Progesterone for Traumatic Brain Injury: What the Two Negative Phase 3 Trials Actually Found
A single patient, recovering well from a complicated mild traumatic brain injury. The disagreement isn't about her care plan — it's about how honestly to answer her mother's question about a hormone therapy two large trials tested, found no benefit from, and never studied in a patient at her level of injury.
S.T., a twenty-six-year-old woman, is most of the way through her first year of a part-time master's program in education, taken on top of a full-time job teaching third grade, when a car runs a red light and T-bones her sedan on the driver's side as she's driving home from an evening class. She has no memory of the impact itself and only a fragmented, unreliable memory of the twenty minutes afterward — paramedics on scene recorded a Glasgow Coma Scale of 13, confused but following commands, with a brief loss of consciousness reported by a bystander who stopped to help.
Head CT in the emergency department showed a small right temporal contusion with no mass effect and no indication for neurosurgical intervention; the trauma team admitted her for observation given her ongoing confusion and amnesia to the event. Her Glasgow Coma Scale of 13 places her in the mild band of the standard scale — mild is 13 to 15, moderate 9 to 12, severe 8 or below — and the contusion on her scan makes hers a complicated mild injury rather than a moderate one: structural damage, but a level of consciousness two full points above where the moderate range even begins. That single number is what quietly removes her from the literature her mother is about to raise. Every progesterone trial worth citing set its ceiling at or below a GCS of 12. She has no significant medical history — no prior head injury, no seizure history, no anticoagulant or antiplatelet use — which is part of why the team already has an unusually optimistic read on her trajectory. By the next morning she is more oriented, still hazy on the hour before the crash, vitals and repeat exam reassuring, on a trajectory that already has the team thinking about a short admission.
Her mother, who has barely left the bedside, has spent the overnight hours reading everything she can find, and has arrived at rounds with a specific question rather than a general one: she found an article, more than a decade old, describing a hormone that in an early trial cut deaths from brain injury nearly in half, and wants to know why her daughter isn't getting it. She is not asking out of panic without basis — the article she found described something that was, at the time it was published, a genuinely exciting result. What she does not know, because the coverage she found predates it, is what happened when that early result was tested at the scale it needed to be tested at.
At rounds, answering the mother's question
I don't think we should treat this as a closed question just because the headline answer is no. The trial her mother found — the original single-center study by Wright and colleagues, published in 2007 — really did show a substantial reduction in mortality with progesterone after traumatic brain injury: a hundred patients, death from any cause 13 percent on progesterone against 30 percent on placebo. It's not unreasonable for a family to ask why that result didn't become practice. I'm not arguing we give it. I'm arguing that the honest answer to her question is more complicated than "it doesn't work," and that a family who did real reading deserves a real explanation of what changed between that trial and now, not a one-line no.
To be direct about where I land: I don't think progesterone belongs in her regimen today. My position here is entirely about how we have this specific conversation with her mother, not a recommendation to prescribe.
You're right that the early result was real and that a one-line no doesn't honor how carefully she's read this. But what changed is about as clean an answer as trial methodology ever gives us: two large, independent phase 3 trials — ProTECT III, run across dozens of U.S. centers, and SYNAPSE, run internationally — both tested progesterone against placebo and both came back negative. I want to be precise about how, because the two failed differently: ProTECT III was stopped early for futility at a planned interim look, 882 patients into a planned 1,140, with favorable outcomes in 51 percent on progesterone against 55 percent on placebo. SYNAPSE was not stopped early at all — it enrolled its full 1,195 patients and reported no clinical benefit on either its primary or its secondary analyses. A completed trial and an abandoned one, arriving independently at the same place, is a stronger answer than two abandoned ones would have been.
And there's a second point that matters more for her daughter than either result. ProTECT III enrolled a Glasgow Coma Scale of 4 to 12; SYNAPSE required 8 or below; the 2007 trial her mother read about enrolled 4 to 12 as well. S.T. is a 13. Whatever those trials established or failed to establish, they established it in patients meaningfully worse off than she is — so the honest answer to her mother is not "we tested it and it doesn't work in patients like your daughter," it's that nobody has tested it in patients like her daughter at all, and the only populations where it was tested showed nothing.
The single-center 2007 result you're citing is the exact shape of finding that phase 3 trials exist to test, and in this case, testing it twice independently produced the same answer both times.
Setting the efficacy argument aside entirely, I'd also want her mother to understand that this was never a costless option to decline. Progesterone in those trials required days of continuous IV infusion — 72 hours in the 2007 study, four days in ProTECT III, five in SYNAPSE — and ProTECT III found infusion-site phlebitis or thrombophlebitis three times more common on progesterone than on placebo. I'll be fair about it: most of those episodes were self-limited and not classified as serious, and SYNAPSE found no meaningful safety difference at all. It isn't a dangerous drug. It's an intrusive one. She's on a two-day trajectory right now: oriented, stable, headed toward discharge. Committing her to a multi-day infusion and the line care and monitoring that comes with it isn't a small ask for a drug two independent trials already found doesn't work. I think that's actually the more honest thing to tell her mother — not just that the evidence says no, but that saying yes would have cost her daughter something real for a benefit nobody has been able to show.
Agreed: no progesterone, and the team returned to her mother directly rather than delivering the decision through a resident — walked through the 2007 result, the two later trials, and why a subgroup hint in a negative study isn't grounds to treat outside one. She asked good follow-up questions and, by the team's read, left the conversation satisfied that her daughter's care hadn't skipped a real option out of institutional inertia.
Not fully agreed: how much of this kind of explanation belongs in every family's hands versus being offered only when specifically asked for, the way it was here. The neurologist would rather build a brief, plain-language explanation of superseded early findings into every TBI admission carrying structural injury on imaging — a group that would have included S.T., whom a moderate-to-severe threshold would have missed — on the theory that families increasingly arrive having read something online regardless of whether they ask about it out loud. The hospitalist thinks that risks manufacturing doubt in families who never had the question in the first place, and would rather keep it responsive, as it was today. Neither view changed the plan for S.T. herself.