Key takeaways
- In a large health system study, only about a third of primary care referrals to specialists ended in a documented completed visit.
- The biggest single drop-off was before scheduling: many referrals never got an appointment date at all.
- Long waits, coverage barriers and poor communication between offices each add their own losses.
- You can measure your own leakage with a simple monthly match of referrals sent or received against visits completed.
Two meanings of referral leakage
Health systems use referral leakage to mean patients referred outside their own network, a revenue concern. For most independent practices, the more pressing meaning is simpler: patients who are referred and never arrive anywhere. The referral is written, maybe sent, maybe received, and then nothing happens. The patient does not get the care the referring clinician thought they needed, and the receiving practice never gains the new patient.
Both meanings share the same causes. A referral that is slow, unclear or hard for the patient to act on is a referral that leaks, whether to another network or out of the system entirely. This article focuses on the second kind, because it is the one every practice, referring or receiving, can do something about.
What the research shows
One of the clearest studies followed more than 100,000 referral scheduling attempts from primary care to specialists in a large US health system. Only 34.8% resulted in a documented completed appointment[1]. The largest single loss came early: 38.9% of referrals had no documented appointment date at all[1]. In other words, most of the leakage happened before anyone sat in a waiting room.
- Documented completed appointment 34.8%
- Appointment date but no documented completion 26.3%
- No documented appointment date 38.9%
Source: Journal of General Internal Medicine (via PubMed Central), 2018[1]
Show the data
| Item | Value |
|---|---|
| Documented completed appointment | 34.8% |
| Appointment date but no documented completion | 26.3% |
| No documented appointment date | 38.9% |
A single health system is not every practice, and documentation gaps account for part of the picture. Some patients were seen without the visit being linked back to the referral. Still, the finding matches what referral coordinators describe: referrals that stall at the first contact, patients who cannot be reached and visits that are booked but never attended.
The five places referrals drop off
| Stage | What goes wrong | Warning sign | First fix |
|---|---|---|---|
| Sending | Order written but not transmitted, or sent to the wrong office | Referring office cannot say when it was sent | Track sent referrals in a log with a confirmation |
| Receiving | Fax or message lands in an unwatched inbox | Patients call asking why nobody contacted them | One monitored queue for all channels |
| Scheduling | Patient not contacted, or contacted once and dropped | Many referrals with no appointment date | Contact within one business day, with a fixed retry cadence |
| Waiting | Next available slot is too far out; patient goes elsewhere or gives up | High cancellation rate on long-lead appointments | Reserve new patient slots and use a cancellation list |
| Attending | Patient does not show, or is seen but nothing goes back to the referrer | Referrer has no record of the visit | Reminders, plus a consult note back to the referrer |
Leak one: the referral that never arrives
Referrals still travel by fax more often than anyone likes to admit. A busy primary care office signs a referral order, someone faxes it, and the receiving fax goes to a machine or inbox that is checked twice a day. If the fax fails, nobody knows. If it arrives with an illegible page, it sits in a pile waiting for clarification that nobody requests.
The fix on the sending side is a referral log with a confirmation step: the order is not done until the receiving office confirms receipt. On the receiving side, it is a single queue for every channel, checked several times a day, with a named owner. The guide on referral intake covers that workflow in detail.
Leak two: nobody calls the patient
The largest loss in the research came at scheduling. A referral received but not acted on for a week has a poor chance of becoming a visit. Patients forget, assume the other office will call, or book with whoever calls first. Receiving practices that call within one business day, text after a missed call and try again at a different time of day convert far more referrals than those that call once and wait.
Referring offices can help by telling the patient plainly what will happen: โBayview Cardiology will call you within two days. If you do not hear from them by Friday, call this number.โ A patient who knows to expect the call answers unknown numbers.
Leak three: the wait is too long
Access is tight in many specialties. In a 2026 poll of medical groups, 28% said new-patient appointment wait times were longer than a year earlier, while 22% said they were shorter[2]. Every extra week between referral and visit gives the patient another chance to cancel, go to urgent care or simply stop pursuing it.
Patients are also finding other routes. Among US adults who use AI tools for health information, 18% said a major reason was not having a regular provider or not being able to get an appointment, rising to 38% among users aged 18 to 29[3]. A long wait does not just delay care; it pushes some patients to look for answers elsewhere.
