Health and Health Services Use in the HRS/AHEAD

描述（由申请人提供）：本研究的目的是深入检查7,447名年龄在70岁或以上的HRS/AHEAD受访者的健康和卫生服务使用情况。我们将分析基线（1993年）和1995年、1998年、2000年、2002年和2004年随访访谈的公共使用数据，并结合来自三个受限访问文件的数据：医疗保险索赔、国家死亡指数和地理编码。有了地理编码文件，我们将使用区域资源文件链接州和县一级的数据，并在本地访问级别使用GIS。我们的四个具体目标是：(1)使用自我报告、医疗保险索赔和它们的对应关系来检查全因住院的风险；(2)评估8种特定重大疾病事件及其后遗症的风险；(3)评估可预防或门诊敏感的医院事件的风险；(4)利用调查小组开发的一种非常有前途的类型学，对住院模式的晚期生命历程轨迹进行建模。我们的理论框架包括社会人口特征、社会经济和其他获取因素、健康信念和生活方式、健康状况和卫生服务使用。在这个框架中包含了四个明确的假设。首先，在建立全因住院模型时，我们预计市场结构和实践模式因素将是最显著的预测因素，其次是地方准入措施。其次，我们预计8种常见疾病的流行病学风险因素将是建模这些疾病时最显著的因素，其次是市场结构和实践模式措施，以及人际初级保健的连续性。第三，当对可预防的或对门诊护理敏感的住院进行建模时，我们预计最显著的预测因素将是教育、人际初级保健的连续性以及当地可获得区域的初级保健医疗服务的供应。最后，当对基线后医院使用模式的晚期生命历程轨迹进行建模时，我们预计基线前模式将是最显著的预测因素，其次是市场结构和实践模式、当地获取因素、功能和社会经济地位。估计技术将主要依赖于多层事件历史模型，如二项和多项逻辑回归，以及比例风险模型。还将酌情使用线性和非线性回归以及随机效应模型。

DESCRIPTION (provided by applicant): The objective of this study is an in-depth examination of the health and health services use of the 7,447 HRS/AHEAD respondents aged 70 years old or older at baseline. We will analyze the public use data from the baseline (1993) and the 1995, 1998, 2000, 2002, and 2004 follow-up interviews merged with data from three of the restricted access files: Medicare claims, National Death Index, and geocodes. With the geocode files we will link State and county level data using the Area Resource File, and at the local access level using GIS. Our four specific aims are to: (1) examine the risk of all-cause hospitalization using self reports, using Medicare claims, and their correspondence; (2) assess the risk for eight specific major morbid events and their sequelae; (3) evaluate the risks for preventable or ambulatory care sensitive hospital episodes; and (4) model the late life course trajectories of hospitalization patterns using a very promising typology developed by the investigative team. Our theoretical framework includes sociodemographic characteristics, socioeconomic and other access factors, health beliefs and lifestyle, health status, and health services use. Embedded in this framework are four explicit hypotheses. First, when modeling all-cause hospitalization, we expect that the market structure and practice pattern factors will be the most salient predictors, followed by the local access measures. Second, we expect that the epidemiologic risk factors for each of the 8 common, morbid conditions will be the most salient factors when modeling those conditions, followed by the market structure and practice pattern measures, and the continuity of interpersonal primary care. Third, when modeling preventable or ambulatory care sensitive hospitalizations, we expect the most salient predictors will be education, continuity of interpersonal primary care, and the supply of primary care health care services in the local access area. Finally, when modeling post-baseline late life course trajectories of hospital utilization patterns, we expect that pre-baseline patterns will be the most salient predictors, followed by market structure and practice patterns, local access factors, and functional and socioeconomic status. Estimation techniques will principally rely on multilevel event history models, such as binomial and multinomial logistic regression, and proportional hazards models. Linear and non-linear regression and random effects models will also be used where appropriate.