On May 7, JAMA Psychiatry published a review of the efficacy of antipsychotics over the short-term, and if the article is carefully parsed, the results reveal that there is no good evidence that antipsychotics provide a clinically meaningful benefit, compared to placebo, over the short-term. This might seem startling to the public and to prescribers, as it is understood to be a given that antipsychotics are effective in curbing acute episodes of psychosis, but in fact it is a finding that can be easily explained, and one that is consistent with an exhaustive 2017 meta-analysis of antipsychotic drug trials.
The authors of the JAMA Psychiatry article conclude that their findings show that antipsychotics are effective over the short-term, which is the same conclusion made by the authors of the 2017 meta-analysis. They do so because both analyses tell of a “statistically significant” drug-placebo difference in the reduction of symptoms on the Positive and Negative Syndrome Scale (PANSS.) What the investigators in each study fail to note is that the drug-placebo difference is quite small (less than 10 points on a 210-point scale), and doesn’t rise to the level of a “minimum clinically important difference.” Indeed, a 9-point drug-placebo difference on the PANSS scale is so small that it most likely wouldn’t be clinically noticeable by either the patient or provider.
In short, here is a summary of the evidence base, compiled over a period of 70 years, for the short-term use of antipsychotics as a treatment for acute episodes of psychosis:
- No good-quality studies have ever been conducted in medication-naïve psychotic patients. Thus, there is no evidence that they are effective for treating patients suffering a first episode of psychosis.
- There are no good-quality studies in patients suffering an “early episode” of schizophrenia that provide evidence that antipsychotics are effective over the short term.
- While there have been hundreds of RCTs in chronic psychotic patients that have found that antipsychotics provide a statistically significant benefit over placebo in terms of reducing symptoms, the drug-placebo difference in these studies does not rise to the level of a “minimum clinically important difference.”
As such, a startling conclusion arises from the two meta-analyses of RCTs of antipsychotic trials: even though antipsychotics have been prescribed to curb acute episodes of psychosis for 70 years, there is no evidence that they provide a meaningful benefit to any group of psychotic patients.
No Trials in Medication-Naïve Patients
When the first antipsychotic, chlorpromazine, was introduced into asylum medicine in 1954, there was no FDA requirement that the manufacturer of a drug had to conduct randomized clinical trials to prove its efficacy. At that time, all a drug manufacturer had to do was run a small trial to assess safety to gain FDA approval. The adoption of antipsychotics in the 1950s was driven by pronouncements from psychiatrists that chlorpromazine was a wonder drug that could reliably knock down psychotic symptoms.
In 1962, the FDA began requiring drug manufacturers to conduct RCTs that assessed the efficacy of a new drug, and since that time hundreds of RCTs of antipsychotics have been conducted. However, researchers who have reviewed this 60-year database of RCTs have not found a single trial that assessed the merits of an antipsychotic in first-episode, medication-naïve patients. Here are two examples of such admissions:
- In a 2017 review, Donald Ross and colleagues concluded that “no placebo-controlled trials have been reported in first-episode patients.”
- That same year, Stefan Leucht and colleagues, in their review of “Sixty Years of Placebo-Controlled Trials in Acute Schizophrenia,” wrote that “there were no studies exclusively examining first-episode patients.”
However, there is one bit of evidence that can be dug up from the research literature that provides a hint of the efficacy of antipsychotics in first-episode patients. In a study of 1,413 first-episode male schizophrenic patients admitted to California hospitals in 1956 and 1957, the California department of mental hygiene found that “drug-treated patients tend to have longer periods of hospitalization . . . furthermore, the hospitals wherein a higher percentage of first-admission schizophrenia patients are treated with these drugs tend to have higher retention rates for the group as a whole.”
That study, while not an RCT, nevertheless tells of how chlorpromazine may have slowed the recovery of first-episode patients, as opposed to helping psychotic patients recover faster.
No Quality Trials in “Early Episode” Schizophrenia Patients
Although researchers today can’t identify a single RCT study of antipsychotics in medication-naïve patients, there were a handful of NIMH studies conducted in the 1970s that sought to assess the merits of antipsychotics in “early episode” schizophrenia patients. In a 2011 Cochrane review, John Bola and colleagues combed through the research literature to find studies of antipsychotics in patients suffering a first or second episode of psychosis, but they could only find five that compared antipsychotic treatment to placebo, psychosocial care or milieu therapy.
The data reporting from these studies was “very poor,” they wrote. The only finding they could extract from the five studies was that those in the placebo group were more likely to leave the study early, while the patients treated with antipsychotics were at increased risk of suffering from the adverse effects of treatment. “Data are too limited to assess the effect of initial antipsychotic medication treatment on outcomes for individuals with an early episode of schizophrenia,” they concluded.
Thus, there is no good-quality RCT evidence that antipsychotics are effective over the short term for treating an “early episode” of schizophrenia.
