文献速递流行病学与生物统计医学人工智能与信息学临床试验与 GCP

神经病学与精神病学药物试验的设计、报告与研究重点(2020-2023):基于注册库的元流行病学研究

Design, reporting, and research focus of neurology and psychiatry drug trials (2020-2023): a registry-based meta-epidemiological study using ClinicalTrials.gov and NLP.
PubMed 文献 · Trials · 2026-07-28

该研究基于ClinicalTrials.gov注册的2000-2023年神经病学与精神病学药物干预试验,利用自然语言处理与人工验证提取并统一试验信息。共纳入18609项试验,多数为早期、单中心、小样本(11-50人),60%已完成,36%由企业资助且比例随时间下降,50%完成后报告结果,平均延迟12个月。研究还通过内容聚类探索药物与疾病类别模式,并提供交互式仪表盘。

要点

  • 多数神经精神药物试验为早期、单中心、小样本(11-50人)。
  • 企业资助占比36%且呈下降趋势,大学资助相应增加。
  • 2007年后完成的试验中50%报告结果,平均延迟12个月。

核心结论

  • 尽管注册要求已实施二十年,神经病学与精神病学药物试验在设计及结果报告方面仍存在持续局限。

研究设计与样本

  • 基于ClinicalTrials.gov注册库的观察性研究,纳入2000-2023年神经病学与精神病学药物干预试验。
  • 使用自然语言处理结合人工验证提取并统一试验信息,并进行内容聚类与可视化分析。

主要发现

  • 共纳入18609项试验,多数为早期、单中心、小样本,典型入组11-50人;60%已完成。
  • 企业资助占36%且份额随时间下降,大学赞助相应增加;10%为非随机试验,38%为开放标签。
  • 2007年后完成的试验中50%报告结果,平均延迟12个月。
  • 最常研究的疾病为疼痛(10%)、精神分裂症(8%)、抑郁症(8%)和阿尔茨海默病(7%)。
  • 药物测试分布广泛,非典型抗精神病药、左旋多巴和氯胺酮各占试验的1-2%。

临床怎么用

  • 注册库统一数据可用于实证评估试验实践,为未来试验设计与报告提供参考。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Clinical trial registries document trial design, funding, and results reporting, but heterogeneous and unstructured entries limit cross-trial comparison. Consequently, it is unclear how neurology and psychiatry drug trials are designed and reported in practice, hampering comparison of studies, understanding of the available evidence, and planning of future trials. The aim of this study was to quantify trial design features, results reporting practices, and research focus in neurology and psychiatry drug trials using harmonized registry data. METHODS: This is a registry-based observational study of interventional neurology and psychiatry drug trials registered on ClinicalTrials.gov between 2000 and 2023. Using natural language processing (NLP) with manual validation, we extracted and harmonized trial information on phase, randomization, masking, sample size, funding source, results reporting, and target diseases and drugs. We also performed content-based clustering and visualization to explore patterns across trials by drug and disease category. We analyzed temporal trends and differences across trial characteristics. RESULTS: We included 18,609 trials. Most were early phase, single-center, and small, typically enrolling 11-50 participants; 60% were completed. Industry funded 36% of trials, with a declining share over time and a corresponding increase in university sponsorship. Ten percent of trials were non-randomized and 38% were open label. Among trials completed after 2007, 50% reported results, with a mean delay of 12 months. The most frequently studied conditions were pain (10%), schizophrenia (8%), depression (8%), and Alzheimer's disease (7%). Drug testing was widely distributed, with atypical antipsychotics, levodopa, and ketamine each accounting for 1-2% of trials. Content-based clustering identified disease-specific groupings of trials. Harmonized data are available through an interactive dashboard. CONCLUSIONS: Registered neurology and psychiatry drug trials show persistent limitations in design and results reporting despite two decades of registry requirements. Registry-based analyses using computational harmonization can support empirical evaluation of trial practices and inform future trial design and reporting.

来源:Trials,查看原文。版权归原作者所有。

引用

  1. Doneva SE, Sick B, Ellendorff TR, et al. Design, reporting, and research focus of neurology and psychiatry drug trials (2020-2023): a registry-based meta-epidemiological study using ClinicalTrials.gov and NLP. Trials. 2026. PMID: 42919398. doi:10.1186/s13063-026-09909-8

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。