U.S. flag An official website of the United States government
  1. Home
  2. Drugs
  3. Science and Research | Drugs
  4. Regulatory Science in Action
  5. Improving the Design of Clinical Trials of Drugs to Treat Schizophrenia
  1. Regulatory Science in Action

Improving the Design of Clinical Trials of Drugs to Treat Schizophrenia

Regulatory Science Impact Story


Based on their analyses of schizophrenia drug trials submitted by drug developers, CDER researchers have identified ways in which these trials could be streamlined while not compromising their sensitivity.

Demonstrating the efficacy of candidate drugs to treat schizophrenia can be especially challenging: developers must contend with high patient dropout in clinical trials for this disease, treatment effects are often modest, placebo response has been increasing in recent years, and there is uncertainty as to which endpoints are most sensitive and specific to capture the drug’s effect on the symptoms of this disease. Given these obstacles, it may not be surprising that several large pharmaceutical companies appear to have scaled back their developmental efforts in regard to schizophrenia, despite its toll on patients and their families. With access to individual data from many drug trials, CDER scientists are uniquely positioned to explore ways to make phase 3 trials of schizophrenia drugs more efficient, less costly, and potentially more likely to identify new and effective treatments. Recent results of CDER research (see Reference, below) suggest some promising strategies to achieve these improvements.

Leveraging Submitted Clinical Data to Improve Schizophrenia Drug Trials

Currently, the standard clinical trial to assess the efficacy of a drug to treat schizophrenia lasts for six weeks after the onset of treatment, and patients are interviewed weekly by a trained rater to assess symptom severity with a test called the Positive and Negative Syndrome Scale (PANSS). Thirty individual symptoms (e.g., delusions, grandiosity, blunted affect, and social avoidance) are measured.

A CDER research team led by the Office of Clinical Pharmacology and Office of New Drugs pooled data from 32 trials of approved schizophrenia drugs representing over 14,000 patients to evaluate two strategies for making schizophrenia trials more efficient: 1) shortening the trials and 2) reducing the number of symptoms on which patients were scored by simplifying the PANSS. To determine the feasibility of shortened trials, the researchers examined the concordance of treatment effects at each week of the trial with subsequent treatment outcome and the time course of treatment-related adverse effects. To assess the potential for simplifying the PANSS instrument, they used an approach derived from item response theory to determine which of the 30 response items were informative (i.e., tended to be strongly correlated with the overall PANSS in the pooled subjects) and which were not. Based on their analyses, the researchers derived a streamlined PANSS instrument consisting of 19 response items.

Analysis of Trial Design: An Advantage from Shortening Trials?

For all drugs in the analysis, the success rates at four weeks (with success defined as a statistically significant improvement in the overall PANSS score) were very similar to those at six weeks, suggesting that four-week trials could be as informative as longer trials for determining efficacy. The analysis also showed that a shorter trial would not impede the detection of common adverse reactions associated with schizophrenia drugs. The researchers reasoned further that the shorter trial (i.e., lasting four weeks) could also reduce patient dropout, an important source of bias in drug trials that can make interpretation more difficult, because patients completing a longer trial may differ from non-completers in important respects; this potential advantage would likely offset any minor decrease in statistical power that might arise from the shortened (i.e., four-week) trial.

Simplifying the Instrument Used to Measure Disease Severity

When patients were assessed using the 19-item PANSS, the results were nearly the same as those obtained for patients using the unaltered (i.e., 30-item) PANSS, suggesting that the modified instrument might be adopted and possibly save time and trial costs. The modified measure also decreased overall variability in patient response, presumably because noninformative items had been removed, and this would have the effect of reducing the trial size needed to see a statistically significant effect of a drug.

CDER researchers summarized their conclusions as follows:

  • Combining a shortened PANSS measurement with a four-week trial may be a practical option for designing new trials of drugs to treat schizophrenia. Shorter-duration trials can accelerate drug development in schizophrenia, and the modified PANSS can be a more meaningful clinical outcome assessment tool.
  • The recommended design would shorten patient exposure to longer duration of placebo treatment.

The researchers also pointed out that while conducting a shorter trial and adopting the modified PANSS did not appear to decrease the required sample size based on power calculations, these changes might provide a substantial reduction in sample size and trial cost for some drugs that are expected to yield high treatment outcomes.

How has this work advanced drug development?

CDER research points to elements of trial design that may shorten schizophrenia trials and reduce their costs. An important result of the research is a means for assessing treatment responses in a more efficient manner that would make the benefits of candidate drugs more readily identified. This research, which has the potential to encourage drug development for this devastating disease, underscores the value of retrospective analyses of clinical trial data submitted to FDA to provide insights into how to advance drug development.

In the 32 trials of approved drugs to treat schizophrenia analyzed by CDER researchers, reductions in disease severity and differences in the magnitudes of the three treatments (comparator drug, the drug being evaluated, and placebo) were observed within the first week. The researchers found that the percentage of patients achieving what was deemed to be a clinically significant response (a 20 % reduction in total PANSS score) did not change appreciably from week 4 to week 6.

This project was supported in part by the Oak Ridge Institute for Science and Education (ORISE) Research Participation Program at the Center for Drug Evaluation and Research (CDER).


Younis IR, Gopalakrishnan M, Mathis M, Mehta M, Uppoor R, Zhu H, Farchione T. Association of End Point Definition and Randomized Clinical Trial Duration in Clinical Trials of Schizophrenia Medications. JAMA Psychiatry. 2020 Oct 1;77(10):1064-1071. doi: 10.1001/jamapsychiatry.2020.1596. PMID: 32609294; PMCID: PMC7330825.

Back to Top