Healthcare Referral Intake Workload Statistics 2026

A referral can appear to be one item in an electronic health record while generating work in several queues. Staff may need to read a fax, find or create the patient record, enter insurance details, check the requested service, collect missing records, route the request for clinical review, obtain authorization, contact the patient, schedule the visit, and return the specialist report.

The available healthcare referral intake workload statistics do not support one national referrals-per-coordinator benchmark. The strongest evidence instead provides measured task time in one intake center, referral volumes in defined health systems, and elapsed time or completion rates for specific workflows. Those measures can inform staffing, but only when their denominators remain clear.

Referral intake statistics at a glance

These results describe different parts of a referral. The 719-second observation is active administrative work. The 15-day result is elapsed cycle time, which includes waiting. The 34.8% result requires both a completed appointment and a report in the referring clinician's record. Combining them into one average would hide where the queue is actually slowing down.

Manual intake took about 12 minutes per referral in one observed center

A 2020 JAMIA Open study examined the referral process at the University of California, San Francisco Pediatric Access Center. Researchers observed seven staff members performing referral intake, patient registration, referral creation, and appointment scheduling. They measured cumulative manual work at 719 plus or minus 48 seconds per referral. That is roughly 12 minutes of active work.

The observed process included receiving and reading faxed material, searching for the patient in the electronic health record, entering demographics, updating insurance coverage, creating the referral, finding patient information, calling the patient, and logging the work. Missing information added calls to referring offices. The study was designed to guide a fax-to-referral automation tool, so its estimate reflects a centralized pediatric access workflow with incoming documents. It is not a national average for every referral type.

The measured mean supports a transparent capacity calculation. At 12 minutes of active work, 100 referrals represent about 20 staff hours before allowances for breaks, meetings, training, interruptions, supervision, leave, or cases that require repeated follow-up.

active intake hours = referral count x active minutes per referral / 60

Using the observed mean for illustration:

This table is arithmetic, not a staffing standard. A clinic that receives complete electronic orders may need less intake time. A specialty that must collect imaging, pathology, clinical notes, or payer documentation may need more.

Referral volume is not the same as a completed workload

The largest study in this review analyzed 103,737 referral scheduling attempts from one large primary care network to 20 high-volume specialties during the year ending June 30, 2016. Only 36,072 attempts, or 34.8%, had a documented completed appointment and a specialist report returned to the referring clinician. Another 38.9% lacked a documented appointment date.

That 34.8% is a closed-loop documentation result, not a national patient no-show rate. An incomplete record could reflect a visit that was never scheduled, a visit that did not occur, care outside the visible network, or a specialist report that did not return to the primary care record. Each state calls for a different staff action.

A separate prospective study followed 776 referred patients from 133 physicians in 81 practices across 30 states. After three months, physicians reported that 79.2% had visited the specialist, while 83.0% of patients said they had completed the referral. Agreement between physician and patient reports was only fair. Physician or staff help with scheduling was a positive predictor of completion.

The difference between those two studies shows why staffing cannot be based on a generic completion percentage. One requires a report in the record, while the other asks patients and physicians whether a visit occurred. A referral team needs a status model that can distinguish receipt, acceptance, authorization, scheduling, attendance, report return, and clinical acknowledgment.

Cycle time needs at least three clocks

Referral cycle time often refers to more than one interval. Intake teams should keep at least these clocks separate:

1. Receipt to first staff review measures queue responsiveness.
2. Receipt to appointment scheduled measures intake, triage, and scheduling work.
3. Referral order to completed visit measures patient access and capacity as well as administrative work.

A 2021 study of centralized scheduling covered cardiology, nephrology, gastroenterology, and neurology referrals in one Midwestern academic health system. Researchers compared six months before and six months after implementation. Overall completion within three months increased from 63.7% to 69.9%, while time to schedule fell from 21 days to 15 days. Results varied by specialty. Gastroenterology improved, neurology and nephrology did not change significantly, and cardiology completion declined.

