measures and analytic design
Admissions were linked to emergency discharges with a different condition 2–9 days earlier. A cohort matched on diagnosis-based (HCC) risk supplied background emergency-use rates. Adjusted signals were expressed per 10,000 emergency discharges. Thirty-day mortality and lost healthy days at home guided prioritization; late-2024 admissions without complete outcome follow-up were excluded.
Robustness checks and interpretation
The supplement listing includes alternative priority thresholds and excluded codes. High priority combined the top half of potential-error incidence with the top quartile of mortality or lost healthy days.
Practical application: build a review queue, not an error verdict
The following are FastHSR implementation considerations, not additional procedures claimed for the study. A quality team could use linked episodes to target chart review and identify conditions needing closer investigation. Preserve the emergency diagnosis, admission diagnosis, dates, and code-list version in each flagged episode so a reviewer can reconstruct the sequence.
Decisions to settle before reuse
Define discharge versus admission dates, transfers, observation stays, repeat admissions, and duplicate professional/facility bills before linking records. Distinguish an emergency discharge denominator from an admission denominator: they answer different questions. Require observable enrollment across the lookback and outcome windows. Check that an apparent return is not one continuous episode of care.
Limitations and local application
These are potential-error signals, not verified diagnostic errors, preventable events, or causal estimates of harm. Older Traditional Medicare patients do not represent every emergency population. For a local application, validate a sample of flagged and unflagged records against clinical charts. Document how often signals reflect disease progression, planned reassessment, or coding differences. Review results with emergency clinicians before ranking providers. Claims screening should support learning and case finding, not unreviewed attribution of blame.
Source article and supplement
Burke LG, Burke RC, Orav EJ, et al. Classification of Emergency Diagnosis–Sensitive Conditions Among Medicare Beneficiaries. JAMA Network Open. 2026;9(10):e2637682. doi:10.1001/jamanetworkopen.2026.37682.
- Read the primary journal article
- Article’s Supplemental Content: technical appendix and ICD-10 lists in Supplements 1 and 2
The full article and supplement listings were reviewed on October 8, 2026. The technical appendix could not be separately retrieved. Obtain the published inclusion and exclusion code lists and confirm linkage rules before replication; this page is not a complete executable phenotype.
Frequently asked question
Does a return admission prove the emergency diagnosis was wrong?
No. A later admission is a screening signal. Chart review is needed to distinguish diagnostic error from progression, planned follow-up, or differences in coding.