Multi-Location Appointment Overflow: A Routing Measurement Study

Research Question and Scope

The research question is narrow: when a request overflows from a full or ineligible location to another branch, clinician, or provider, what must be recorded on the same request record for cross location comparisons to be defensible? Scope covers multi site service networks that keep one visible request record while routing internally. It excludes requests that are abandoned at intake, requests handled entirely by an external referral partner with no shared record, and walk in traffic with no prior request. The unit of analysis is the request, not the appointment, because one request can generate multiple offers and one appointment can satisfy multiple requests in some referral arrangements. Required intake fields include request identifier, originating location, requested service or clinician type, date and time of first contact, channel, and stated constraints such as travel radius, language, and availability windows. Required routing fields include eligibility determination, reason for overflow, offer owner at each step, offer timestamp, and expiration. Required disposition fields include final state, final location, time to disposition, and reason for closure. The CDC Program Evaluation Framework supports stating evaluation questions and intended use before data collection, which is why these fields are specified in advance rather than reconstructed after the fact. The U.S. Census Bureau Statistical Quality Standards support documenting definitions, sources, and known limitations so that downstream users can judge fitness for use.

Methodology and Data Quality

The design is a retrospective cohort of request records over a defined observation window, paired with a prospective validation sample. The window should be long enough to include normal demand variation and short enough that routing rules and location capacity are stable; document any rule changes inside the window and treat them as strata or exclusions. Sampling follows NIST Engineering Statistics Handbook guidance: define the population, state the sampling frame, and describe how records were selected. A simple approach is a census of overflow requests in the window, with a random sample of non overflow requests as a comparison group. Data quality checks follow the U.S. Census Bureau Statistical Quality Standards: each field has a definition, an allowed value set, an owner, and a completeness threshold. Key checks include duplicate request identifiers, timestamps that precede first contact, offers with no owner, dispositions with no timestamp, and eligibility flags that conflict with the stated reason for overflow. Missingness must be reported by field and by location, because differential missingness across locations can create apparent performance gaps. The AHRQ CAHPS Improvement Guide supports linking measurement to a defined improvement cycle, so the study should name the decision the data will inform, such as whether to change referral eligibility rules or offer ownership. Any statistic reported should carry its denominator, its observation window, and its exclusion list.

Analysis Plan and Inference Boundary

Analysis proceeds in three layers. First, descriptive counts: requests by originating location, overflow reason, eligibility outcome, offer count, and final disposition, each with a denominator and a time window. Second, time measures: time from first contact to first offer, first offer to acceptance or decline, and first contact to final disposition, reported as distributions rather than single averages because scheduling delays are typically skewed. Third, comparison across locations, using the same request definition and the same disposition states. Inference is bounded by the design. A retrospective cohort can describe what happened in the window; it cannot establish that a routing rule caused an outcome unless assignment to routing paths was effectively random or a suitable comparison was constructed. Confounding is likely: locations differ in service mix, staffing, payer mix, and geography, and these differences travel with the request. The CDC Program Evaluation Framework supports separating findings from interpretation and stating alternative explanations. Practical rules: report the number of requests excluded and why; do not compare locations on acceptance rate without also reporting offer volume and offer ownership; do not treat a disposition of no appointment as a single category when it can mean ineligible, declined, expired, or unreachable. Where a number is not established by a source, state the measure to be collected rather than supplying a figure.

Limitations and Alternative Explanations

Several failure modes recur in overflow measurement. First, denominator drift: counting appointments instead of requests, or counting only requests that reached an offer, which inflates apparent success by dropping early exits. Second, ownership ambiguity: an offer made by one location but attributed to another, which makes accountability untraceable and distorts location comparisons. Third, travel constraint omission: if travel radius and mode are not captured at intake, a declined offer may be recorded as patient choice when it was a distance problem. Fourth, disposition collapse: merging ineligible, declined, expired, and unreachable into one state hides the operational lever that would change the outcome. Fifth, window mismatch: comparing locations over different date ranges or with different capacity events inside the window. Sixth, eligibility rule changes mid window without stratification. Alternative explanations for any observed difference include case mix, staffing levels, referral partner behavior, and data entry practice. The U.S. Census Bureau Statistical Quality Standards call for documenting these limitations with the data. The AHRQ CAHPS Improvement Guide supports testing changes in small cycles before broad conclusions, which is a reasonable response when confounding cannot be removed. NIST guidance on sampling reinforces that a sample cannot repair a biased frame; if the frame excludes requests that never reached routing, the study describes routed requests only.

Responsible Use of Findings

Findings from this study should be used to improve record keeping and routing decisions, not to rank locations on a single headline number. A responsible use is to publish a small set of definitions, then report each location against those definitions with denominators, windows, and exclusions visible. Another responsible use is to route improvement work to the specific field that is failing, such as eligibility capture or offer ownership, rather than issuing a general performance message. Scheduling roles already carry substantial administrative load; BLS describes medical secretaries and administrative assistants as handling records, correspondence, and scheduling related duties, so adding fields without removing duplicate entry is a real cost that should be measured. The AHRQ CAHPS Improvement Guide supports iterative measurement tied to a defined aim, and the CDC Program Evaluation Framework supports stating intended use and audiences up front. Limits to state plainly: this brief does not report observed rates, does not compare named organizations, and does not claim that any routing configuration is superior. Where a decision depends on a number that is not established by a cited source, the correct next step is to define the measure, assign an owner, set the observation window, and collect it before acting.

Research methodology

The study design is a retrospective cohort of appointment request records over a fixed observation window, with a prospective validation sample to test field completeness and definition stability. The record unit is the request, identified by a single request identifier that persists across locations, offers, and final disposition. Each record carries intake fields, routing fields including eligibility and offer ownership, and disposition fields with timestamps. Records are included when a request was created in the window and reached at least one routing decision. Records are excluded when the request was abandoned at intake, when no shared record exists, or when routing rules changed inside the window without stratification. Descriptive analysis reports counts, distributions, and missingness by field and location.

Data sources and methodology

This brief reports published findings as stated by each source. It does not combine study populations into a new benchmark; local operators should treat the figures as context and measure their own workflow.

1. BLS Occupational Outlook Handbook: Medical Secretaries and Administrative Assistants: Occupational duties and outlook for medical scheduling roles.
2. AHRQ CAHPS Improvement Guide: Structured measurement and quality improvement guidance.
3. NIST Engineering Statistics Handbook: Descriptive analysis, sampling, and measurement design.
4. U.S. Census Bureau Statistical Quality Standards: Data quality and documentation standards.
5. CDC Program Evaluation Framework: Transparent evaluation questions and interpretation.

Related content

• Multi-Location and Multi-Provider Scheduling
• Inbound Appointment Booking

Common questions