Automating referral intake at a fertility clinic: from the fax queue to a booked consult

A referral is the moment a fertility clinic first learns a patient exists. It is also the moment that patient's information is at its most fragile: a fax that has arrived as an image, in a queue nobody quite owns, addressed to a physician who is in clinic all day. Everything the clinic later does for that patient starts here.

This is a guide to automating that stage sensibly: what a referral's first two weeks look like today, what the evidence says about referrals and delay, the requirements a clinic can hold any vendor to, and how we have built it.

The first two weeks of a referral, today

A family physician's office faxes a referral on a Tuesday. It arrives as a PDF or a scanned image in a shared inbox, or on paper in a tray: a cover sheet, a letter, sometimes a page of results, occasionally a partner's details on the back. Nobody reads it that day.

On Thursday a coordinator opens the queue and types what she can read into the EMR: name, date of birth, health card number, the referring physician, the reason. The handwriting on the second page is a guess. She searches for the patient, finds two possible matches spelled two ways, and picks one. She has no way of knowing that the same physician faxed the same referral last month, because that one was filed under the other spelling.

The patient, meanwhile, was told "the clinic will call you". She has heard nothing since. Ten days in, she calls to ask whether the referral arrived, and the person who answers cannot tell her without going to look. The clinic has not lost the referral. It has lost ten days, created a duplicate chart, and given the patient a first experience of the clinic that consists of a wait she cannot see the end of.

What the evidence says

Fax is still the front door. Ontario's health-care providers send an estimated 152 million faxes a year, according to the province's Ministry of Health, and the Ontario Medical Association says more than 90% of doctors still use fax machines; specialist referrals are among the documents most commonly faxed[1]. The province intends to phase fax out over several years; until then, referrals will keep arriving as images.

Referrals break at every step. A 2011 review in The Milbank Quarterly put the share of patients in the United States referred to a specialist each year at more than a third, and found that many referrals include no transfer of information either to or from the specialist and that there are breakdowns and inefficiencies in every component of the referral process[2]. A 2018 analysis of 103,737 referral scheduling attempts across one large US health system found that only 36,072, 34.8%, resulted in a documented completed appointment; 38.9% of scheduling attempts had no appointment date at all, and specialties with longer waits had lower completion rates[3]. The study does not report fertility separately, but the mechanism, a referral that is received and never becomes a booking, is exactly the one an unattended intake queue produces.

Patients leave before treatment starts. A Dutch cohort of 1,391 couples referred to a secondary-care fertility service found that 319 dropped out, 76.8% of them on their own initiative, and close to half before any treatment had begun[4]. The stretch between referral and first consultation is where a clinic is least visible to the patient, and it is where a good part of the loss happens.

Time is not neutral in fertility. The joint ACOG and ASRM committee opinion on female age-related fertility decline states that fecundity decreases gradually but significantly from about age 32 and more rapidly after 37, and recommends that women older than 35 receive an expedited evaluation and begin treatment after six months of trying, or earlier if clinically indicated, with more immediate evaluation and treatment warranted after 40[5]. A referral that sits for two weeks in a queue, then waits for a call-back, then waits for a slot, is spending the one resource the guideline says should be spent quickly.

What a good intake process does

These are requirements rather than features. A clinic can hold any vendor to them, and its own process too.

Read the document; do not re-key it. The patient, the partner if there is one, the referring physician and the reason for referral should come off the page into fields, and the system should say which values it is unsure of rather than presenting a guess as a fact.

Match before creating. Look for the patient in the existing records before a new chart is created. Where there are several candidates, ask a person. Never merge silently.

Flag duplicates and mismatches. The same patient referred twice inside a window; a cover letter that names one person and attached results that name another; a referral that looks like the other half of a couple.

Route every uncertain field to a person. Nothing should be confirmed until someone has decided each flagged value. If the system accepts anything on its own, it must be labelled as such, kept where staff can see it, be reversible, and be measured. A clinic should be able to see how often the automatic path was right.

