Identifying Safe Stimulant Prescribing Practices to Protect Patients, Inform Key Program Initiatives, and Assist Providers

背景：弗吉尼亚州处方兴奋剂(PS)的使用量一直在飙升。PS对许多人都有好处 但退伍军人也会导致严重的风险，从心脏问题，精神病，躁狂，物质 滥用药物和过量用药。这些风险在退伍军人事务部内外都没有得到很好的量化，而且因为 退伍军人的身心健康状况比率很高，这些风险在 退伍军人。PS将成为退伍军人管理局2022年1月开始的下一项精神药物安全倡议(PDSI)的重点。 意义：PS是有效且有用的药物，但并不是对所有患者都同样安全。这项研究将 使用退伍军人管理局非常庞大的全国数据库来全面量化PS的风险并确定 那些不良反应风险最大的患者组。因此，这项研究将帮助提供商决定谁将 从兴奋剂开始，以及一旦开始如何管理患者。由于即将到来的倡议，我们的研究 将对退伍军人管理局有用，因为它寻求改善PS安全，并可能对改善护理产生快速影响。 创新和影响：这项研究将提供：1)首次全面了解 许多PS风险的频率；2)独特的资源，以支持退伍军人管理局的计划办公室制定质量 改进目标，并寻求改进在PDSI期间和之后的PS处方实践；3)和 有机会证实或驳斥已报道的PS与接受药物治疗的患者死亡风险之间的关联 4)首次对PS治疗过程中的事件进行系统调查，包括毒物筛查和 检测重复的处方，这可能表明不良反应的风险增加；5)图表审查 检查未诊断的物质使用是否可能导致PS风险；6)制定风险评分，以 将我们的发现传达给供应商，并帮助他们与患者讨论PS风险。 具体目标：目标1：确定PS的特殊风险患者；目标2：确定其他描述 在PS开始时或之后与PS风险增加相关的做法；目标3：制定风险评分(RS)以 促进风险感知和协作决策报告：PS。 方法：本研究主要使用退伍军人事务部的大型和详细的临床数据库，由High-Ware分析。 维度倾向得分和边际结构模型。这些方法有助于解决混淆的问题， 这有时会导致从数据库信息得出不准确的结论。其中一些方法还 可以快速实施，以提供广泛的风险调查，允许退伍军人使用特定于退伍军人的数据来 随着PDSI的进行，确定他们的安全工作的优先顺序。关于副本的自动海图提取 有关药物使用的处方和手工表检查将增加这些PS风险的细节 评估。将通过评估已确定的风险因素来构建几种不同结果的风险分数 在目标1和目标2中，采用迭代混合效应Logistic回归随机森林方法识别目标1和目标2 模特们。将选择单个最有用的风险分数，或几个结果的组合的风险分数 使用计划办公室的输入，该自动生成的风险分值的格式将类似于退伍军人管理局现有的 阿片类药物风险评分，以便于使用方案办公室的临床仪表板类似使用风险评分。 下一步：考虑到正在进行的PDSI倡议，这项研究的发现有可能立即提供帮助 退伍军人通过团队与计划办公室的密切关系，后者拥有良好的 与供应商沟通。由于类似阿片类药物工具已经很熟悉，后续工作可能包括退伍军人事务部 通过混合试验实施风险评分，或由于已经使用了类似的阿片类药物工具， 通过在退伍军人管理局广泛采用有关患者安全事件的事前/事后数据收集。其他 后续研究可能会检查PDSI没有如预期那样改变做法的领域，或者 开发新的方法与供应商沟通，或开发新的工具来帮助他们。这项研究将 帮助改善对每年接受处方兴奋剂治疗的数千名退伍军人的护理。

Background: VA use of prescribed stimulants (PS) has been skyrocketing. PS are beneficial for many Veterans but can also cause serious risks ranging from cardiac problems, psychosis, mania, substance abuse, and overdose. These risks have not been well-quantified inside or outside of the VA, and because of Veterans’ high rates of physical and mental health conditions, these risks may be particularly high in Veterans. PS will be the focus of VA’s next Psychotropic Drug Safety Initiative (PDSI) starting January 2022. Significance: PS are potent and useful medications, but are not equally safe for all patients. This study will use the VA’s very large nationwide databases to comprehensively quantify the risks from PS and identify those patient groups at greatest risk of adverse effects. Thus, this study will help providers decide who to start on stimulants, and how to manage patients once started. Because of the upcoming initiative, our study will be useful to the VA as it seeks to improve PS safety and may have rapid impacts on improving care. Innovation and Impact: This study will provide: 1) the first comprehensive look at the magnitude and frequency of many PS risks; 2) a unique resource to support VA’s program offices as they set quality improvement targets and seek to improve PS prescribing practices during and after the PDSI; 3) an opportunity to confirm or refute the reported association between PS and mortality risks in patients receiving PS “off label”; 4) the first systematic investigation about events during PS treatment, including tox screens and the detection of duplicate prescriptions,that may signal increased risks for adverse effects; 5) a chart review examining whether undiagnosed substance use may contribute to PS risks; 6) development of a risk score to communicate our findings to providers, and help facilitate their discussions of PS risks with patients. Specific Aims: Aim 1: Identify Patients at Particular Risk from PS; Aim 2: Identify Other Prescribing Practices at or after PS Initiation Associated with Increased PS Risks; Aim 3: Develop Risk Scores (RS) to Facilitate Risk Perception & Collaborative Decision-Making re: PS. Methodology: This study primarily uses the VA’s large and detailed clinical databases, analyzed by high- dimensional propensity scores and marginal structural models. These methods help address confounding, which can sometimes lead to inaccurate conclusions from database information. Some of these methods also can be implemented rapidly to provide a broad survey of risks, allowing the VA to use Veteran-specific data to prioritize their safety efforts as the PDSI proceeds. Automated chart extraction concerning duplicate prescriptions and a manual chart review concerning substance use will add detail to these PS risk assessments. Risk scores for several different outcomes will be constructed by assessing risk factors identified in Aim 1 & 2 plus those identified by random forest methods by iterative mixed effects logistic regression models. The single most useful risk score, or risk score for a composite of several outcomes, will be chosen with program office input, and this automatically-generated risk score will be formatted similar to VA’s existing opioid risk score to facilitate similar use of the risk score using the program office’s clinical dashboard. Next Steps: Given the ongoing PDSI initiative, this study’s findings have the potential to immediately help Veterans through the team’s close relationship with program offices, who have well-established pathways for communicating with providers. Since a similar opioid tool is already familiar, follow-up work may include VA implementation of the risk score through a hybrid trial or, since a similar opioid tool is already in used, through widespread adoption at VA’s with pre/post data gathering concerning patient safety events. Other follow-up research may examine areas in which the PDSI did not change practices as much as expected, or develop new approaches to communicate with providers, or new tools to assist them. This study will contribute to improving care for thousands of Veterans receiving prescribed stimulants each year.