Project 3: Suicide Risk Identification in Jails using Data Linkage and Automation

项目 3：使用数据链接和自动化识别监狱中的自杀风险

• 批准号: 10441875
• 负责人: Brian Kenneth Ahmedani
• 金额: $ 27.96万
• 依托单位: MICHIGAN STATE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-08-22 至 2027-07-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10441875
• 关键词: Accreditation; African American population; Algorithms; Area; Automation; Behavior; Books; Caring; Cessation of life; Communities; Community Health Systems; Continuity of Patient Care; County; Coupled; Criminal Justice; Crowding; Data; Data Analyses; Data Linkages; Detection; Disease; Early Intervention; Electronic Health Record; Ethnic Origin; Evidence based intervention; Familiarity; Family; Family member; Future; General Population; Geography; Goals; Health; Health Care Visit; Health Insurance; Health Insurance Portability and Accountability Act; Health Personnel; Health system; Hybrids; Individual; Infrastructure; Institution; Insurance Carriers; Intake; Intercept; Intervention; Jail; Justice; Lead; Length; Link; Machine Learning; Medicaid; Medical Records; Mental Health; Methods; Michigan; Modeling; Population; Process; Provider; ROC Curve; Race; Recording of previous events; Records; Relative Risks; Reporting; Research; Risk; Sampling; Screening procedure; Severities; Standardization; Suicide; Suicide attempt; Suicide prevention; System; Targeted Research; Testing; Thinking; Time; United States Dept. of Health and Human Services; Validation; Work; base; cost; cost effectiveness; data sharing; design; effectiveness implementation study; effectiveness implementation trial; evidence base; experience; future implementation; health care availability; health care service utilization; health data; health record; high risk; hybrid type 2 trial; implementation determinants; implementation evaluation; implementation outcomes; implementation process; implementation strategy; improved; insurance claims; model development; predictive modeling; reducing suicide; risk prediction model; scale up; screening; suicidal morbidity; suicidal risk; suicide model; time use

Abstract Project #3 Although professional and accreditation standards exist to guide identification of suicide risk, few jails effectively screen for such risk at booking (Intercept 2). Given that individuals booking into jails may be less forthcoming in reporting thoughts and behaviors to correctional officers, current identification practices are insufficient. It may be possible to enhance identification methods in jails, replicating a method developed in community health systems. Using a general population sample from seven health systems, the Mental Health Research Network developed a suicide risk model to predict suicide attempts/deaths using electronic health records and insurance claims data. Claims records were used to create the model that resulted in a risk score that could be available for medical personnel, alerting them to the possibility of heightened suicide risk. Replicating this validation, using a jail population with integrated Medicaid claims data, could result in a similar identification process for justice- involved individuals available at jail intake (booking) that could assist in detecting who among those entering jail could be at risk for suicide attempts and suicide deaths. Risk identified through the model will be compared to the practice-as-usual identification within the jail. Because there is no standardized process for identification of suicide risk within jails, each jail’s screening process will be assessed separately. This proposal would leverage three geographically and demographically diverse jails in one state, increasing the generalizability of the findings. Aim 1. Validate the suicide risk model with a jail population sample (three jails; on all of those who enter during a specific length of time), using Medicaid claims and vital record data. Aim 2. Compare the risk flag to the current suicide risk identification process (e.g. practice as usual) within 3 diverse jails. Aim 3. Evaluate implementation factors to inform the design of a future hybrid trial and integration within jails, working with state Medicaid and the Department of Health and Human Services. Improved suicide risk identification in jails could decrease the adverse impacts that suicide has on those who are detained, family members, correctional staff, the institution and community (i.e. liability, costs). Our long-term goal of this research targets jail systems by implementing an automated ‘suicide risk flag’ – derived from prior health records, resulting in improved detection at intake that would lead to intervention to reduce suicide attempts and suicide deaths within the jail. The assembled team has experience with development of the model, familiarity and experience implementing screening tools within jails, and integrating and analyzing jail and Medicaid data. The project leverages an established partnership between the team and criminal justice system. This project will inform an R01 hybrid effectiveness-implementation trial to assess whether the use of a suicide risk flag derived from this algorithm results in access to evidence based intervention within the jail resulting in a reduction in suicide attempts and death within these jail settings and post-release.

摘要项目#3 尽管存在专业和认证标准来指导自杀风险的识别，但很少有监狱有效地 在预订时筛查此类风险(拦截2)。考虑到进入监狱的个人可能不那么愿意 在向惩教人员报告思想和行为方面，目前的辨认做法是不够的。它可能 有可能加强监狱中的识别方法，复制在社区卫生中开发的方法 系统。使用来自七个健康系统的普通人群样本，精神健康研究网络 开发了一个自杀风险模型，使用电子健康记录和保险来预测自杀企图/死亡 索赔数据。索赔记录被用来创建模型，该模型产生了可用的风险分数 对于医务人员，提醒他们自杀风险增加的可能性。复制此验证，使用 拥有完整的医疗补助索赔数据的监狱人口，可能会导致类似的司法鉴定过程- 涉及可在监狱接收(预订)时可用的个人，这可能有助于在进入监狱的人中发现谁 可能有自杀未遂和自杀死亡的风险。通过模型识别的风险将与 监狱内照常进行的身份鉴定。因为没有标准化的程序来鉴定 监狱内的自杀风险，每个监狱的筛查过程将分别进行评估。这项提议将对 在一个州的三个地理和人口结构不同的监狱，增加了调查结果的概括性。 目的1.用监狱人口样本(三个监狱；对所有在此期间进入监狱的人)验证自杀风险模型 特定的时间长度)，使用医疗补助索赔和生命记录数据。目标2.将风险标志与当前 在3个不同的监狱内进行自杀风险识别过程(例如，照常进行)。目标3.评估执行情况 为未来监狱内混合试验和整合的设计提供参考的因素，与州医疗补助和 卫生与公众服务部。监狱自杀风险识别的改进可能会降低 自杀对被拘留的人、家庭成员、惩教人员、机构的不利影响 和社区(即责任、成本)。我们这项研究的长期目标是通过 实施根据以前的健康记录自动设置的自杀风险标志，从而改善 在进食时进行检测，将导致采取干预措施，减少自杀未遂和自杀死亡人数 监狱。组建的团队对模型的开发有经验，熟悉和有经验 在监狱内实施筛查工具，整合和分析监狱和医疗补助数据。该项目 利用小组和刑事司法系统之间已建立的伙伴关系。该项目将为 R01混合有效性-实施试验，以评估自杀风险标志的使用是否源自 算法导致在监狱内获得基于证据的干预，从而减少自杀 在这些监狱环境下和释放后的未遂和死亡。