Health Utility Estimates and Co-Occurring Conditions

描述（由申请方提供）：多种健康状况的存在使成本效益分析中使用的特定状况的健康相关生活质量（HRQoL）影响估计复杂化。该项目的总体目标是找到最佳方法来估计多种情况下单一健康状况的影响。之前的研究重点是对汇总健康评分进行建模;我们的研究将尝试通过对用于计算这些评分的各个问题进行建模来改进汇总健康评分估计。具体而言，我们将（1）开发和验证HRQoL测量中共现条件的问题级模型，（2）根据最佳汇总评分模型评估此问题级模型，（3）构建117种慢性疾病的HRQoL影响的全国代表性目录，以及（4）演示使用该目录从新样本中删除共现条件的影响。我们将使用一个大型的医疗保险数据库来开发，验证和评估我们的模型。我们将使用来自医疗支出小组调查（一项大型全国代表性调查）的18岁及以上的受访者，对117种慢性病的影响进行分类。这两个数据集包括一些AHRQ定义的优先人群，如妇女，种族和少数民族，低收入，农村和内城个人。我们将使用基于临床的心力衰竭患者样本来证明我们的方法在较小样本中的使用。我们将使用比例优势模型预测问题的回答，该模型使用年龄，性别和健康状况作为预测因素。拟议的研究将提供有价值的健康测量方法，为卫生政策提供信息。

DESCRIPTION (provided by the applicant): The presence of multiple health conditions complicates Health Related Quality of Life (HRQoL) impact estimates for specific conditions used in cost-effectiveness analyses. The overall goal of this project is to find the best method to estimate the impact of a single health condition when multiple conditions are present. Previous research focused on modeling summary health scores; our research will attempt to improve summary health score estimates by modeling the individual questions which are used to calculate these scores. Specifically, we will (1) develop and validate a question-level model of co-occurring conditions in HRQoL measures, (2) evaluate this question-level model against the best summary score model (3) construct a nationally representative catalog of HRQoL impact for 117 chronic conditions, and (4) demonstrate the use of the catalog to remove the effect of co-occurring conditions from a new sample. We will use a large Medicare database to develop, validate, and evaluate our model. We will use respondents aged 18 and over from the Medical Expenditures Panel Survey, a large nationally representative survey, to catalog the impact of 117 chronic conditions. These 2 datasets include a number of AHRQ-defined priority populations such as women, racial and ethnic minorities, low income, rural, and inner city individuals. We will use a clinic-based sample of heart failure patients to demonstrate use of our methods in smaller samples. We will predict question responses with a proportional odds model that uses age, sex, and health conditions as predictors. The proposed research will provide valuable methods for health measurement which informs health policy.