Risk of death among veterans with depression

描述(由申请人提供)： 背景：全球疾病负担研究预测，到2020年，抑郁症将成为全球第二大残疾原因，来自社区研究的大量(但不是明确的)数据表明，抑郁症是导致死亡的关键风险因素。无论是老退伍军人还是新退伍军人，都有很高的心理健康服务需求，退伍军人特别容易受到伤害，其抑郁(约12%的患病率)、贫困、并存疾病和残疾的发生率高于普通美国人。建议的研究基于我们现有的研究，即抑郁症对退伍军人自杀的影响(IIR 04-211)和退伍军人抑郁症治疗的成本效益(CD2 07-206)，以研究退伍军人服务使用者中与抑郁症相关的所有死亡原因。我们的研究团队是唯一合格的，拥有必要的数据资源和专业知识来检查总体死亡率和特定原因的死亡，同时通过其广泛的管理、调查和死因数据来源考虑基线和时变风险因素以及其他潜在影响的社会人口和临床特征。目的：本研究的目的是：1)检测VA患者中与抑郁症相关的全因死亡率和原因特异性死亡率的潜在风险；2)描述可改变的健康行为对抑郁症与全因死亡率、心血管疾病死亡率和癌症死亡率之间关系的影响；3)研究接受抑郁症治疗与死亡率之间的关系。方法：为了实现这些目标，我们将创建一个新的面板数据集，其中包含2003财年起全国所有退伍军人服务用户(N~4,700,000)的健康和社会人口学信息，并进行长达5年的跟踪。我们将对来自现有数据源的合并的二级数据进行分析。退伍军人管理局的国家患者护理数据库(NPCD)将提供详细的住院和门诊临床数据，确定抑郁症和其他疾病的诊断以及包括处方药在内的医疗保健利用情况。将为符合条件的患者链接医疗保险利用率数据。国家死亡指数(NDI)将提供死亡数据，包括死亡原因和日期。2003年患者医疗体验调查(SHEP)将提供行政资源无法提供的社会人口学和健康行为信息。我们将使用生存分析技术，特别是COX回归模型，以及Logistic回归分析，在我们的VA患者队列中检查抑郁对时间和死亡原因的影响。我们将研究抑郁症与死亡的独立关联，以及潜在混杂因素的影响，包括内科和精神科共病、健康行为和社会人口学特征。在使用统计方法控制可能的治疗选择偏差的分析中，我们将检查抑郁症治疗对死亡率的影响。在敏感性分析中，我们将检查抑郁症与死亡时间、原因和相关因素之间的关系的区域差异。通过这项工作，我们不仅能够从有史以来研究过的最大的卫生系统范围的队列中提供关于抑郁症如何与死亡率相关的详细信息，而且我们还将创建一个独特的数据资源，可能会提供给其他退伍军人管理局的研究人员和政策制定者，以确定退伍军人管理局患者群体的死亡原因。

DESCRIPTION (provided by applicant): Background: The Global Burden of Disease study projected that depression will be the second leading cause of disability worldwide by 2020, and there is substantial (but not unequivocal) data from community studies indicating that it is a key risk factor for mortality. Both older veterans and newly returning veterans have high mental health service needs, and VA patients are particularly vulnerable, with higher rates of depression (~12% prevalence), poverty, comorbid medical illness, and disability than the general US population. The proposed study builds on our existing research on the impact of depression on Veteran suicide (IIR 04-211) and cost-effectiveness of VA depression treatment (CD2 07-206) to examine all causes of death associated with depression among VA service users. Our research team is uniquely qualified and has the necessary data resources and expertise to examine overall death rates as well as deaths from specific causes while accounting for baseline and time-varying risk factors and other potentially influential sociodemographic and clinical characteristics through its extensive administrative, survey, and cause of death data sources. Objectives: The objectives of this study are: 1) to examine the potential risks of all-cause and cause-specific mortality associated with depression among VA patients, 2) to characterize the impact of modifiable health behaviors on the relationships between depression and all-cause mortality, and mortality due to cardiovascular disease and cancer, and 3) to examine the associations between receipt of depression treatment and mortality. Methods: To achieve these objectives, we will create a new panel dataset with health and sociodemographic information on all VA service users nationwide (N~4,700,000) from FY03 followed for up to five years. We will conduct analyses with merged secondary data from existing data sources. The VA's National Patient Care Database (NPCD) will provide detailed inpatient and outpatient clinical data identifying depression and other disease diagnoses as well as health care utilization including prescription medications. Medicare utilization data will be linked for eligible patients. The National Death Index (NDI) will provide mortality data, including cause and date of death. The 2003 Survey of Healthcare Experiences of Patients (SHEP) will provide sociodemographic and health behavior information not available from administrative resources. We will use survival analysis techniques, specifically Cox regression models, as well as logistic regression analysis to examine the impact of depression on timing and causes of death in our VA patient cohort. We will examine the independent association of depression with death, as well as the impact of potential confounding factors, including medical and psychiatric comorbidity, health behaviors, and sociodemographic characteristics. In analyses using statistical methods to control for possible treatment selection biases, we will examine the impact of depression treatment on mortality. In sensitivity analyses, we will examine regional variation in the relationship between depression and timing, causes, and correlates of death. Through this work, we will not only be able to provide detailed information from the largest health-system- wide cohort ever studied on how depression is associated with mortality, but we will also be creating a unique data resource that can potentially be made available for other VA researchers and policy makers to identify causes of death in the VA patient population.