Quick answer
Referral concentration analysis asks how much admissions activity depends on sources that could change together. Group related sources, use completed admissions with a defined period, and test interruption and replacement scenarios. Translate lost arrivals into census only through an explicit timing model. Patient choice and clinical suitability remain boundaries on any commercial response.
A source can produce excellent conversion and still create a fragile business dependency. When several hospitals share an ownership group or discharge process, a single operational change may affect them together. The question for a regional director is how to prepare for that disruption without turning residents into guaranteed referral volume.
Hospital discharge planning must respect patient choice and use relevant post-acute information. A network relationship does not create entitlement to placements. Keep any contingency plan consistent with those boundaries. 42 CFR 482.43(a)(8) and (d).
Build the dependency map before calculating shares
Use a defined outcome, such as completed admissions in the last 90 days. Keep referrals, accepted offers and completed admissions separate. Deduplicate repeat messages and establish how a transfer back to the facility is classified. Record missing source attribution as its own category until resolved.
Map each source to the dependencies relevant to the scenario: common parent, shared discharge center, specialty service, geography or exchange channel. These groupings answer different questions. Two unrelated hospitals using one unavailable messaging platform may share a technology dependency without sharing corporate control.
Public nursing-home counts and average census can contextualize the service area, but they do not disclose current hospital referral streams. CMS's dictionary makes the public fields and periods explicit. Use the facility's reconciled internal admissions history for this analysis. CMS nursing-home data dictionary.
Calculate concentration at two levels
Consider 100 synthetic completed admissions: Hospital A supplied 40, Hospital B 20, Hospital C 20 and community sources 20. A and B share a parent. The largest individual source represents 40%; the common-parent group represents 60%.
For an optional descriptive concentration index, square the decimal shares and sum them. At the four-source level, 0.40² + 0.20² + 0.20² + 0.20² = 0.28. At the grouped level, 0.60² + 0.20² + 0.20² = 0.44. The increase reveals the effect of grouping. These are internal comparison measures, not antitrust findings or universal thresholds for acceptable dependence.
Stress-test a specific interruption
Assume the common-parent group would ordinarily produce 60 admissions in the next 30 days. In a scenario where its arrivals fall by half, 30 admissions are lost. If eight suitable replacement admissions actually occur, the net shortfall is 22. Keep the 50% reduction and eight replacements visible as assumptions, not predictions.
Translate that shortfall into resident-days using explicit arrival dates and expected days present within the scenario window. If the 22 missing admissions would each have contributed an average of ten days during that 30-day window, the illustrative shortfall is 220 resident-days, or 7.33 residents in average daily census. This is not the day-30 census loss and is not a claim that their full length of stay is ten days.
Extend to 60 days only after modeling continuing arrivals, departures and recovery. Doubling the 30-day census effect ignores those movements.
Choose actions the organization can deliver
Separate disruption causes. A communication outage calls for a tested alternate route; a hospital service closure calls for a demand reassessment; an internal response backlog calls for operational work. Broad outreach will not correct an unavailable clinical service or an unresolved information-transfer failure.
The software requirement is a dated source hierarchy, preserved historical attribution, scenario inputs and a traceable link to completed admissions. Test a parent-group change without rewriting prior reports. Require both raw-source and grouped exports so an attractive dashboard cannot hide correlated exposure.
Use the result to prioritize relationship continuity and service planning, with patient preferences and safe admission decisions intact. Ease's skilled nursing offering is planned; concentration analytics would require a defined and demonstrated future scope.
Questions & answers
Why can ten hospital referral locations still represent concentrated SNF demand?
Several locations may share a parent organization, discharge platform or service line that changes together. Report both individual-source and common-dependency views rather than assuming every location is independent.
Does losing ten admissions mean census immediately falls by ten?
No. Census effects depend on when those residents would have arrived, how long they would have stayed, existing departures and replacement arrivals. Use a dated scenario rather than equating monthly admissions with point-in-time census.
Primary sources
Use the cited source and its applicable version when making a facility-specific decision. These guides explain operational evaluation methods; they are not individualized clinical, coding, or legal advice.