SNF Referral Conversion: Cohorts, Open Cases and Source-Mix Effects

Define the opportunity before counting it

A portal message, a referral opportunity, a resident and a stay are different units. Record how repeated submissions are linked, how a genuinely new opportunity is recognized, and how uncertain matches are reviewed. Preserve source references so the denominator is reproducible.

Choose the outcome explicitly. Administrative acceptance, scheduled arrival and actual arrival answer different questions. This guide concerns measurement, not clinical admission selection or coverage determination. The examples and analytical models are original and hypothetical; they are not benchmarks or evidence that a workflow causes more admissions.

Work a complete referral cohort

A facility receives 50 submissions during a defined week. Ten are updates or duplicates under the approved rule, leaving 40 unique opportunities. At the observation cutoff, 20 have arrived, 12 have other resolved outcomes and eight remain open.

The last row is an arithmetic bound, not a forecast or confidence interval. It assumes the cohort and resolved outcomes stay fixed and only the eight open cases can add arrivals. If cases reopen, data is corrected or the inclusion rule changes, recalculate. Do not present 62.5% as an improvement over 50%; it uses a different denominator.

Record the report as “20 of 40 arrived, eight still open, observed through [date].” The cutoff belongs beside the percentage because recent referrals have had less time to reach an outcome.

Keep period activity separate from cohort outcomes

In a hypothetical April, 30 new referrals arrive. Twelve of them become arrivals in April. An additional 24 April arrivals come from March referrals. The same-month activity ratio is 36 arrivals/30 new referrals = 120%.

April's new-referral cohort conversion observed through April 30 is 12/30 = 40%. The 24 earlier-cohort arrivals explain the difference. A 120% ratio is not proof of impossible data, but calling it April referral-cohort conversion would be incorrect.

For trend review, compare cohorts at a consistent observation age where feasible and preserve still-open counts. A seven-day-old cohort and a thirty-day-old cohort are not equally observed merely because both appear on a monthly chart. A period activity report remains useful for workload and arrivals; label it accordingly.

Test whether source mix reverses the apparent trend

Consider two fictional referral source groups with consistently defined and fully observed cohorts:

Both source-group rates fell ten percentage points, but the combined rate rose 22 points because more referrals came from Group A. An overall improvement does not establish improvement within each source or a software effect. This simple source-mix reversal is why the report should retain both grouped counts and the aggregate.

It also explains why ranking sources by a single percentage is weak. A source with three arrivals from four referrals has an observed 75% share, but that small cohort cannot establish superior partnership quality. Review volume, observation time, capacity and documented coordination issues before directing outreach.

Record outcomes without inventing motives

Use separate fields for final outcome, primary documented reason, contributing factors and unknown reason. A referral may wait for information, encounter a later capacity change and then be withdrawn by the source. Preserve the sequence; do not call it three declines or infer an undocumented clinical reason.

If 12 cases have nonarrival outcomes and five lack documented reasons, show seven documented and five unknown. Assigning the most common reason to those five makes the chart look complete while removing what the data actually says.

The AHRQ On-Time transfer and intake report handouts illustrate reporting built from defined intake and transfer elements. Their clinical event definitions are not substitutes for this referral cohort definition. The ONC Health IT Playbook connects workflow design and measurement; applying that principle here means changing capture and staff instructions when a report needs a distinction the current record does not contain.

Use the analysis to choose a specific investigation

If open cases cluster in awaiting-verification status, examine internal review capacity. If they cluster in awaiting-source-response status, inspect request specificity and receipt routing. If the source mix changed, examine its contribution before attributing the aggregate result to faster work.

An independent facility can keep a small weekly cohort ledger and review the same cases at successive cutoffs. A group needs a common opportunity definition and a method for referrals moving between sites. A regional opportunity routed to two facilities should not automatically become two regional opportunities; local workload and regional conversion can use different units when labeled explicitly.

Retain the duplicate rule, source normalization and definition version with the report. When an integration begins detecting duplicates more reliably, a smaller denominator can raise conversion without any added arrivals. Describe that as a measurement change, then rebuild comparable trends if the underlying evidence supports it.