Novel Quality Measures for Primary Care Management of Attention-Deficit/Hyperactivity Disorder

项目摘要/摘要 注意力缺陷/多动障碍(ADHD)影响着8%-10%的美国儿童。初级保健提供者(PCP) 照顾大多数ADHD儿童，但ADHD治疗质量差距，社会人口差异是 潜在的司机，可能会导致终生疾病和/或不必要的治疗。有迫切的需要 制定ADHD治疗的高质量措施，作为缩小差距和改善健康的先决条件 结果。这项提议的目标是利用机器学习(ML)方法的最新进展- 实现对整个患者群体的电子健康记录(EHR)数据的分析-以开发强大的 ADHD治疗的质量措施，并为质量改进干预措施做准备。这份K23建议书 将加速班尼特博士向独立内科科学家的转变，朝着他的长期目标前进 改善以社区为基础的初级卫生保健，为发育和行为障碍儿童提供服务。他的 多学科导师团队包括Heidi Feldman(ADHD研究导师)、C.Jason Wang(医疗保健 技术和医疗服务联合导师)和Grace Lee(质量改进和实施科学联合 导师)。这个国家公认的内科科学家团队将确保班尼特博士实现他的目标， (1)应用机器学习技术评估护理质量，同时减轻偏见；(2)推进研究 高级统计和定性方法方面的技能，(3)建立质量改进和质量改进方面的专门知识 实施科学的方法，以及(4)提高专业技能和向独立过渡。Dr。 班尼特的临床和研究经验、他的指导团队以及斯坦福大学的环境，使他 以达到提案的目的。根据他在分析电子健康记录数据和试点成功方面的经验 作为一条自然语言处理管道，班尼特博士有以下具体目标：(1)制定指导方针- 基于质量度量，将自由文本的ML分析与结构化EHR数据相结合来评估PCP 4-11岁儿童ADHD的治疗，(2)评估PCP对循证指南的依从性 对于ADHD治疗，并检测护理方面的差异并将ML模型中的相关偏差降至最低，(3)确定优先顺序 旨在改善ADHD护理并缓解家庭和家庭差异的质量改进干预措施 临床医生利益相关者认为可行、可接受和重要。与NIMH的战略计划相一致，这 提案将(1)加强利益攸关方之间的合作，不断改进循证证据 初级保健环境中的做法，(2)确定规划的初级保健方案和系统级质量的目标并确定其优先顺序 旨在标准化ADHD护理和缓解差异的改进干预措施，以及(3)应用新的 提供高质量ADHD护理的实时反馈和持续监控的技术。带着未来 R01资助，班尼特博士将在全国儿科网络中交叉验证开发的质量衡量标准 医疗保健系统，同时实施数据驱动的质量改进干预措施。

PROJECT SUMMARY / ABSTRACT Attention-Deficit/Hyperactivity Disorder (ADHD) affects 8-10% of US children. Primary care providers (PCPs) care for most children with ADHD but quality gaps in ADHD treatment, with sociodemographic disparities as a potential driver, may lead to life-long morbidity and/or unnecessary treatments. There is an urgent need to develop quality measures for ADHD treatment, as a prerequisite for mitigating disparities and improving health outcomes. The objective of this proposal is to leverage recent advances in machine learning (ML) methods – enabling the analysis of electronic health record (EHR) data of an entire patient population – to develop robust quality measures for ADHD treatment, and to prepare for quality improvement interventions. This K23 proposal will accelerate Dr. Bannett’s transition into an independent physician scientist, towards his long-term goal to improve community-based primary health care for children with developmental and behavioral disorders. His multidisciplinary team of mentors include Heidi Feldman (ADHD research mentor), C. Jason Wang (health care technology & health services co-mentor), and Grace Lee (quality improvement & implementation science co- mentor). This nationally recognized team of physician scientists will assure Dr. Bannett achieves his goals, to (1) apply machine learning techniques to assess quality of care while mitigating bias, (2) advance research skills in advanced statistics and in qualitative methods, (3) build expertise in quality improvement and implementation science methods, and (4) enhance professional skills and transition to independence. Dr. Bannett’s clinical and research experiences, his mentoring team, and the environment at Stanford, position him to achieve the proposal’s aims. Building upon his experiences in analyzing EHR data and successes in piloting a natural language processing pipeline, Dr. Bannett has the following specific aims: (1) to develop guideline- based quality measures that combine ML analysis of free text with structured EHR data to assess PCP treatment of children aged 4-11 years with ADHD, (2) to assess PCP adherence to evidence-based guidelines for ADHD treatment and to detect disparities in care and minimize related bias in ML models, (3) to prioritize quality improvement interventions aimed at improving ADHD care and mitigating disparities that family and clinician stakeholders consider feasible, acceptable, and important. Aligned with the NIMH’s strategic plan, this proposal will (1) strengthen collaboration between stakeholders to continuously improve evidence-based practices in primary care settings, (2) identify and prioritize targets for planned PCP- and systems-level quality improvement interventions aimed at standardizing ADHD care and mitigating disparities, and (3) apply novel technologies that provide real-time feedback and continuous monitoring of high-quality ADHD care. With future R01 funding, Dr. Bannett will cross-validate developed quality measures in a national network of pediatric healthcare systems, and, in parallel, implement data-driven quality improvement interventions.