Structure, Process, and Outcomes in Cancer Surgery

描述（由申请人提供）：癌症手术与高手术死亡率相关，特别是在手术量低的医院。由于对观察到的量-结果关系的机制缺乏了解，这些中心提高质量的努力受到阻碍。本研究将分三步探讨这一问题： 1.确定高容量和低容量医院之间的区别。使用100%的国家样本从1994-2001年医疗保险数据库（n约800，000），我们将比较结构和过程变量在高和低容量的医院与每8个程序。癌症护理的具体结构变量将包括放射和化疗服务的现场可用性，医院参与癌症试验，以及外科医生在癌症手术中的“专业化”。然后，我们将评估与患者选择和其他围手术期护理过程相关的变量。 2.确定哪些因素有助于解释观察到的体积-结果关系。然后，我们将研究8种手术的结构、过程和结局（手术死亡率）之间的关系。使用嵌套模型，我们将评估观察到的体积-结果关系如何随着各种结构和护理过程变量添加到模型中而变化。我们将使用SEER-Medicare链接数据库重复这些分析，以更好地控制病例组合（肿瘤分期）。 3.为研究护理的细节做好准备。为了最好地计划基于医疗记录或现场访问的结构，过程和结果的研究，我们必须首先了解各医院手术死亡率的变化是否可归因于特定的死亡原因。我们将在北方新英格兰的两家医院进行一项试点研究，开发并测试一种基于图表的工具，用于评估死亡原因。

DESCRIPTION (provided by applicant): Cancer surgery is associated with high operative mortality rates, particularly at hospitals with low procedure volumes. Efforts to improve quality at these centers are hindered by lack of understanding about mechanisms responsible for observed volume-outcome relationships. This study will approach this question in 3 steps: 1. Determine what's different between high volume and low volume hospitals. Using 100% national samples from the 1994-2001 Medicare database (n approximately 800,000), we will compare structure and process variables at high and low volume hospitals with each of 8 procedures. Structural variables specific to cancer care will include on-site availability of radiation and chemotherapy services, hospital participation in cancer trials, and surgeon "specialization" in cancer surgery. We will then assess variables related to patient selection and other processes of perioperative care. 2. Identify, which factors help explain observed volume-outcome relationships. We will then examine relationships between structure, process, and outcomes (operative mortality) with the 8 procedures. Using nested models, we will assess how observed volume-outcome relationships change as various structural and process of care variables are added to the model. We will repeat these analyses using the SEER-Medicare linked database to better control for case-mix (tumor stage). 3. Prepare the groundwork for studying the fine details of care. To best plan a study of structure, process, and outcomes based on medical records or site visits, we must first understand whether variation in operative mortality rates across hospitals is attributable to specific causes of death. We will develop and test a chart-based instrument for assessing cause of death in a pilot study at two hospitals in northern New England.