A family physician described the experience with a precision that no health services paper has improved on:
> "I'll do up a complete assessment on a patient with a really bad knee, refer her to orthopedics, and three months later I hear back that that specialist only does shoulders."
Three months. A complete assessment. A patient in pain. And the referral fails not because of a clinical error but because the referring physician did not know a fact about the specialist that the specialist has known for years.
Now hold that alongside a second fact from the same body of research. The referring doctor "usually has no idea whether that specialist has a short or long waiting list," which is why referrals get made, in the words of that reporting, "on a 'who you know' basis."
Two pieces of information determine whether a referral works: how long the wait is and whether this specialist actually handles this problem.
Neither is available to the person making the decision.
The waits, and where they are going
The scale of what patients are being routed into has grown considerably.
From AMN Healthcare and Merritt Hawkins survey work in 2025:
- Average new-patient wait across 15 large US metropolitan areas: 31 days, up from 26 in 2022 and 21 in 2004.
- Obstetrics and gynecology: 41.8 days. Gastroenterology: 40 days.
- Individual waits recorded up to 231 days.
In Canada, where the measurement tradition is longer, the Fraser Institute found the median wait from general practitioner referral to treatment reached 30.0 weeks in 2024, up from 27.7 in 2023 and 9.3 weeks in 1993.
So waits are long, lengthening, and vary enormously between individual specialists in the same city and specialty.
That variance is the operative fact. If every orthopedic surgeon in a metro had a 31-day wait, opacity would cost little. The reality is that one has a two-week wait and another has four months, and the referring physician is choosing between them with no information.
The one factor that matters and cannot be known
Research in Annals of Family Medicine asking primary care physicians what determines specialist selection found appointment timeliness rated of major importance by 55.4 percent, placing it among the top considerations alongside medical skill and prior personal experience.
It is the single most important referral factor with no available data anywhere.
Physicians know it matters. They have no way to act on it. So they use proxies: memory, habit, the practice that seemed responsive last year, whoever the staff can reach on the phone.
The most instructive evidence about what happens when this changes comes from a BMC Primary Care study in Ontario. Researchers showed family physicians actual wait data derived from their own electronic records for the specialists they were referring to.
The measured median wait was 42 days, with a 75th percentile of 80 days.
The physicians' response was immediate and specific. They said they would re-route non-urgent cases. One captured the reasoning exactly:
> "I don't want them to wait 4 months to suffer needlessly if they could see somebody within 2 weeks."
Given the information, referring physicians change behavior straight away. They are not indifferent to wait times. They have simply never been able to see them.
Why no specialist publishes a wait time
Here is the structural trap, and it is a genuinely hard one.
Imagine you are a specialist with a two-week wait in a market where the average is two months. Publishing that fact would be excellent marketing.
For about a month.
Then every referrer in the region routes to you, your wait becomes four months, and the advantage evaporates while your practice becomes unmanageable.
Now imagine you have a four-month wait. Publishing that guarantees referrals go elsewhere, permanently, including the complex cases you actually want.
In both directions, publishing is against the specialist's interest. So nobody publishes, and the equilibrium is universal opacity. This is not obstruction; it is a rational response to the incentives, and no amount of exhortation will change it.
Scope opacity has a similar structure. The orthopedic surgeon who only does shoulders is listed as an orthopedic surgeon because that is what the directory field contains. Publishing a narrower scope reduces inbound volume, and no directory has a field granular enough to hold it anyway.
The referrer, who needs both facts most, is the only party with no way to obtain either and nothing to trade for them.
The two systems that have partly solved it
It is worth noting that this is not unsolvable, because parts of it have been solved elsewhere.
Ontario and Alberta have built electronic referral systems that display wait times to referring physicians. The mechanism works because a single-payer system can mandate participation, which resolves the incentive problem by removing the choice.
The periodic secret-shopper survey conducted by AMN and Merritt Hawkins produces the US wait-time numbers cited above. It is genuinely valuable and it is a snapshot of metro averages every few years, which is not what a physician making a referral on a Tuesday needs.