Acknowledge receipt. The patient should be able to learn that the referral arrived and what happens next without a staff member going to look. So should the person who answers the clinic line.

Record every touch. From arrival to booking: who reviewed it, what was corrected, who called, what happened. One timeline, in order, on one record.

Freeze what arrived. Once accepted, the original referral should remain a read-only record of what was received, separate from the outreach and booking that follow.

Measure the gap. Time from referral to booking should be a number the clinic can see, and a referral nobody has picked up should announce itself.

How Fertiligent does it

Referrals reach the inbox by fax, email or upload, including from an existing patient's own page so that the document arrives already attached to that patient. Each one is read as it arrives: the letter details, the patient, a partner where there is one, the referring physician and the reason for referral are extracted into fields, and each row in the inbox shows a confidence score and, where flagged, why.

The inbox is organised into lanes: Needs me for the referrals waiting for a person, Failed for the ones the system could not process, each with a Retry, and Confirmed and Rejected as history. Opening a referral gathers anything the system is unsure of at the top, under "Needs your attention", and the reviewer must accept or edit each flagged value before confirming. Nothing is re-keyed unless the system misread it.

Matching happens before anything is created. A clear match links to the existing patient; several candidates are put to the reviewer; no match means confirming will create a new record. Banners carry the rest: a mismatch between the cover letter and the attached results, which must be acknowledged before confirming; a possible duplicate within the clinic's duplicate window; the patient's other referrals; and a suggested couple link, which a person confirms or declines. If the patient was previously marked declined, lost or do-not-contact, the system stops and asks for an explicit decision before anyone re-engages them.

High-confidence referrals can be accepted on the clinic's behalf, and they are not filed silently: they sit in their own Auto-confirmed lane with a 24-hour window to undo, and a summary strip at the top of the inbox shows how many were handled automatically, the recent auto-confirm accuracy and the health of the pipeline. The clinic can see what the system is doing and step in.

Confirming is the hand-off: the patient is linked or created at status Referred, a prospect is created or linked with a "Referral received" note on its timeline, and the referral freezes as a read-only record of what arrived. Reaching the patient and booking the consultation happen on the prospect, where every call, text, note and stage change sits in order. A prospect nobody has picked up is flagged overdue once the referral has sat longer than the clinic's turnaround target, and marking it booked records the appointment date, so time from referral to booking is a figure the clinic can report on. One patient, one record (the referral, the outreach and the chart together) is the principle behind our approach to data management for fertility clinics.

The patient is not left guessing. When she calls the clinic line, the voice agent matches her by phone number, verifies her date of birth, and can say one of a few safe things: the referral was received and is in review; it is confirmed and staff will call to schedule; or, if it cannot tell, that a callback has been queued for a person. If details are missing (a health card number, an email address or a mobile number); it can collect them on the same call, saved as a draft for a staff member to review and apply. Nothing a caller says is written onto the referral automatically. What happens after intake (outreach, reminders and the parts that must stay human) is covered in our guide to automating patient follow-up in a fertility clinic.

The fax is not going away this year. What can go away is the fortnight between its arrival and the first moment anyone at the clinic, or the patient, knows what is in it.

See it yourself: take a referral through to a booking in the live demo, a fictional clinic with synthetic data.

References

• [1] Why doctors still use faxes, and why Ontario aims to phase them out of health care, CBC News (2023)
• [2] Dropping the baton: specialty referrals in the United States, The Milbank Quarterly (2011)
• [3] Closing the Referral Loop: an Analysis of Primary Care Referrals to Specialists in a Large Health System, Journal of General Internal Medicine (2018)
• [4] When and why do subfertile couples discontinue their fertility care? A longitudinal cohort study in a secondary care subfertility population, Human Reproduction (2009)
• [5] Female age-related fertility decline. Committee Opinion No. 589, ACOG and ASRM, Fertility and Sterility (2014)

Related:

• Automating patient follow-up in a fertility clinic: what to automate, what to keep human
• AI in Fertility Clinics: Benefits, Implementation, and ROI

Sergei Gorlovetsky, CEO, Fertiligent