Fluoxetine for Depersonalization Disorder: A Real Trial, A Real Negative Result
A nurse with six years of depersonalization has been offered fluoxetine on general principle. The real evidence for that specific drug in that specific condition is a negative randomized trial — and the team has to decide what that negative result should actually mean for her.
L.F. came to psychiatry for a second opinion about a prescription she has not started. Her primary care physician, working from the general reputation SSRIs carry across mood and anxiety symptoms, offered her fluoxetine and suggested she "just try it." She is 31, an emergency department nurse of eight years, most of it on nights because the pace suits her better than daytime traffic does, and she wanted to know what the drug was actually for before she took it. Six years ago, early in her career, she worked a mass-casualty response during a difficult rotation. She describes the days after it more vividly than the event itself — a stretch where nothing felt quite real, sound and light arriving slightly delayed, her own hands looking unfamiliar performing tasks she had done a hundred times. Unlike an acute stress reaction, it never resolved. It is chronic now, present most days at low intensity, worse when she is tired or under unusual pressure, and notably absent during the acute moments of an actual emergency; she has told her PCP, half-joking, that she is somehow more present during a real code than during an ordinary quiet Tuesday. Most Sunday mornings she is at the same climbing gym before her shift that evening, a routine she has kept for six years almost without exception. She has no diagnosable depression and no anxiety disorder meeting full criteria, though she acknowledges background work stress that most ED nurses would recognize.
The evidence that bears on her question is narrower and more specific than SSRI reputation suggests. Simeon and colleagues' 2004 trial in the British Journal of Psychiatry randomized 50 patients with depersonalization disorder to ten weeks of double-blind fluoxetine (mean dose 48 mg/day) or placebo, and found no meaningful separation on the core depersonalization outcomes. It was adequately designed rather than underpowered or poorly run, which makes a negative result in the exact condition being asked about a different kind of evidence than no evidence at all. The trial did report one finding that cuts the other way, and it is the finding her case turns on: depersonalization was significantly more likely to improve in patients whose comorbid anxiety disorder improved. She does not have one.
Before starting a drug with a negative trial behind it
I wouldn't start fluoxetine here primarily for the depersonalization. Simeon and colleagues' 2004 trial randomized 50 patients with depersonalization disorder to ten weeks of double-blind fluoxetine at a mean of 48 mg a day or placebo, and found no significant separation on the core outcome. That's a meaningfully different kind of evidence than general SSRI efficacy in depression or anxiety — it's a specific negative result in the exact condition we're being asked about. Without a separate depression or anxiety diagnosis to justify it independently, we'd be exposing her to real side-effect burden without the evidence base actually behind the target we'd be treating.
The trial is real and I'm not disputing its result. But a population-level null doesn't mean nobody responds, and the strongest version of that argument comes from inside your own trial rather than from clinical anecdote: Simeon reported that depersonalization was significantly more likely to improve where a comorbid anxiety disorder improved. That is a real signal about who responds, from the same 50 patients. She acknowledges background stress around her shift work. A time-limited trial with explicit tracking seems reasonable to me rather than treating one randomized trial as the final word for the individual patient in front of us.
That finding is precisely why I'd hold. What Simeon reported was improvement tracking with improvement in a comorbid anxiety disorder — a diagnosis, with something to treat and something to measure. She doesn't have one. Background stress that most ED nurses would call unremarkable isn't a sub-threshold version of that finding; it's a different thing entirely, and there's nothing there for fluoxetine to improve first. Her presentation is close to the isolated depersonalization population the negative trial most directly describes, not the edge case the exception argument depends on.
I'd refer her to specialized depersonalization-focused psychotherapy — a real, if smaller, evidence base exists there too; an open study by Hunter and colleagues found meaningful benefit from a structured cognitive-behavioral approach targeting depersonalization specifically. Given she's already managing full-time night shifts, avoiding an unnecessary medication trial with no clear target has real value on its own.
Agreed: defer fluoxetine, refer to specialized depersonalization-focused cognitive-behavioral therapy as the first step, and revisit a medication trial explicitly only if a genuine comorbid depression or anxiety target emerges over time.
The psychiatrist's concern about individual variation within a negative trial wasn't dismissed — it was folded into the plan as an explicit condition for revisiting medication later, rather than argued down entirely. All three left agreeing on both the immediate plan and on what would change it, which resolved the disagreement more completely than a split decision would have.