Return-to-play decision

How can a doctor clear an athlete when no test can show the injury has healed?

A player takes a blow to the head. He is examined, held out, monitored, and eleven days later declared fit. But a concussion leaves no fracture line to watch knit together. Standard imaging of a concussed brain is typically normal — that is close to the definition of the injury. There is no scan, no blood marker in routine use, no measurement that returns "healed."

So what exactly did the doctor learn between the injury and the clearance? If no instrument can see the thing being decided, on what basis was the decision made — and is it a decision at all, or a guess wearing a white coat?

Begin by noticing that "we cannot measure the state directly" is not the same as "we know nothing." Consider how you would decide whether a house's wiring is repaired when you cannot open the walls. You would not stare at the plaster. You would switch things on — one at a time, from the least demanding load upward — and watch for a fault.

That is precisely the structure of the answer. The clinician cannot observe brain recovery, but he can observe the brain's response to load. And the injured brain, unlike the healed one, produces symptoms when exerted. So the unobservable state is inferred from an observable behaviour under deliberate provocation.

This reframes the whole exercise. The doctor is not running a test of healing. He is running a sequence of stress tests, each of which can only fail — and a run of non-failures is the evidence.

Why a sequence rather than one hard test? Two reasons, and both are about the cost of being wrong. First, a graded ladder finds trouble at low load, where a symptom is a headache rather than a second head injury on the field. Second, each step that passes is weak evidence, but weak evidence accumulates — passing light aerobic work, then sport-specific movement, then non-contact drills, then full contact is a chain of independent-ish observations, and the chain is far stronger than any link.

The international consensus formalises this. The Amsterdam statement from the 6th International Conference on Concussion in Sport (2023) sets out a graded return-to-sport strategy of six stages, with a minimum of about 24 hours at each, alongside a parallel return-to-learn or return-to-work progression. Symptom recurrence at any stage sends the athlete back a step. Note that the timing rule is doing real work: the delay exists because symptoms can be delayed too, so a same-day pass would not be evidence of much.

Now the part that makes this a genuinely hard decision rather than a checklist. Two errors are possible, and they are wildly asymmetric. Clearing too early risks a further injury to a brain that is still vulnerable, with recovery times that lengthen and, in the worst and thankfully rare case, catastrophic outcomes — though the mechanism usually named there, second impact syndrome, remains debated in the literature as to whether it is a distinct entity. Holding too long costs games, wages and a career window. When errors are asymmetric, the rational threshold is not "more likely than not." It is deliberately shifted towards the cheaper mistake, which is why protocols are conservative by design and why they are stricter for children, whose recovery is slower.

One more piece. Almost all of the evidence is self-reported. The athlete says whether he has a headache — and he has enormous incentive to say no. That is why the decision is placed with an independent physician rather than the team's, why leagues introduced spotters and mandatory removal rules, and why baseline testing exists: a pre-season measurement so that a post-injury score has something honest to be compared against. The protocol is engineered around a witness with a motive.

Three refinements.

First, clearance is not a claim of certainty. It is a statement that, having applied a graded series of provocations without reproducing symptoms, the probability of persisting injury has fallen below an agreed threshold. That threshold is a policy choice, not a biological fact, which is why protocols differ between leagues and why they have tightened over time as the costs of the false clear were better understood.

Second, feeling well is not the same as being recovered. Symptoms often resolve before physiological measures normalise, which is exactly why the ladder continues past the point of symptom resolution rather than ending there. An athlete who says he is fine on day three has supplied one piece of evidence, not a conclusion.

Third, the shape of this reasoning is general. It is the same move made whenever a state cannot be observed but its behaviour under load can: pressure-testing a pipeline, staged rollout of software, a cardiac stress test. In each case the certificate says "it did not fail under increasing, deliberately applied demand" — never "it is sound."

• Blood biomarkers such as GFAP and UCH-L1, and how close they are to giving concussion an objective test.
• Why paediatric protocols are more conservative, and what the evidence on developing brains actually supports.
• The same inference structure elsewhere: ACL return-to-sport criteria, which do have objective strength and hop-test measures, and what changes when the state becomes partly observable.

Every term the collection defines is gathered in the glossary.