The longest trial behind each drug’s approval

The longest trial behind each approval

Weeks, from each drug’s approval package as quoted on its journey page [1]. The right column is people filling at least one US prescription in 2024, from the federal survey file [5]; seven drugs show no count because AHRQ masks them or too few respondents reported them. (Until 2026-09-29 this column came from a third-party site and showed a number for every drug.) Sorted shortest trial first.

• aripiprazole (Abilify)6 weeks · approved 2002 · 1,780,154 patients
• lisdexamfetamine (Vyvanse)4 weeks · approved 2007 · 1,619,998 patients
• esketamine (Spravato)4 weeks · approved 2019 · clinic-administered; no survey count

  The 2019 approval rested on one positive 4-week trial (TRANSFORM-2) plus SUSTAIN-1, an event-driven relapse-prevention study with no fixed horizon; two of the three short-term trials missed their primary endpoints (FDA summary review)
• zolpidem (Ambien)5 weeks · approved 1992 · 2,420,368 patients
• lurasidone (Latuda)6 weeks · approved 2010 · masked in the federal survey file; no count
• bupropion (Wellbutrin SR)6 weeks · approved 1996 · 8,042,018 patients

  1996 Is the Wellbutrin SR approval (NDA 020358, 10/04/1996); the original immediate-release Wellbutrin was approved 12/30/1985 and marketed from 1989 after a seizure signal in a bulimia trial
• citalopram (Celexa)6 weeks · approved 1998 · 2,876,646 patients
• olanzapine (Zyprexa)6 weeks · approved 1996 · too few survey respondents to count
• quetiapine (Seroquel)6 weeks · approved 1997 · 1,849,066 patients
• methylphenidate (Concerta)7 weeks · approved 2000 · 2,985,530 patients

  2000 Is the Concerta approval (NDA 021121, 08/01/2000); the original Ritalin approval was 12/05/1955, before the 1962 law that required efficacy trials at all
• escitalopram (Lexapro)8 weeks · approved 2002 · 9,460,411 patients
• vortioxetine (Trintellix)8 weeks · approved 2013 · masked in the federal survey file; no count
• desvenlafaxine (Pristiq)8 weeks · approved 2008 · too few survey respondents to count
• lithium8 weeks · approved 1970 · too few survey respondents to count
• risperidone (Risperdal)8 weeks · approved 1993 · too few survey respondents to count
• clonazepam (Klonopin)9 weeks · approved 1975 · 2,154,594 patients

  The 9-week dose-response trial belongs to the 1990s panic-disorder indication; the original 1975 approval (NDA 017533, 06/04/1975) was for seizure disorders
• modafinil (Provigil)9 weeks · approved 1998 · masked in the federal survey file; no count
• duloxetine (Cymbalta)9 weeks · approved 2004 · 4,071,231 patients
• alprazolam (Xanax)10 weeks · approved 1981 · 3,426,968 patients
• venlafaxine (Effexor XR)12 weeks · approved 1997 · 3,429,959 patients

  1997 Is the Effexor XR approval (NDA 020699, 10/20/1997); immediate-release Effexor was approved 12/28/1993
• mirtazapine (Remeron)40 weeks · approved 1996 · 1,458,927 patients
• paroxetine (Paxil)52 weeks · approved 1992 · 1,681,958 patients
• lamotrigine (Lamictal)78 weeks · approved 2003 · 2,474,525 patients

Where we cannot show you a number

Thirteen drugs have no single stated pivotal-trial duration in our record. A cleaner page would drop these rows; an honest one lists them with the reason, because absence here is a fact about the public record, not a finding about the drug.

• sertraline (Zoloft) — our record carries no stated trial duration for the 1991 approval; the journey page explains the gap rather than guessing across it.
• fluoxetine (Prozac) — same gap for the 1987 approval — no duration we can quote, so none appears.
• gabapentin (Neurontin) — approved for seizures in 1993; its psychiatric use is off-label, so there is no psychiatric approval trial to time.
• trazodone — a 1981 approval with a thin public record — the absence is in the record, not in our reading of it.
• buspirone (BuSpar) — the current label format carries no study durations for the 1986 approval.
• hydroxyzine (Vistaril) — approved in 1956, before efficacy trials were required at all — the oldest drug on this site predates the rules.
• lorazepam (Ativan) — a 1977 label that never printed its trials — and that states effectiveness beyond four months was never systematically assessed.
• diazepam (Valium) — approved 1963, once the most prescribed drug in America; the trials were never printed on the label.
• temazepam (Restoril) — a 1981 approval with no printed trials and a 7-to-10-day indication attached to a refilled-for-years drug.
• buprenorphine (Suboxone film) — the film’s label carries no clinical-studies section at all — the 2010 approval was bridged pharmacokinetically to the earlier tablets.
• atomoxetine (Strattera) — real trials on the label (6–10 weeks across ages, plus maintenance) but no single longest figure stated cleanly enough to chart.
• guanfacine (Intuniv) — five monotherapy trials plus maintenance, ages 6–17 only — a range, not a single number, and adult use rides on extrapolation.
• ziprasidone (Geodon) — acute trials of 4–6 weeks plus a maintenance trial — and a label that plainly says one pivotal trial failed to beat placebo.

What this does and does not mean

Read the table in both directions. Short pivotal trials are not a scandal: a placebo-controlled trial of a few weeks is how the FDA establishes that a drug works at all, and it is the same standard for nearly every medicine you have ever taken. Some of these drugs also ran long trials — lamotrigine’s 78 weeks is a maintenance design testing relapse prevention, which is exactly the question long-term use raises.

What a six-week trial cannot tell you is what year six looks like: long-term side-effect rates, what stopping is like, whether the benefit persists. That evidence accumulates after approval, in studies with a patchier and less mandatory record — which is why each drug’s page on this site carries its post-approval research separately, and why the stopping evidence gets its own page. If you are weighing a prescription, the question this table hands you for your prescriber is short and fair: “the approval trials ran {n} weeks — what do we know about people who take it as long as I might?”

Common questions

How long are antidepressant clinical trials?

The pivotal trials behind FDA approvals typically run six to twelve weeks. A 2025 systematic analysis in The American Journal of Medicine put the median antidepressant trial at 8 weeks — against a median real-world duration of use of roughly five years.

Does a short trial mean the drug is unsafe?

No. Short placebo-controlled trials are the standard the FDA uses to establish that a drug works at all, and some drugs on this page also ran long maintenance trials — lamotrigine’s 78 weeks is one. What a short trial cannot tell you is what ten years on the drug does; that evidence comes from post-approval studies, which is a different and patchier record.

How many people take these drugs long term?

In CDC survey data from 2011–2014 covering ages 12 and over, about one in four people who took an antidepressant in the past month had been taking one for ten years or more. Overall, 13.2% of US adults used an antidepressant in the past 30 days in 2015–2018.

Why is there no number for sertraline or fluoxetine?

Because our label-sourced record carries no stated pivotal-trial duration for them, and we do not guess. Each drug’s journey page explains the specific gap. Absence of a number in our table is a fact about the record, not about the drug.

Where do the patient counts come from?

From the federal Medical Expenditure Panel Survey itself: we computed them from AHRQ’s public-use 2024 Prescribed Medicines file (HC-254A, released July 2026). Each is the weighted number of people with at least one purchase of the drug in the United States in 2024. Where AHRQ masks a drug in the public file, or fewer than 60 survey respondents reported it (AHRQ’s own threshold for publishing an estimate), we show no number and say which. An earlier version of this page took these counts from ClinCalc, a third-party site.

Sources

Last verified 2026-09-29. Corrections change this date.