Short-Term Results in Chronically Ill Patients
In their 2017 review of antipsychotic drug trials, Leucht and colleagues identified 167 RCTs published from 1955 to 2016 that met their inclusion criteria for review. The short-term studies, typically lasting six weeks, involved 28,102 patients, with a mean age of 38.7 years and a mean duration of illness of 13.4 years.
As Leucht and colleagues were unable to find any good-quality studies in first-episode or early-episode schizophrenia, they noted that the outcomes were “representative only for chronically ill, often previously treated patients.”
They found an “effect size” for reduction of symptoms of 0.47, which, they wrote, translates to a drug-placebo difference of 9.6 points on the Positive and Negative Syndrome Scale (PANSS). As will be seen below, this drug-placebo difference falls below a “minimum clinically important difference.”
Even without that minimal clinically important difference reference point, the finding of an effect size of 0.47 does not tell of a treatment that provides a benefit to the majority of patients. Here is a graphic that shows the symptom overlap between two groups with an effect size of 0.47:
There is an 81% overlap in the reduction of symptoms with an effect size of 0.47. As Leucht noted, this produces a “number needed to treat” (NNT) of six, which means that six patients need to be treated to produce one patient who benefits from the treatment. The other five are exposed to the adverse effects of antipsychotics without any additional benefit beyond placebo in symptom reduction, and thus could be said to be harmed by the treatment.
However, for the purposes of this report, on whether the clinical trials of antipsychotics provide evidence of a clinically meaningful benefit, the key statistic is the drug-placebo difference of 9.6 points on the PANSS. While that difference may be statistically significant, it does not rise to the level of a minimum clinically important difference.
No “Minimum Clinically Important Difference”
As a 2012 article in The Journal of Clinical Psychiatry explained, the “concept of the minimum clinically important difference” emerged as a way of assessing the clinical relevance in “standardized instrument scores” used in RCTs. The minimum clinically important difference is defined as “the smallest difference in a score in the domain of interest which patients [or providers] perceive as beneficial and which would mandate, in the absence of troublesome side effects, and excessive cost, a meaningful change in the patient’s management.”
The importance of the minimum clinically important difference, they added, is to determine “if small statistically significant differences in measurement scores in studies with large sample sizes are great enough to be considered clinically meaningful.”
This 2012 paper was authored by researchers at Yale and Columbia University, one of whom was Robert Rosenheck, who is well known for his schizophrenia research. PANSS rates the severity of 30 symptoms on scores of 1 to 7, such that the possible range of scores is 30 to 210. Rosenheck and colleagues determined that the drug-placebo difference on the PANSS must be at least 15 points to be clinically meaningful.
As can be seen, in Leucht’s meta-analysis of 60 years of antipsychotic studies, the drug-placebo difference of 9.6 points on the PANSS fell short of the minimum clinically important difference. Moreover, this was the difference in trials where the placebo group is mostly composed of chronic patients who have been withdrawn from their medication, with such drug withdrawal known to put psychotic patients at increased risk of relapse for the next few months.
JAMA Psychiatry’s 2025 Study
The study published in the May issue of JAMA Psychiatry provides further details about the drug-placebo difference on PANSS in short-term trials of antipsychotics. The findings, when depicted in a graphic, provide a visual understanding of why an 8- or 9-point drug-placebo difference isn’t clinically meaningful. (Leucht was one of the authors of this meta-analysis as well.)
The study was designed to assess if short-term outcomes differed according to whether the patients had not taken an antipsychotic in the previous month, compared to those who had been “recently treated” with an antipsychotic. The researchers found 12 studies that enabled this comparison, with 692 patients in the “not recently treated” group, and 2,089 in the “recently treated group.”
Here are the PANSS scores for the not recently treated group:
As can be seen, the drug-placebo difference on symptom reduction on the PANSS scale was 8.3 points in the “not recently treated” group (27.3 – 19.0).
Here are the PANSS scores for the recently treated group:
The drug-placebo difference for this group was 9.6 points (19 – 9.4). Thus, in neither grouping of patients was the drug-placebo difference near the 15 points required for a minimum clinically important difference.
The graphics below depict these drug-placebo differences, which provide a visual sense of how the outcomes for those treated with antipsychotics failed to separate from the outcomes for the placebo groups. If a treatment is clinically meaningful, then there should be a clear separation from outcomes for the placebo cohort.
Thus, in this 2025 meta-analysis, the data tell of a treatment that failed to provide a minimum clinically important difference, and graphing the results on a 210-point scale visually reveals how the drug treatment failed to notably separate from placebo. Yet, in both cases, the drug-placebo point differences were “statistically significant,” which led Leucht and colleagues to conclude that antipsychotics were effective in both groups of patients.