A larger NYC Health + Hospitals study analyzed 50,260 requests at 19 specialty clinics across seven facilities and six specialties. After eConsult implementation, 13% of requests were resolved electronically. Among requests that still required a visit, successful scheduling increased from 66.5% to 82.3%. Mean appointment wait fell from 61.0 days to 54.1 days, but completion within 90 days did not improve significantly, moving from 38.4% to 37.9%.

Those findings separate process improvement from patient access. A team can schedule more requests and shorten the wait without increasing the share of visits completed within 90 days. Staffing reviews should therefore report scheduling and attendance separately.

Pediatric referral data shows where the open queue accumulates

A 2023 retrospective cohort study examined 38,334 referrals from nearby primary care clinics to a large children's hospital between March 2019 and March 2021. Sixty-two percent were scheduled and 54% were completed. Completion was lower for Black patients at 45%, Native Hawaiian or Pacific Islander patients at 48%, Spanish-speaking families at 49%, and publicly insured patients at 47%.

The study also found longer time to scheduling and completion for several groups after adjustment. These results do not identify how many staff minutes each referral required. They do show that one overall queue average can conceal differences in access. A useful workload report should segment open referrals by language needs, payer, specialty, urgency, and days since the last action, while protecting patient privacy and avoiding assumptions about the reason for delay.

Intake, authorization, scheduling, and closure are different jobs

Referral operations become hard to staff when all work is recorded as "referral processing." The stages have different inputs, owners, and completion rules.

CMS uses a similarly specific endpoint for Quality ID 374. The 2026 measure counts patients whose referring clinician received a report from the first clinician to whom the patient was referred. It does not treat an order, authorization, or scheduled appointment as a closed loop.

The federal SAFER Clinician Communication Guide, updated in 2024, recommends tracking referral status at the patient level until closure, including expected time to scheduling and completion. It also calls for clear follow-up accountability, standardized handoffs, and acknowledgment of referral receipt with an expected turnaround time.

Prior authorization can sit inside a referral workflow, but it is not the referral itself. The authorization queue tracks payer requirements and decisions. The referral queue tracks the handoff between referring and receiving care teams. A request can be authorized but not scheduled, or scheduled without a returned specialist report. Combining the queues makes both staffing and delay analysis less reliable.

Administrative staffing data provides a cost input, not a referral ratio

The Bureau of Labor Statistics estimated 961,610 medical secretary and administrative assistant jobs in the United States in May 2025. The national mean wage was $22.50 per hour and the median was $22.08. That occupation is broader than referral coordination and includes staff who schedule appointments, maintain records, and support medical offices.

BLS does not publish a national referral-coordinator employment count or a recommended caseload. Its wage estimate can serve as one input to a local labor-cost model, but a loaded hourly cost should also include employer payroll costs, benefits, paid leave, and local wage conditions.

monthly referral labor cost = sum of active hours by role x loaded hourly cost by role

Keep licensed clinical time separate from administrative time. Clinical triage, urgency decisions, and medical-necessity documentation may require a nurse or clinician. Patient registration, document indexing, status tracking, and routine scheduling can often follow an approved administrative procedure. The staffing model should show both rather than assigning every minute to one blended role.

Build a staffing model from local queue events

Start with eight to twelve weeks of referral data. Count every new referral and each repeated event that consumes staff time. Useful event categories include document review, missing-information outreach, eligibility checking, authorization submission, patient contact, scheduling, rescheduling, status follow-up, and report reconciliation.

For each event, record active handling minutes and elapsed waiting time separately. Use the median for routine capacity planning and the 90th percentile to understand difficult cases. Then calculate paid capacity after subtracting recurring non-queue time such as meetings, training, breaks, and leave.

required productive FTE = total monthly active referral hours / productive hours available per FTE per month

If a team records 500 referrals per week at the UCSF study's approximate 12-minute intake mean, intake alone represents 100 active hours. That calculation excludes clinical triage, authorization, repeated patient calls, appointment wait time, and loop closure. It should be a starting estimate, not a promise that a fixed headcount can safely complete the whole pathway.