US payers show nothing. Their directories, as covered earlier in this series, struggle to maintain accurate addresses, let alone live availability.
So the demonstrated solution requires either a mandate or a party that both sides trust. American healthcare has neither, which is why the problem persists here and not everywhere.
The insight the Canadian study points to
The Ontario research contains a suggestion that most readers pass over, and it may be the most practical path available.
The wait data shown to those physicians was derived from their own electronic records. It came from observing what actually happened to patients they had already referred.
Which means the information already exists, distributed across every referring practice in the country. Every practice knows, from experience, roughly how long its patients wait to see each specialist it uses, and which referrals get bounced back as out of scope.
Nobody pools it.
That reframes the problem substantially. Wait-time transparency does not have to come from specialists, who have every reason not to provide it, or from governments, which in the US will not mandate it. It can come from referrers pooling what they already learn the hard way.
The referrer's data has a useful property too: it measures the thing that actually matters, which is not the practice's stated availability but the realized wait for a real patient with a real insurance plan and a real problem.
And it has an obvious failure mode that determines the design. A pooled wait-time resource that practices can see and influence will be gamed. Which means it only works among verified referring clinicians, reporting their own observed experience, in a space practices cannot manipulate.
The rural case, where opacity becomes triage
Everything above describes an inefficiency in well-supplied metropolitan markets. In rural practice the same opacity becomes something closer to a clinical safety problem.
A rural family physician referring for a specialty that has three plausible options within a two-hour drive is not choosing between a two-week wait and a two-month wait. She may be choosing between a four-month wait and a six-month wait, with a fourth option that requires an overnight stay for the patient.
In that setting, knowing which of the three has capacity this month is not an optimization. It determines whether the patient is seen this year.
The evidence on rural specialist access underlines how thin the options are. In one rural region study, roughly 70 percent of primary care physicians reported having no psychiatrist to readily refer to, and half of the counties studied had no psychiatrist at all.
And rural referrers have the least ability to compensate for opacity through personal networks, because the referral literature is clear that referral choice defaults to who you know, and a physician practising in a small community has fewer accumulated relationships at distant referral centers than an urban colleague who trained locally.
The information gap is widest exactly where the consequences are largest, which is the recurring pattern in almost every coordination failure in this series.
What would work
Sub-specialty scope at the granularity that matters. Not "orthopedics." Shoulders. Not "gastroenterology." Advanced endoscopy. The three-month knee referral to a shoulder surgeon is a taxonomy failure before it is anything else.
Realized wait times contributed by referrers. What your patients actually experienced, by specialist, updated continuously. Verified contributors only.
Specialist-declared scope and current capacity, visible to verified referrers only. This is the trade that makes participation rational for the specialist. Declaring "I do shoulders, I currently have a six-week wait, send me the complex revisions" produces better-matched referrals rather than simply more of them. Specialists do not object to volume; they object to inappropriate volume, and this is the mechanism that reduces it.
A bounce-back record. Out-of-scope referrals are pure waste for everyone, and tracking which specialists bounce which referral types is the fastest route to fixing the taxonomy.
And crucially, not public. A public wait-time database creates the gaming incentive that keeps this from existing. A peer-visible one, contributed by verified clinicians, does not.
A note on urgency signaling. Part of what makes wait times so damaging is that referrals carry almost no reliable urgency information. Every referral marked urgent by a worried referrer competes with every other one, and specialist offices learn to discount the flag. A shared, verified convention for what urgent means, honored in both directions, would do more for actual triage than any additional capacity.
What you can do now
If you refer
Track your own realized waits. For every referral, note the date sent and the date the patient was actually seen. Within three months you will have a better dataset than anything published, specific to your patients and their insurance.
Ask the scope question directly, once. "What do you actually want to see, and what should I send elsewhere?" A five-minute call per specialist eliminates the shoulder-surgeon-knee-referral category of failure permanently.
Ask for the current wait when you call. Front desk staff will usually tell you. Nobody asks, because nobody thinks to.
