ED Use Patterns Antecedents and Consequences in Older Adults

描述（由申请人提供）：在2006年美国卫生系统紧急护理未来委员会的报告中，IOM直截了当地宣布，急诊科和创伤中心人满为患是全国流行病。用简单的经济学术语来说，供需之间存在着严重的不平衡。不幸的是，很少有基于人群的研究关注个人ED就诊的需求因素，也没有关于ED使用模式随时间变化的需求因素的研究发表。我们建议在全国具有代表性的医疗保险受益人样本（具体目标1）中发展一种或多种反映老年人ED使用模式的类型（而不是专注于单次ED访问），并检查这些ED使用模式的前因（具体目标2）和后果（具体目标3）。我们把重点放在老年人上，因为在2003年，联邦医疗保险支付了1.14亿次急诊科就诊中的16%，这为它提供了通过支付政策进行干预的垄断能力。我们将从四个来源链接数据。第一个是1993年的基线和1995年、1998年、2000年、2002年和2004年对7,447名参加“最年长老人的资产和健康动态研究”（AHEAD）的非机构个人的随访访谈。这些研究对象在1924年之前出生，年龄在70岁或以上。第二个数据来源是国家死亡指数（NDI），其中包含截至2004年12月31日有关AHEAD主题的重要状态信息。第三个数据源是每波数据收集中AHEAD主题的地理代码标识符。最后的数据来源是医疗保险A部分和B部分的索赔，这些索赔将从1989年到2004年的日历年可用。具体目标1的类型将根据CPT相对强度代码和ICD9-CM诊断代码，以及它们与标准有效性措施（如住院率、救护车到达、急诊科接受的所有医疗保健的比例份额和死亡率）的关系来制定。具体目标2将首先使用分层（三个层次：急诊科访问量、受试者、县）广义线性模型评估个人急诊科访问量。虽然将包括一组全面的协变量，但我们对教育、当地卫生保健提供系统的特点和护理的连续性特别感兴趣。然后，我们将重点关注使用类似的两级（人和县）模型对ED使用模式进行建模。在具体目标3中，适当的两水平、多变量统计模型（Cox脆弱回归用于第一个日期事件的时间，如ACSCs住院或死亡率，二项回归用于受限范围计数变量，如住院次数，泊松回归用于倾斜的扩展范围变量，如医生就诊和总费用），将首先估计给定ED使用模式虚拟变量集的粗略影响。然后通过顺序引入标准协变量来分解这些效应。在2006年美国卫生系统紧急护理未来委员会的报告中，IOM直截了当地宣布，急诊科和创伤中心人满为患是全国流行病。不幸的是，很少有基于人群的研究关注个人ED就诊的需求因素，也没有关于ED使用模式随时间变化的需求因素的研究发表。我们建议在全国具有代表性的医疗保险受益人样本（具体目标1）中发展一种或多种反映老年人ED使用模式的类型（而不是专注于单次ED访问），并检查这些ED使用模式的前因（具体目标2）和后果（具体目标3）。

DESCRIPTION (provided by applicant): In its 2006 Report of the Committee on the Future of Emergency Care in the United States Health System, the IOM bluntly declared that there is a national epidemic of overcrowded EDs [Emergency Departments] and trauma centers. In simple economic terms, a critical imbalance exists between supply and demand. Unfortunately, little population-based research exists that focuses on demand factors for individual ED visits, and no studies have been published on demand factors for ED use patterns over time. We propose to develop one or more typologies that reflect an older adults pattern of ED use over time (as opposed to focusing on single ED visits) among a nationally representative sample of Medicare beneficiaries (Specific Aim 1), and to examine the antecedents (Specific Aim 2) and consequences (Specific Aim 3) of these ED use patterns. We focus on older adults because in 2003 Medicare paid for 16% of the 114 million ED visits, providing it with a monopsonistic ability to intervene via payment policy. We will link data from four sources. The first is the 1993 baseline and 1995, 1998, 2000, 2002, and 2004 follow-up interviews with the 7,447 non-institutionalized individuals who participated in the Study of Assets and Health Dynamics Among the Oldest Old (AHEAD). These subjects were born before 1924 and were 70 years old or older at baseline. The second data source is the National Death Index (NDI), which contains vital status information on the AHEAD subjects through December 31, 2004. The third data source is the geocode identifiers for the AHEAD subjects at each wave of data collection. The final data source is the Medicare Part A and B claims for the AHEAD subjects that will be available for calendar years 1989 through 2004. The typologies for Specific Aim 1 will be developed based on CPT relative intensity codes and ICD9-CM diagnostic codes, as well as their relationship to criterion validity measures, such as hospital admission rates, arrival by ambulance, proportional share of all health care received in the ED, and mortality. Specific Aim 2 will be evaluated focusing first on individual ED visits using hierarchical (three levels: ED visits, subject, county) generalized linear modeling. Although a comprehensive set of covariates will be included, we are especially interested in education, the characteristics of the local health care delivery system, and continuity of care. We will then focus on modeling ED use patterns using similar two-level (person and county) models. In Specific Aim 3 the appropriate two-level, multivariable statistical models (Cox frailty regression for the time to the first dated event like hospitalization for ACSCs or mortality, binomial regression for constrained-range count variables like the number of hospital episodes, and Poisson regression for skewed extended-range variables like physician visits and total charges), will first estimate the crude effects of the given set of ED use pattern dummy variables, and then decompose these effects by sequentially introducing standard covariates. In its 2006 Report of the Committee on the Future of Emergency Care in the United States Health System, the IOM bluntly declared that there is a national epidemic of overcrowded EDs [Emergency Departments] and trauma centers. Unfortunately, little population-based research exists that focuses on demand factors for individual ED visits, and no studies have been published on demand factors for ED use patterns over time. We propose to develop one or more typologies that reflect an older adults pattern of ED use over time (as opposed to focusing on single ED visits) among a nationally representative sample of Medicare beneficiaries (Specific Aim 1), and to examine the antecedents (Specific Aim 2) and consequences (Specific Aim 3) of these ED use patterns.