Juniper Vale: Mark, hi, you look like you've got something on your mind. Did you scroll past a headline again?
Mark Delaney: Ha, I did. Three drugs beat Prozac. And I kept thinking about some person, call her Dana, forty-one, three years on fluoxetine, reading that in her kitchen. Thumb over the pharmacy app.
Juniper Vale: Wait, and then what? Does she switch?
Mark Delaney: That's what I wanna sort out. Like, why does this study say that, and can we trust it?
Juniper Vale: Start with the eight weeks. Andrea Cipriani's 2018 meta-analysis measured outcomes over roughly eight weeks. Most trials test a drug for weeks, and people take it for years.
Mark Delaney: Oh. So, uh, picking a drug that way is like choosing a running shoe after a sprint, and then wearing it for a marathon.
Juniper Vale: That's it. So this new Lancet Psychiatry paper, November 10th, went for the longer stretch.
Mark Delaney: And with Prozac as the yardstick, who came out ahead?
Juniper Vale: Escitalopram, which is Lexapro, duloxetine, which is Cymbalta, and agomelatine, Valdoxan. Each had higher sustained response than fluoxetine.
Mark Delaney: So those three beat Prozac. Okay, but uh, who ran this thing? And how big are we talking?
Juniper Vale: Dr. Paul van Haaren and Dr. Eric Ruhé at Radboud University Medical Center in the Netherlands, with the University of Oxford. They pooled 115 randomized trials. 30,428 adults with major depressive disorder, 21 different antidepressants.
Mark Delaney: Huh. That's a lot of people.
Juniper Vale: It is. Now, Dana sees the numbers. Escitalopram, 1.62. Duloxetine, 1.57. Agomelatine, 1.40. And she thinks, sixty-two percent more people got better on Lexapro.
Mark Delaney: Well, yeah. Isn't that what it says? I mean, one point six two, that's sixty-two percent, you know?
Juniper Vale: No. An odds ratio is not sixty-two percent more people getting better. It compares odds, the chance of responding divided by the chance of not responding. So 1.62 means the odds are about one and a half times higher than on fluoxetine. It doesn't tell you how many extra people out of a hundred.
Mark Delaney: Wait, no, hang on. So it's a real edge, but kinda... the size of it in actual people, we can't read off the number?
Juniper Vale: Right, and that's the piece to hold onto. Now here's what makes it practical. Escitalopram had an odds ratio of 0.60 for all-cause discontinuation, duloxetine 0.64. Below one means fewer people quit.
Mark Delaney: Oh, that's the good part. A pill that looks great on paper doesn't help if you stop taking it in month three, uh, because it makes you feel awful.
Juniper Vale: Exactly why it matters. So Dana's looking at all this, and her real question is simple: is my drug worse?
Mark Delaney: But hold on, is it, though? Worse, I mean. Because uh, how do they even know, if they didn't put Prozac up against all of them directly?
Juniper Vale: Good catch, and mostly they didn't. Only 35 trials, about 9,888 people, were true head-to-head. The rest is a network meta-analysis. Picture lining people up by height, but nobody ever stands next to each other. Everyone only gets measured against one third person. Then you work out who's taller from that.
Mark Delaney: Huh. So, uh, Lexapro stood next to the third guy, Prozac stood next to the third guy, and the math fills in the gap between them?
Juniper Vale: Yes. And every link adds a little wobble. Add up enough wobble and a clean ranking gets fuzzy.
Mark Delaney: Okay, but, I mean, how much do I trust the wobble? Is it like, a little, or, you know, a lot?
Juniper Vale: Depends on the quality of the trials feeding it. More than half were rated high risk of bias. That means design flaws that could distort results, like poor blinding or heavy dropout. So some of the gap between drugs may be study quality, not the drugs.
Mark Delaney: Wait, really? So Prozac is recommended in guidelines almost everywhere, and this giant study can't show it's worse with clean data?
Juniper Vale: Right, and that's the part that matters. Then there's the clock. Only six of the 115 trials followed people past six months. For a paper sold as long-term, that's thin. And agomelatine isn't FDA-approved in the US, so one of the three winners isn't on the table for American listeners. The part coming up is what Dr. Ruhé tells people like Dana, and it flips how you read the headline.
Mark Delaney: So what does the author say Dana should do? Because I read this bit three times. Dr. Ruhé says if you're already stable on fluoxetine, or whatever you're on, uh, don't switch. Based on these results, don't switch.
Juniper Vale: Wait, really? The lead author is telling you not to act on his own headline. I'll be honest, I wasn't expecting that.
Mark Delaney: Ha. Right? So then who's it for?
Juniper Vale: People starting treatment. Ruhé says the findings could eventually influence prescribing for them. Picture someone in a clinic next month, doctor says two options feel about equal. A ranking like this could tip that close call.
Mark Delaney: Okay, so starting, yes. Switching, no. Because, uh, Dana's already got the thing the study is chasing. She's stayed well. For three years, you know?
Juniper Vale: That's the part that matters. And there's one more catch. These rankings describe groups. They can't tell you how any one person responds. Dana could be the person Prozac suits better than Lexapro.
Mark Delaney: So her move is, what, she types a message to her doctor, and leaves the pharmacy app alone?
Juniper Vale: Yes. Ask at her next review. Don't stop or swap anything on her own.
Mark Delaney: Were there any losers in the pile? Besides Prozac, I mean, which I still don't wanna call a loser.
Juniper Vale: Bupropion and reboxetine. Both showed lower sustained response against four other drugs in pairwise comparisons. Same caveats apply, though.
Mark Delaney: And the fix is just... better trials?
Juniper Vale: That's what the authors say, and so does the accompanying Lancet Psychiatry commentary. More direct, long-term head-to-head trials. Drugs actually standing next to each other, for years.
Mark Delaney: So who pays for that? Years-long trials, drug against drug. Uh, nobody gets rich off that, do they? I mean, a company wants its own pill against a placebo, not against the guy next door.
Juniper Vale: That's the open question, and I don't think anyone knows yet. Will anyone fund long, direct head-to-head trials? Until somebody does, the honest uncertainty is the finding. Back to Dana for a second. Tuesday morning, coffee going cool, and she puts the phone face down on the counter. The other thing she needs is a review. Someone sitting down with her once a year, asking how the drug is doing, whether she still needs it, whether she wants to change anything.
Mark Delaney: Huh, and that's kinda the small thing I actually buy. Imperfect evidence still beats none. And a headline isn't a prescription. A check-in with her doctor is, you know, a real thing she can count on. She can sip the coffee, even if it's lukewarm.