A worker tells their supervisor their shoulder is bad. What happens in the next 10 minutes will influence the length of that claim more than anything a clinician does in the following 10 weeks.
That is an uncomfortable claim to make about a conversation, so here is the evidence for it.
Delay is the variable
Of everything that can be changed about how an employer handles an injury, the interval between the injury and the start of treatment is the one with the clearest published relationship to outcome.
RRRadj = 1.4, 95% CI 1.03, 1.8
Raised risk of not returning to work when treatment is delayed
Claims do not run long because the tissue is unusually stubborn. They run long because the clock started and nothing happened. Every week of waiting adds deconditioning, adds fear of movement, and adds the worker's own growing suspicion that this is permanent.
The Ontario low back pathway
The most instructive Canadian example is a low back pathway that changed nothing about the treatment and everything about the waiting.
a mean of 6 (range 1-19) months compared with 38 days
Wait to spine surgical assessment, before and after the pathway
Assessment wait went from a mean of 6 months to 38 days. Imaging volumes fell. Nobody invented a new technique. The gain came from removing the queue, which is a scheduling problem dressed up as a clinical one.
Return-to-work timing, at a hospital
The same shape shows up in a return-to-work program at a Canadian hospital, where the change was early contact and supervisor involvement rather than a new therapy.
45% improvement
Improvement against a peer group of 29 hospitals, which improved 25%
The peer comparison is the useful part. 29 other hospitals improved 25 per cent over the same period; this one improved 45. The difference was the process, not the medicine.
What the supervisor should actually do
4 things, and none of them require a clinical qualification.
- Listen to the whole sentence. The most common failure is a supervisor who hears "my shoulder is bad" and starts problem-solving before the worker has said which movements make it worse. That detail is the entire content of the report.
- Do not diagnose, and do not reassure. "That's probably just a strain" is both outside the supervisor's scope and, if wrong, the sentence the worker will quote back for the next 18 months. "I don't know what that is, let's find out today" costs nothing and commits to nothing.
- Do something the same day. Not necessarily treatment - a booked appointment, a task swap, a note in the log. The point is that the worker leaves the conversation with the clock visibly running rather than stopped.
- Write down what was said. Not for the file, for the plan. Whoever builds the accommodation in 3 weeks needs to know which movements hurt on day 1.
Why objective capacity data changes this conversation
A supervisor asked to place someone on modified duty is being asked a clinical question with no clinical information. They guess, usually low, because guessing low feels safer. The worker ends up in a role well below what they can actually do, gets bored, gets treated as fragile, and stays in the accommodation longer than they needed to.
A measured capacity number replaces the guess. It says what this person can lift, push and hold today, benchmarked, which turns "what should we let him do" into a lookup rather than a negotiation.
The 4 steps above are our own practice guidance, not a published protocol. The figures are published Canadian findings from named programs, each linked to its source, and none of them is a wereset result.
Further reading
Return to work sets out the statutory duty by province and what an accommodation plan contains. The insurer view covers the same evidence from the carrier's side of the table.
