Colorectal Cancer Screening:Evaluating Trends & Outcomes

描述（由申请人提供）拟议的研究开发微观模拟模型，以检查在华盛顿西部的大型HMO团体健康合作组织（GHC）的50多万名成员的限定人群中进行的筛查对观察到的结直肠癌（CRC）发病率和死亡率趋势的影响。 自动化临床数据系统采集了多个医疗保健事件，包括实验室检查（FOBT）和内窥镜检查程序的日期和结果、药房使用情况、门诊医生访视和住院情况，可获得有关该人群的广泛信息。 将使用GHC和Westem华盛顿SEER之间已建立的联系来识别该定义人群中的CRC病例，并使用基于人群的GI病理学登记研究来识别GHC成员中的非恶性结局（例如晚期腺瘤）。 拟议的研究将解决3个主要目标： 目标1：开发一个描述自然史、筛查、诊断和死亡率的结直肠癌微观模拟模型：我们提出的模型将分布函数作为未知参数的输入，而不是固定值。 与以往的方法相比，该模型至少有四个主要优点：1）模型输入中包含了不确定性; 2）模型参数之间的相关性可以通过分布的形式体现出来; 3）模型参数的分布形式可以使敏感性分析自动化，这与单变量敏感性分析相比具有重要的优势; 4）该模型产生输出的分布估计，量化其不确定性。 目标二：在50岁及以上的健康人群中，建立筛查和息肉切除术对结直肠癌发病率和死亡率趋势的影响模型。 我们的模型将使用描述在该人群中筛查传播的数据，使我们能够检查GHC筛查与比率趋势之间的关系。 目标3：模拟HRT、筛查和息肉切除术对50岁及以上女性健康人群CRC发病率趋势的综合影响。 HRT已被证明可以降低结肠直肠癌的发病率，并可能提高死亡率。 我们的第三个目标将使我们能够探索这种风险降低的机制，例如腺瘤性息肉风险的变化，临床前癌症转变的减少以及逗留时间的延长。

DESCRIPTION (provided by applicant)The proposed research develops microsimulation models to examine the effect of screening as practiced on observed trends in colorectal cancer (CRC) incidence and mortality in a defined population of over 500,000 members of Group Health Cooperative (GHC), a large HMO in western Washington. Extensive information about this population is available from automated clinical data systems that capture multiple health care events, including dates and results of laboratory tests (FOBT) and endoscopy procedures, pharmacy utilization, outpatient physician visits and hospitalizations. An established link between GHC and the westem Washington SEER will be used to identify CRC cases in this defined population and a population-based registry of GI pathology will be used to identify non-malignant outcomes (e.g. advanced adenomas) among GHC members. The proposed research will address 3 primary aims: Aim 1: Develop a microsimulation model for colorectal cancer that describes natural history, screening, diagnosis, and mortality: The model we propose takes distributional functions as inputs for unknown parameters, rather than fixed values. The proposed model has at least four key advantages over previous strategies: 1) The model incorporates uncertainty in model inputs; 2) It allows incorporation of correlation between model parameters through their distributional forms; 3) Its use of distributions as inputs essentially automates sensitivity analyses and offers an important advantage over univariate sensitivity analysis; and 4) The model produces distributional estimates for outputs, quantifying their uncertainty. Aim 2: Model the effect of screening and polypectomy on the trends in CRC incidence and mortality within the Group Health Population aged 50 and older. Our model will use data describing dissemination of screening within this population, allowing us to examine the relationship between screening at GHC and trends in rates. Aim 3: Model the combined effect of HRT, screening and polypectomy on trends in CRC incidence within the female Group Health Population aged 50 and older. HRT has been shown to reduce the incidence and perhaps improve mortality from colorectal cancer. Our third aim will allow us to explore mechanisms for this reduction in risk, such as changes in the risks for adenomatous polyps, decreases in their transition to preclinical cancer, and lengthening of sojourn time.