Share what you learn with your colleagues. Your practice partners are referring blind to the same specialists. A shared internal document is the highest-yield fifteen minutes available.
If you receive referrals
Publish your scope to your referrers. Not to the public. A single page to the practices that refer to you, stating what you want and what you would rather they send elsewhere. This reduces inappropriate referrals, which is the volume you actually want to lose.
Tell referrers your current wait. Bad news delivered early is better than a patient waiting three months and then being triaged out. Referrers will route around a long wait for non-urgent cases and still send you the urgent ones, which is the correct outcome.
Bounce out-of-scope referrals quickly and specifically. "Not my area, send this to X" within a week is enormously more useful than a decline after eight weeks.
If you run a practice or region
Measure your realized referral waits. The Ontario study demonstrated this is derivable from records you already hold. Almost no US practice does it.
Consider a regional pooling arrangement. A handful of practices sharing observed waits and scope for the specialists they all use would replicate most of the value of a formal system, without requiring anyone to build one.
Frequently asked questions
How long do patients wait to see a specialist? Average new-patient waits across 15 large US metropolitan areas reached 31 days in 2025, up from 26 days in 2022 and 21 days in 2004, with obstetrics and gynecology at 41.8 days, gastroenterology at 40 days, and individual waits recorded up to 231 days. In Canada, median wait from GP referral to treatment reached 30 weeks in 2024.
Do referring physicians know specialist wait times? Generally not. Research and reporting consistently find referring physicians have no reliable visibility into how long a given specialist's wait is, which is why referral choice frequently defaults to personal familiarity. Appointment timeliness is rated of major importance by 55.4 percent of primary care physicians and is the top factor with no available data.
What happens when physicians are shown wait time data? They change behavior immediately. In an Ontario study where family physicians were shown wait times derived from their own records, showing a median of 42 days and a 75th percentile of 80 days, physicians said they would re-route non-urgent referrals to shorter queues.
Why don't specialists publish their wait times? Because publishing works against them in both directions. A short wait, once published, attracts enough referrals to eliminate the advantage and overwhelm the practice. A long wait, once published, diverts referrals permanently including the complex cases the specialist wants. Universal opacity is the rational equilibrium.
Why do referrals get sent to the wrong sub-specialist? Because directory taxonomies operate at specialty level while practice operates at sub-specialty level. An orthopedic surgeon who only treats shoulders is listed as an orthopedic surgeon, and nothing in the referral pathway surfaces the distinction until the patient has waited months.
Has anyone solved wait time transparency? Partially. Ontario and Alberta have built electronic referral systems displaying wait times, which works because a single-payer system can mandate participation. In the US, periodic secret-shopper surveys produce metro-level averages every few years, and payer directories do not show availability at all.
The bottom line
Two facts determine whether a referral succeeds: how long the wait is, and whether this specialist handles this problem.
Neither is knowable by the person making the decision. Both are known perfectly well by the specialist's office. And the specialist has a rational interest in publishing neither.
So a physician performs a full assessment, selects a specialist from memory, and finds out three months later that the patient waited to see someone who does not treat their condition.
Meanwhile every referring practice in the country is learning these facts continuously, one disappointed patient at a time, and throwing the knowledge away.
The information exists. It is generated daily, by the people who most need it, at considerable cost to their patients. Nobody has ever pooled it, because pooling requires a group of verified professionals willing to tell each other the truth about a system that no single one of them can fix alone.
Part of a series on the missing professional infrastructure of healthcare. Previously: "Who Has a Cohort of X?"
Evidence note: US wait time figures come from AMN Healthcare and Merritt Hawkins survey work (2025), which uses a secret-shopper methodology across 15 large metropolitan areas. Canadian wait figures come from the Fraser Institute's annual survey (2024). Physician response to wait-time data comes from BMC Primary Care (2022). Referral selection factor rankings come from Annals of Family Medicine (2004). Practitioner quotations come from Healthy Debate reporting (2017). Wait time measurement methodologies vary considerably between sources and are not directly comparable.