“This study’s findings confirm antipsychotics as a treatment option for both ‘not recently treated’ and ‘recently treated’ individuals,” they wrote. “This supports current treatment guidelines in schizophrenia, which recommend antipsychotics as a first-line pharmacological treatment option.”
Summing up the Evidence for Short-Term Use of Antipsychotics
With this brief review, it is now easy to sum up the evidence base for use of antipsychotics to curb acute episodes of psychosis:
- There is no evidence supporting their use in medication-naïve patients.
- There is no evidence supporting their use in early-episode schizophrenia.
- Sixty years of clinical studies show that antipsychotics do not provide a “minimum clinically important difference” in chronic patients.
Efficacy in Schizophrenia Drugs Approved Since 2010
As new antipsychotics come to market, they are often touted as providing an additional benefit compared to antipsychotics already on the market. However, Leucht’s 2017 meta- analysis included studies of the atypical antipsychotics that came to market from the early 1990s to 2010, which did not show that these “second-generation” antipsychotics produced a “minimum clinically important difference.” A review of schizophrenia drugs approved since 2010 reveals that that remains the case for the newer drugs as well.
In the table below, the drug-placebo difference in PANSS score for each drug is either from a pooled analysis of pivotal trials, or from a single trial at the most effective dose.
The drug companies that brought these six new schizophrenia drugs to market conducted multiple phase II and phase III trials, and regularly tested multiple doses to find an “optimal dose” that showed greatest efficacy over placebo. Yet, there wasn’t a single study for any of these drugs that showed the antipsychotic provided a clinically meaningful benefit.
RCTs as a Source of Clinical Illusion
Randomized double-blind clinical trials are considered the gold standard for assessing the short-term efficacy of a medical intervention. In the case of antipsychotics, the RCTs are presented to the public and to prescribers as evidence of the efficacy of antipsychotics for treating acute episodes of psychosis, and as a treatment for schizophrenia. That bottom-line conclusion leads to a one-size-fits all clinical practice, with the routine prescribing of antipsychotics for all those with “psychotic” symptoms.
In essence, the RCTs have fostered an illusion of efficacy. It is also easy to see, from a careful parsing of the RCT data, why clinicians perceive antipsychotics as somewhat helpful, which helps reify the belief that their short-term use is evidence-based.
If you look at the change in PANSS scores for those treated with antipsychotics, in the absence of a placebo comparator the diminishment in symptoms over six weeks does rise to the level of clinical importance. For instance, in the meta-analysis published in JAMA Psychiatry in 2025, the PANSS scores for the medicated patients decreased more than the “minimum clinical important” threshold of 16 points, which is a clinically noticeable difference. However, prescribing physicians do not see the drug-placebo difference, because in regular practice they are treating everyone with antipsychotics, and so they are blind to the improvement that occurs in non-medicated patients.
Once this clinical experience is understood, it is easy to see how RCTs can foster harmful clinical practices. The RCTs of antipsychotics have been conducted in a subset of patients (chronic patients), and yet are presented as applicable to all psychotic patients. And while RCTs of antipsychotics may tell of a statistically significant reduction of symptoms, which is cited as evidence of their benefit, the failure of that drug-placebo difference to provide a clinically meaningful benefit is almost never mentioned, either in pronouncements to the public or in the medical literature. As a result, rather than serve as a source of illumination, RCTs in this case have served as a source of darkness, used by psychiatry to present to the public and to prescribers an unfounded story of efficacy.
The failure of the existing research to guide use of the drugs is so all encompassing that, in order for psychiatry to claim that its short-term is “evidence-based,” it needs to start over its testing of this class of drugs. It needs to conduct RCTs in medication-naïve patients and early-episode patients, as well as in chronic patients, and rather than rely on “statistical significance” as a marker for efficacy, rely on the “minimum clinically important difference” as the standard.
What use of antipsychotics to curb acute episodes of psychosis would emerge from robust RCTs of that kind?
One likely answer is that, particularly in first-episode psychosis, or early-episode schizophrenia spectrum disorders, initial use of antipsychotic drugs would be delayed, a practice developed in Northern Finland in the early 1990s as part of its Open Dialogue therapy. This delay in initiation of antipsychotic medication enabled two-thirds of their patients to recover without being exposed to the drugs, and enabled five-year outcomes for Open Dialogue patients that were far superior to outcomes in the U.S. and the rest of the “developed” world. As for their use in chronic patients, the existing literature suggests that their use should not be a first-line therapy, but rather reserved for those who fail to improve with non-drug therapies.
However, in the absence of RCTs that can answer such questions, the public and prescribers need to become aware of the startling conclusion that emerges from meta-analyses of short-term RCTs of antipsychotics, which is that 70 years of RCTs have failed to provide evidence that they provide a clinically meaningful benefit to first-episode patients, early-episode patients, and chronic patients. Once the shock of that conclusion sunk in, then the door to a best-use model could begin to open.