SSRIs for Somatic Symptom Disorder Without Comorbid Depression
A single patient with fourteen months of unexplained multisystem symptoms and a normal depression screen. The disagreement isn't whether an antidepressant might help — it's whether prescribing one requires first finding a depression to treat, and if not, which drug class actually fits what she has.
M.A., a 41-year-old woman, has run the events department at a downtown hotel for eleven years — work that trained her to notice when something is off before anyone else does, an instinct that hasn't made the last fourteen months any easier. It started with a dull ache across her shoulders she blamed on standing through back-to-back weddings, then abdominal bloating she attributed to a new diet, then a tremor in her hands she was sure meant something serious. Fourteen months, three specialists, and one unremarkable colonoscopy later, she has a folder of normal results and a symptom list that has only grown: joint pain that migrates, headaches most afternoons, a persistent sense that her heart is beating too hard even when her monitor strips read completely normal. Her PHQ-15 today comes back at 18, moderate-to-severe somatic symptom burden, but her PHQ-9 sits at 6, below threshold for even mild major depression, and she pushes back hard on any suggestion that this is depression she hasn't recognized in herself. She isn't wrong to push back: nothing in her history — sleep, appetite, her interest in her work, her marriage of nine years — reads as depressed. What she has is a felt, physical symptom burden serious enough that she has cut her hours twice this year, without a depressive syndrome underneath it to explain the prescription anyone might reach for next.
The evidence for antidepressants in exactly this situation — real, disabling somatic symptom burden without a depressive syndrome to anchor the prescription to — isn't incidental to her case, it's the actual question in front of the team. Kroenke's 2007 review in Psychosomatic Medicine, which pooled 34 randomized trials in somatoform disorders, found antidepressants beneficial in 4 of the 5 trials that tested them — but Kroenke was explicit that the review could not say whether that benefit works by reducing depression and anxiety generally or by acting on somatic symptoms directly, and named that as the open question the field still had to answer. The trial that speaks most directly to her situation is Muller and colleagues' 2008 study, which recruited 51 outpatients with multisomatoform disorder rather than major depression and found escitalopram separating from placebo on PHQ-15 symptom burden from week six onward — a population selected on somatic symptoms, not on a mood diagnosis. The mechanism most often proposed, descending pain- and symptom-modulation running through serotonergic and noradrenergic pathways, would predict an effect on her diffuse, migrating pain specifically, and might not touch her tremor or her bloating the same way — a genuinely different bet from prescribing an antidepressant simply because a screening tool crossed a line.
Follow-up, after the third normal workup
Fourteen months of a rising symptom count and a PHQ-9 that doesn't clear threshold isn't a reason to wait — it's exactly the population this literature studied. Kroenke's 2007 review pooled 34 randomized trials in somatoform disorders and found antidepressants beneficial in 4 of the 5 that tested them. I'll say plainly what that review does not settle, because it matters here: Kroenke himself flagged that nobody yet knows whether the effect runs through mood or acts on somatic symptoms directly. What closes that gap for me is Muller and colleagues' 2008 trial, which enrolled 51 patients on multisomatoform disorder — not on a depression diagnosis — and still saw escitalopram separate from placebo on PHQ-15 burden by week six. The entry criterion there was her symptom picture, not her PHQ-9. If we wait for her PHQ-9 to cross a threshold that trial never asked its own patients to cross, we're treating the wrong number.
You're right that Muller's patients were enrolled on symptom burden rather than a mood diagnosis — I'm not contesting that data. But notice you had to concede the point yourself: Kroenke declined to say whether these drugs work on symptoms or on mood, and that leaves the whole depression-independence case resting on 51 patients at a single site in one twelve-week trial. That's not nothing. It also isn't the settled foundation "we're treating the wrong number" implies. What concerns me more than effect size is what starting a psychiatric medication tells her about what's wrong with her, fourteen months into a workup she's already experienced as dismissive.
She told us directly she doesn't want to hear this is depression she hasn't recognized in herself — a prescription framed as fixing a chemical problem can land exactly that way with a patient this attuned to being taken seriously, and CBT for somatic symptom disorder has its own real trial base without that framing risk.
Both of you are arguing about whether to prescribe, and I think you're each right about a different slice of her symptom list — which is why I'd narrow the question rather than settle it. Her pain is migrating and diffuse in a way that looks less like general somatization and more like a centrally-mediated pain syndrome, the same population duloxetine and other SNRIs carry dedicated trial evidence in, distinct from the SSRI-in-somatization literature either of you has been citing.
If we start something today, an SNRI targeted at her pain phenotype specifically does real, mechanistically distinct work regardless of which of you turns out to be right about the broader medication-versus-CBT question — and if the answer really is CBT-first, that doesn't have to mean nothing happens about her pain in the meantime.
Agreed: start duloxetine today at a low dose, targeted specifically at the pain component, with a CBT referral placed in parallel rather than sequentially — nobody is asking her to complete psychotherapy before her pain gets addressed.
Not agreed, and carried forward rather than smoothed over:
Duloxetine continues for the pain component alone; the escitalopram question is treated as settled by the psychotherapy response, not revisited.
The psychiatrist's original case for adding escitalopram is reopened — the reason for holding it shouldn't cost her more months of untreated symptom burden while she waits.
The psychiatrist reads today's SNRI start as fully compatible with adding an SSRI later if access stalls; the psychologist reads it as buying CBT the time it needs to actually work, not as a placeholder for a foregone conclusion. Neither treats today's plan as having resolved which of them is right.