Duloxetine for Stress Incontinence: A Drug Approved in Europe, Never Approved Here
The same drug, the same trial data, and two regulators who looked at it and reached opposite conclusions — leaving a US clinician to decide what an off-label prescription actually means here.
Angela T., 52, has coached her town's youth soccer league for eleven years and had to stop running drills with her own team last season after leaking through her shorts during a sprint in front of the players' parents — the moment, she says, that finally made her call a doctor instead of just buying darker-colored athletic wear. She is otherwise healthy, perimenopausal with regular cycles still, no diabetes, and has never been treated for depression, anxiety, or any other psychiatric condition; her only regular medication is a daily multivitamin.
Her six completed months of supervised pelvic floor physical therapy, with only modest improvement, are what place her inside the European label's own entry condition — moderate to severe stress incontinence unresponsive to conservative measures — rather than merely near it; the leakage still occurs with coughing, sneezing, and any running or jumping motion, bothersome enough that she has quietly stopped several activities rather than risk it in public again. A midurethral sling is the next guideline step, but she has read enough about mesh litigation in the popular press to decline surgery outright for now, asking instead whether "the pill version" she found described online might work. That pill is duloxetine, approved in Europe and Canada for exactly her condition, moderate to severe stress incontinence unresponsive to conservative measures — but never approved for this indication in the United States. Eli Lilly withdrew its US application in 2005 after the FDA identified a higher-than-expected rate of suicide attempts in the open-label extension phase of the same trial program that supported European approval. A later independent reanalysis of that program's randomized phase — Maund, Guski and Gøtzsche, CMAJ 2017, working from the clinical study reports themselves — recorded no suicidality, violence or akathisia events at all in the controlled data, which narrows the FDA's specific claim considerably; it also found one patient harmed for every seven treated by an activation event and concluded outright that duloxetine's harms outweighed its benefits. Her own psychiatric history is entirely negative, so the disputed signal is not the part of that reanalysis that describes her — the harm-benefit verdict is.
Two regulators, one trial program, and her own clean history
I wouldn't prescribe duloxetine for this. The FDA didn't reject this indication over a routine dispute — it identified a higher-than-expected rate of suicide attempts specifically in the open-label extension of Lilly's own controlled trials for stress incontinence, and Lilly withdrew the application rather than contest it. Prescribing off-label against the agency's own stated finding for this exact indication is a different kind of decision than ordinary off-label use of an already-approved drug.
I'd draw the line differently on the specific claim, though not in the direction you might expect. Open-label extension data is unblinded, uncontrolled, and already enriched for patients who chose to continue therapy — not the randomized comparison the drug was actually tested against. Maund, Guski, and Gøtzsche went back to the European Medicines Agency and obtained the full clinical study reports for the randomized phase itself — four placebo-controlled trials, 1,913 patients — and recorded no suicidality, violence, or akathisia events in that controlled data at all.
But I won't let that be heard as a defense of the drug, because it isn't one. The same paper put the number needed to harm at 7 for stopping because of an adverse effect and 7 for an activation event, described two duloxetine patients with five serious events of severe depression, panic attacks and severe anxiety, and concluded that the harms outweighed the benefits. So my position is narrower than yours and narrower than it sounds: the FDA's particular suicide-attempt finding does not survive contact with the randomized data, and the best independent read of those same data still argues against prescribing. Those are two separate conclusions and I hold both.
I'd add one more layer before either of you finalizes a recommendation. Angela has no personal or family psychiatric history — no depression, no prior suicidality, nothing that would put her inside whatever population, if any, actually carried the original signal. That's a real, checkable fact specific to her, distinct from the population-level regulatory dispute.
It doesn't resolve whether we should prescribe an indication the FDA specifically hasn't approved — that's still her decision to make once she understands both readings of the evidence, not something her clean history settles on its own.
Agreed: Angela was given the regulatory history in full — the FDA's stated concern, the CMAJ reanalysis that found no suicidality or violence events in the randomized phase, and that same reanalysis's separate conclusion that duloxetine's harms outweighed its benefits — and asked to take the decision home rather than have it made for her in the room.
Not agreed among the three voices: the urogynecologist would not personally write the prescription regardless of what Angela ultimately wants, citing the FDA's specific, indication-matched finding; the clinical pharmacologist would prescribe it only if she asks for it after hearing that the same reanalysis clearing the suicidality charge also judged the drug's harms to outweigh its benefits, and treats her negative psychiatric history as narrowing one risk rather than settling the question; the primary care physician's position is that whichever physician does prescribe it should document the disputed regulatory history explicitly in her informed consent, regardless of which way the decision goes.