CANCER SURVEILLANCE USING HEALTH CLAIMS BASED DATA

广泛的目标：开发、评估和测试 联邦医疗保险和退伍军人事务部(VA)索赔 创建高质量的国家癌症监测系统的数据 (NCSs)适用于美国老年人口。 具体目标：1)发展--结合来自 12种重要癌症的医疗保险和退伍军人保险；2)评估-- 比较有关发病率和分期的数据的质量 在NCS中使用相同的信息诊断出癌症 SEER方案于1993年报告；和3)利用率--(A) 开发和分析12种癌症的全国发病率 作为各州的具体比率，(B)描述三种治疗模式 癌症部位，以及(3)一种癌症的模型治疗结果 地点。 相关性：目前没有统一的国家系统 确诊时报告癌症发病率和癌症分期。是这样的 这些信息将对规划和 对癌症预防和控制干预措施的评估 作为对治疗的描述和评估。 研究设计：1989年联邦医疗保险和退伍军人管理局数据 到目前为止，将合并以创建事故率和 确定老年医疗保险受益人的癌症阶段。 联合检测的敏感性、特异性和阳性预测值 NCSS数据将使用链接的SEER-Medicare文件确定。 三种癌症的治疗模式将使用 多项概率比特方程组，以及为描述 使用OLS回归和/或对其中一例患者的治疗结果 考克斯比例风险技术。

Broad Objectives: to develop, evaluate, and test the utility of Medicare and Department of Veterans Affairs (VA) claims-based data to create a high quality National Cancer Surveillance System (NCSS) for the elderly U.S. population. Specific Aims: 1) Development--combine administrative data from Medicare and the VA for 12 important cancers; 2) Evaluation-- compare the quality of the data regarding incidence and stage of cancer at diagnosis in the NCSS with the same information reported by the SEER program in 1993; and 3) Utilization--(a) develop and analyze national incident rates for 12 cancers, as well as state specific rates, (b) describe treatment patterns for three cancer sites, and (3) model treatment outcomes for one cancer site. Relevance: Currently there is no uniform national system for reporting cancer incidence and stage of cancer at diagnosis. Such information would be extremely valuable for the planning and evaluation of cancer prevention and control interventions, as well as describing and evaluating treatment. Research Design: Medicare and Va administration data from 1989 to the present will be combined to create incident rates and determine the stage of cancer for elderly Medicare beneficiaries. The sensitivity, specificity and positive predictive values of the NCSS data will be determined using linked SEER-Medicare files. Treatment patterns for three cancers will be described using multinominal probit equations, and a model developed to describe treatment outcomes for one of them using OLS regression and/or Cox proportional hazards techniques.