Knee Osteoarthritis With Diffuse Pain: Duloxetine as Both Treatment and Diagnostic Test
Her knee imaging doesn't match how widely her pain is described. Duloxetine is approved for exactly this picture — but the group disagrees about whether prescribing it settles the question or begs it.
Denise M., 58, processes insurance claims from a desk she says she now dreads sitting down at, not because of her job exactly, but because sitting for any length of time has become part of what sets her knees off, and standing up afterward has become its own small ordeal. She has worked the same job for over twenty years, through a divorce and, more recently, the stress of a teenage son who has struggled at school, both of which she mentions in passing rather than as complaints. What brought her in was knee pain, and her imaging shows real but modest disease — Kellgren-Lawrence grade 2 in both knees, nothing that obviously explains the picture she actually describes. The ache, by her account, doesn't stay in her knees; it spreads through her legs and lower back some days, she isn't sleeping well, and she describes a tiredness that feels different from simply being in pain, more like something underneath it.
That mismatch — mild-to-moderate structural disease against a diffuse, amplified symptom picture — is the profile the pivotal duloxetine trials for knee osteoarthritis were built around, and the drug carries an FDA approval for chronic musculoskeletal pain broadly, plus a conditional ACR recommendation specifically for knee osteoarthritis. Chappell's 2009 trial, 231 patients over thirteen weeks, found duloxetine superior to placebo on average daily pain from the first week onward. A companion trial escalated patients who had not reached a 30% pain reduction by week seven to 120mg — and a later pooled analysis of both trials found that escalation bought those non-responders nothing measurable, which makes week seven a point at which to judge whether the drug is working rather than a point at which to give more of it. But nothing about Denise's evaluation has actually tested whether her diffuse pain reflects a centrally-mediated component of her osteoarthritis specifically, or something else that happens to look similar from the outside — and that open question is exactly what the group is now arguing about before reaching for the prescription pad.
Diffuse pain, modest imaging, and an unconfirmed mechanism
Denise's picture is close to what the pivotal duloxetine trials for knee osteoarthritis were designed around — pain out of proportion to imaging, poor sleep, a diffuse quality that a purely peripheral anti-inflammatory approach hasn't touched. She's already tried NSAIDs and acetaminophen without meaningful relief.
Duloxetine has a real FDA approval for chronic musculoskeletal pain and an ACR conditional recommendation specifically for knee osteoarthritis. I'd start at 60mg, which is the dose Chappell's trial actually established, and hold her there. I want to flag one thing early so nobody reaches for it later: the reflex with an SNRI non-responder is to double the dose, and in this indication that reflex has been tested and failed — the pooled analysis of both pivotal knee-osteoarthritis trials found escalating non-responders to 120mg gave no additional benefit over staying at 60.
What gives me pause is that nobody has actually asked Denise directly about her mood, her stress, or her sleep as their own subject, separate from the knee complaint that brought her in. She told us herself she's under real situational stress right now.
Diffuse pain, poor sleep, and fatigue describe a lot of things, not just centrally-sensitized osteoarthritis. Starting an SNRI because the pattern fits a trial population isn't the same as confirming that pattern's actual source in her case, and if there's an underlying mood or sleep disorder driving this, duloxetine could blunt the picture without addressing what's actually going on.
I don't think we have to resolve the mechanism question before we act, if we're honest about what the trial itself is actually testing. Start duloxetine at 60mg, and use the same threshold the pivotal trials used — a 30% pain reduction by week seven, not just 'feeling a bit different.'
And the escalation point matters more than it looks. If 120mg were on the table for a non-responder, week seven would stop being a clean read — we'd be extending an ambiguous result instead of interpreting it. Because the evidence says escalating buys nothing here, week seven stays a genuine fork: continue, or stop and turn to the mood and sleep workup.
If she clears that bar, that response is real information suggesting a centrally-mediated component was genuinely contributing, worth continuing on its own terms. If she doesn't, that's evidence pointing the other way, and it becomes the reason to pursue the mood and sleep evaluation directly rather than the reason we skipped it. Either way, we learn something we don't currently know.
Agreed: start duloxetine 60mg daily and hold that dose, with a week-7 assessment against a 30% pain-reduction threshold — continuing if she clears it, stopping and redirecting if she doesn't, rather than escalating to 120mg, which the pooled trial data found does not help non-responders in this indication. A follow-up specifically addressing sleep and mood, separate from the knee conversation, was also scheduled regardless of how she responds to the medication.
Not resolved: whether a good response to duloxetine at week seven should be read as confirming a centrally-mediated pain component, as the rheumatologist proposed, or whether it would remain genuinely ambiguous — since SNRIs can improve mood and sleep directly, a response wouldn't cleanly rule out the concern the primary care physician raised. Both agreed the scheduled mood and sleep evaluation should proceed either way, rather than being canceled if the medication happens to help.