Personalized Provider Selection to Reduce Surgical Disparities

结直肠癌（CRC）是2019年老年人死亡的第二大原因，诊断为145，600例 病人，并负责51，020死亡。在没有转移性疾病的情况下，手术是标准的 为90%以上的CRC患者提供治疗。从现有的文献和我们的初步研究的见解表明， CRC中最重要的手术差异与手术风险的种族效应和强大的医院有关- 死亡率和发病率的相关差异。CRC手术结局存在显著差异， 医院（例如死亡率0.6%-14.7%），已知的差异对黑人病人产生不利影响。黑色 患者的手术利用率较低，手术结局较差，与 白色病人。黑人患者更有可能使用低质量，低容量的医院进行手术，即使 更高质量的选择可以在离家更近的地方找到。这些差距不仅限于种族，还包括居住环境 (e.g.农村）和其他患者特征。获得更高质量的医院是实现这一目标的关键障碍。 在人群中的手术公平性。推动医院选择的数据有限。我们的初步研究 这表明，大多数黑人患者（86%）有一个更高质量的医院位于他们的附近， 通过数据驱动的转诊，有可能将差距减少30%以上，同时改善 人口。现有的风险分层工具，以协助在医院选择过程中缺乏必要的 促进合理决策的因素组合，包括：1）疾病特异性，2）注意 复杂的医患互动，3）医院质量信息，4）比较统计。我们 初步数据表明，可以进行符合这些标准的准确风险预测。拟议 研究，我们将完善个性化的预测模型，将其扩展到国家一级，并开发工具， 使统计比较成为可能。由于差距不再仅仅是弱势群体的问题， 使用反事实模拟展示数据驱动的转介在社会福利方面的收益。我们还 将使用情景测试来模拟数据驱动的转诊对转诊提供者意愿的影响 为了提高质量而牺牲便利性和声誉。这些信息对于推动政策改革至关重要， 推进外科公平。我们的目标是通过将年龄较大的黑人CRC患者转介到更高质量的治疗来减少差异 通过1）开发个性化的风险模型来区分医院（或外科医生），2）提供 为旨在激励数据驱动转诊的政策提供依据，以及3）制定促进数据的战略 推动CRC转诊。这项开创性的工作将提供1）新的风险分层方法，2）估计 在制定新政策时，决策者可以从数据驱动的转介中获得社会福利效益， 手术差异和3）关于医生偏好的新知识，以告知干预措施，以促进采用 数据驱动的转介。这项工作将作为一个模板，为以后的努力，以扩大数据驱动的转介 所有手术治疗的实体器官恶性肿瘤。

Colorectal cancer (CRC), the second leading cause of death in older adults in 2019, was diagnosed in 145,600 patients and was responsible for 51,020 deaths. In the absence of metastatic disease, surgery is the standard of care for more than 90% of CRC patients. Insight from existing literature and our preliminary studies suggest that the most essential surgical disparities in CRC are related to race effects in surgical risk and strong hospital- associated differences in mortality and morbidity. Significant variation in CRC surgical outcomes exists across hospitals (e.g. mortality rates 0.6%-14.7%) with known disparities adversely affecting black patients. Black patients have lower surgical utilization rates, worse surgical outcomes, and lower survival rates compared to White patients. Black patients are more likely to use lower quality, lower volume hospitals for surgery, even when a higher quality choice can be found closer to home. These disparities extend beyond race to residential setting (e.g. rural) and other patient characteristics. Access to higher quality hospitals is a critical barrier to achieving surgical equity across the population. Data to drive hospital selection is limited. Our preliminary studies demonstrate that most Black patients (86%) have a higher quality hospital located within close proximity of their home and the potential to reduce disparities by >30% with data driven referrals while improving outcomes across populations. Existing risk stratification tools to assist in the hospital selection process lack the requisite combination of factors to facilitate rational decision-making including: 1) disease specificity, 2) attention to complex patient-provider interactions, 3) information on hospital quality, and 4) comparative statistics. Our preliminary data suggest that accurate risk prediction can be performed that meet these criteria. In the proposed study, we will refine the personalized prediction models, scale them to the national level, and develop the tools to make statistical comparisons possible. As disparities are no longer a problem for the vulnerable alone, we demonstrate the gains in Societal Welfare of data driven referrals using counterfactual simulation. Further, we will use scenario testing to simulate the effects of data driven referrals on the willingness of referring providers to trade-off convenience and reputation for enhanced quality. This information is critical to drive policy reform to advance surgical equity. Our goal is to reduce disparities by referring older, black CRC patients to higher quality hospitals by 1) developing personalized risk models to differentiate across hospitals (or surgeons), 2) providing evidence to inform policies designed to incentivize data driven referrals, and 3) setting strategies to promote data driven referrals for CRC. This pioneering work will provide 1) new methods of risk stratification, 2) an estimate of the Societal Welfare benefits of data driven referrals for policy makers when designing new policies to minimize surgical disparities and 3) new knowledge on physician preferences to inform interventions to promote adoption of data driven referrals. This work will serve as a template for subsequent efforts to extend data driven referrals across all surgically treated solid organ malignancies.