Track backlog by age as well as count. Fifty referrals received yesterday are not operationally equivalent to 50 referrals untouched for three weeks. A practical dashboard can show:

• new referrals received and accepted;
• referrals awaiting missing information;
• referrals awaiting clinical triage or payer action;
• median and 90th-percentile time to first review;
• median and 90th-percentile time to scheduling;
• scheduled, completed, and closed-loop rates;
• staff touches and active minutes per referral;
• open referrals by age band and next owner.

What administrative support can own

Administrative staff can receive and index referral documents, verify demographic fields, collect missing nonclinical information, maintain work queues, contact patients, schedule visits, monitor status, and reconcile returned reports under a documented workflow. They should have a clear escalation path for urgent symptoms, unclear clinical criteria, medical-necessity questions, and privacy or consent concerns.

A healthcare virtual assistant can support defined nonclinical steps when access controls, privacy safeguards, supervision, and escalation rules are in place. The services page explains broader managed support. Related research covers healthcare referral coordination staffing and healthcare prior authorization staffing costs.

The staffing decision should follow the measured queue. A team with a large missing-document backlog needs a different mix of work than a team whose main constraint is specialist appointment capacity.

Frequently asked questions

How long does referral intake take?

One UCSF Pediatric Access Center time-and-motion study measured 719 plus or minus 48 seconds, or about 12 minutes, of cumulative manual work per referral. The study covered seven staff members and a specific centralized pediatric workflow. Local times will vary with document quality, referral type, system integration, and missing information.

How many referrals can one coordinator process?

There is no authoritative national ratio. Divide measured referral work hours by the productive hours available to the role, then add separate capacity for clinical triage, authorization, follow-up, and supervision. Do not convert an eight-hour shift into eight hours of uninterrupted queue work.

Is referral cycle time the same as specialist wait time?

No. Referral cycle time may end at first review or scheduling. Specialist wait time usually runs until the appointment. Name the start and end event whenever reporting days.

Does a scheduled appointment close the referral loop?

Not under the CMS closing-the-loop measure. CMS Quality ID 374 requires the referring clinician to receive a report from the clinician who received the referral. A scheduled or completed visit without a returned report does not meet that endpoint.

Should prior authorization be counted as referral intake work?

Track it as a separate stage. Authorization may be necessary before scheduling, but it has its own payer submissions, status checks, decisions, and appeals. Separating it shows whether delay sits with intake, the payer, scheduling, appointment capacity, or report return.

Sources and method notes

This review prioritizes federal guidance and statistics plus peer-reviewed health-services studies. Each statistic retains its study year, setting, and denominator. Active staff time is not presented as elapsed patient wait time. Scheduling, visit completion, and receipt of a specialist report remain separate outcomes.

1. Odisho and colleagues, referral automation design and time-and-motion study, published in 2020 using observations of seven UCSF Pediatric Access Center staff members.
2. Patel and colleagues, closing the referral loop, published in 2018 using 103,737 scheduling attempts from July 2015 through June 2016.
3. Bongers and colleagues, centralized scheduling, published in 2021 using four specialties and six-month preimplementation and postimplementation periods.
4. Gaye and colleagues, NYC Health + Hospitals eConsult implementation, published in 2022 using 50,260 requests from 2016 through 2020.
5. Srinivasan and colleagues, pediatric specialty referral scheduling and completion, published in 2023 using 38,334 referrals from March 2019 through March 2021.
6. Forrest and colleagues, ASPN Referral Study, published in 2007 using 776 patients, 133 physicians, 81 practices, and 30 states.
7. CMS, 2026 Quality ID 374 specification, version 10.0 dated December 2025.
8. Office of the National Coordinator for Health Information Technology, SAFER Clinician Communication Guide, August 2024 edition.
9. Bureau of Labor Statistics, May 2025 Occupational Employment and Wage Statistics, published in 2026.
10. Ramelson and colleagues, enhanced referral management system, published in 2018 and reporting continued post-pilot use at 2,000 to 2,500 referrals per month.

The article was verified on September 21, 2026. The 2026 title identifies the editorial edition. It does not imply that every underlying observation was collected in 2026.