Recurrent Carpal Tunnel Syndrome: One More Injection, or Time to Refer for Surgery?
A repeat corticosteroid injection worked once before. New EMG evidence of early nerve damage raises whether trying it again is a reasonable second attempt, or a delay the nerve itself can't afford.
Harold P., a 63-year-old retired carpenter, first developed numbness and tingling in his right thumb, index, and middle fingers about fourteen months ago, worse at night and when driving. A corticosteroid injection into the carpal tunnel six months ago gave him roughly four months of real relief before the numbness crept back, now joined by something new: he has started dropping small objects, a spool of thread, his car keys, without warning. On exam his thenar eminence looks subtly flatter compared to the opposite side, and repeat nerve conduction studies done this week, compared against the study from fourteen months ago, now show a measurable drop in compound motor action potential amplitude alongside the persistent slowing — a finding consistent with early axonal loss, not just the demyelinating conduction delay his first study showed.
That distinction is the whole disagreement in today's visit. Corticosteroid injection has real short-term efficacy here — Ly-Pen and colleagues' 2005 randomized trial found injection matched surgical decompression for symptom relief at one year — and his own four months of genuine relief is a data point some clinicians read as predicting he would respond again. Two findings complicate reading it that way. The same group's two-year results showed the advantage had shifted to surgery, and when Andreu and colleagues examined that trial's neurophysiologic outcomes in 2014, only the surgical arm improved on nerve conduction parameters at twelve months: injection relieved what Harold felt without changing what his nerve was doing. Meanwhile the features long identified as predicting poor injection response and early relapse — thenar wasting, symptoms running beyond a year, and motor findings on electrodiagnostic testing — are precisely the three he has acquired since the injection that worked. A steroid injection reduces the synovial inflammation contributing to compression; it does not restore axons already lost. The nerve in front of the team today is not the nerve his first study described, and it is the second study, not the first response, that describes what an injection would now be treating.
Hand clinic follow-up, six months after the first injection
His first injection gave him roughly four months of real, meaningful relief, and that's not nothing — a genuine positive response to injection is something some clinicians read as predicting he'd respond to a repeat course as well, and Ly-Pen and colleagues' randomized trial did find injection holding its own against surgery for symptom relief out to a year. I'd consider a second injection before jumping straight to a surgical referral, especially since he tolerated the first one without any complication.
I understand the logic, and I agree his first response was real.
But the EMG this week isn't describing the same problem his first study did. Fourteen months ago, the conduction slowing was there but his motor amplitude was preserved — a compressed but structurally intact nerve, exactly the kind of finding an injection's anti-inflammatory effect can meaningfully help. This week's study shows a measurable drop in that same amplitude, and it isn't an isolated lab finding — it lines up with something he told us himself, that he's started dropping small objects without warning. An injection reduces inflammation around the nerve; it does nothing to reverse axons that have already been lost, and repeating it treats last year's problem, not this year's.
I'd frame this as urgency, not as a binary between one more injection and nothing else. The part I'd lean on is Andreu and colleagues' 2014 neurophysiologic analysis of the Ly-Pen trial: injection relieved symptoms without improving nerve conduction at twelve months, and only the surgical arm moved those numbers. In a nerve already shedding motor amplitude, buying symptom relief without changing conduction is buying time we don't know he has. I want him on the surgical schedule this week.
That doesn't mean he has to be in pain while he waits for the appointment — an interim injection for comfort isn't unreasonable, as long as everyone, including him, understands it's a bridge to surgery, not an attempt to avoid it.
Expedited surgical referral placed this week given the new EMG evidence of axonal loss. A bridging corticosteroid injection offered and accepted for interim comfort, explicitly framed to Harold as a bridge to surgery rather than an alternative to it.