Leak four: coverage barriers
Insurance shapes who gets scheduled. A meta-analysis of 34 secret shopper studies found that Medicaid coverage was associated with a 1.6-fold lower likelihood of successfully scheduling a primary care appointment and a 3.3-fold lower likelihood for specialty care, compared with private insurance[4]. When the receiving practice does not accept the patient's plan, the referral often dies quietly instead of being redirected.
Referring offices can reduce this leak by keeping a short list of specialists who accept each major plan in their area, including Medicaid, and checking it before sending. Receiving offices that decline a referral for coverage reasons should say so to the referrer the same day, so the patient can be sent elsewhere quickly.
Leak five: the visit happens but nobody knows
Communication between offices is uneven in both directions. In a national physician survey, 69.3% of primary care physicians said they always or most of the time send a patient's history and reason for referral, but only 34.8% of specialists said they always or most of the time receive it. And while 80.6% of specialists said they send consult results back, only 62.2% of primary care physicians said they receive them[5]. A visit the referrer never hears about looks like leakage from their side, and a referral without history makes the specialist's visit less useful.
The remedy is covered in closing the referral loop: a scheduled notice, a no-show notice and a consult note, each sent to the referrer on a set timeline.
What leakage costs a receiving practice
For a specialty practice, referrals are often the main source of new patients, and each lost referral is a lost first visit plus everything that would have followed: procedures, follow-ups and the referring clinician's confidence. The loss rarely shows up in any report, because a referral that never became a visit leaves no claim behind.
Example: a fictional dermatology practice receives 120 referrals a month. If half of them become completed first visits, and each new patient leads to an average of three visits over the following year, then improving completion by ten percentage points adds 12 new patients a month and roughly 36 visits a year from each month's referrals. The numbers are illustrative, but the arithmetic is worth doing with your own figures before deciding how much staff time the referral queue deserves.
There is also a relationship cost. Referring offices notice when their patients are not seen, and they quietly start sending patients to the practice that calls back faster. Fixing leakage is as much about keeping referrers as it is about the visits themselves.
Measure leakage in your own practice
Pick one month
Pull every referral you sent, or every referral you received, in a recent month. Start with one direction.
Match to visits
For each referral, check whether a visit was booked and whether it was completed. Use the patient's name, date of birth and the receiving office.
Classify each loss
Assign each unbooked or uncompleted referral to one of the five stages above.
Calculate three rates
Share booked, share completed and share with information returned to the referrer.
Fix the biggest stage first
Choose the stage with the most losses and change one process, then repeat the measurement in three months.
Example: a fictional gastroenterology practice matched one month of received referrals to visits and found that most losses came from patients contacted once and never again. It added a second call and a text, plus a day-five notice to the referring office. Three months later, the share of referrals with a booked visit had risen and two large referring practices had started sending more patients. Spreadsheet or software? See tracking referrals: spreadsheet vs software.
Frequently asked questions
What is a good referral completion rate?
There is no universal benchmark, and published rates vary with how completion is documented. Measure your own baseline, then track whether booked and completed shares rise after each change.
Who is responsible for a referral, the sender or the receiver?
Both. The sender is responsible for transmitting a complete referral and confirming receipt; the receiver for contacting the patient, booking and reporting back. Shared logs make the handoff visible.
How quickly should a specialist office contact a referred patient?
Within one business day for routine referrals, the same day for urgent ones, with a retry cadence of calls and texts if the patient does not answer.
Does insurance affect whether referrals are completed?
Yes. Research found Medicaid patients had a markedly lower likelihood of successfully scheduling specialty appointments than privately insured patients. Checking plan acceptance before sending reduces this leak.
References
Journal of General Internal Medicine (via PubMed Central) (2018). Closing the Referral Loop: an Analysis of Primary Care Referrals to Specialists in a Large Health System (Journal of General Internal Medicine) (opens in a new tab) โ
Medical Group Management Association (MGMA) (2026). New-patient wait times largely hold flat in 2026 as some groups add providers in bid to meet demand (opens in a new tab) โ
KFF (2026). KFF Tracking Poll on Health Information and Trust: Use of AI For Health Information and Advice (fielded Feb 24 - Mar 2, 2026) (opens in a new tab) โ
Inquiry: The Journal of Health Care Organization, Provision, and Financing (via PubMed Central) (2019). Medicaid Patients Have Greater Difficulty Scheduling Health Care Appointments Compared With Private Insurance Patients: A Meta-Analysis (Inquiry, 2019) (opens in a new tab) โ
Archives of Internal Medicine (via PubMed) (2011). Referral and consultation communication between primary care and specialist physicians: finding common ground (Archives of Internal Medicine) (opens in a new tab) โ