Health-Economic Evaluation of the Disease Management Program for Chronic Obstructive Pulmonary Disease in Germany

在过去的十年中，德国对慢性阻塞性肺病的管理逐渐发生了变化。疾病管理方案（DMP）已被提出并在COPD患者中实施。这些项目的目的是改变治疗模式，从而改善治疗结果，同时提高稀缺医疗资源的有效利用，甚至节省成本。在国际上，许多研究评估了遵循疾病管理方法治疗COPD的干预措施。然而，这些干预措施的设计是高度异质性的，无法与德国的环境相比较。虽然有许多研究评估了德国的DMP 2型糖尿病，包括四项研究收集了疾病基金的回顾性索赔数据，但几乎没有证据表明德国所有其他DMP，包括DMP COPD。作为德国医疗保健背景下的首个同类研究，本研究的主要目的是比较3年时间内DMP COPD与COPD标准治疗的成本效益。为此，我们将回顾性地评估来自BARMER/GEK的行政索赔数据，BARMER/GEK是一家拥有860万保险的大型德国疾病基金。患者将被纳入研究(a)如果至少有一个住院ICD-10主要诊断为J41的COPD。J42。J43。-或J44。- (b)至少有一种住院主要诊断为J12。-到J18。-和J20。-到J22。-与J41 COPD的二次诊断相结合。J42。J43。-，或J44。-或(c)至少有两次门诊诊断为J41的COPD。J42。J43。-，或J44。-在2010年的180天内然后将患者队列分为干预组（即患者是DMP COPD的成员）和对照组。我们将随访患者3年至2013年，同时使用2008年至2010年的数据进行风险调整。有效性将根据死亡率、（紧急）医院就诊次数和工作能力来衡量。成本将从付款人的角度来衡量。由于使用观察性数据可能导致治疗选择偏倚，因此将采用两步风险调整。首先，倾向评分匹配将减少干预组和对照组之间的异质性，包括社会人口统计学变量、COPD特异性变量和一般合并症测量（Elixhauser-Score）。其次，根据匹配的样本，估计每个结果的回归模型，即死亡率的Cox比例风险模型，（紧急）医院就诊次数和无法工作天数的计数数据模型，以及成本的伽马模型。结果可能证明了结构化护理计划的成功，它可以用来鼓励患者积极参与和登记，并刺激提供者的参与。使用大型数据集，该研究还可能能够确定从DMP COPD中获益最多的患者亚组。

Over the last decade, the management of COPD in Germany has gradually changed. Disease management programs (DMP) have been proposed and implemented in patients with COPD. The aim of these programs was to modify treatment patterns, thereby improving outcomes of care whilst increasing the efficient use of scarce health care resources, or even generating cost savings. Internationally, a number of studies evaluated interventions that follow the disease management approach for the management of COPD. However, the design of these interventions is highly heterogeneous and not comparable to the German setting. While there have been numerous studies evaluating DMP type 2 Diabetes in Germany, including four studies gathering retrospective claims data from sickness funds, little evidence is available for all other German DMPs, including DMP COPD.Being the first study of its kind in the German healthcare context, the primary aim of this study is to compare the cost-effectiveness of DMP COPD with standard care for COPD over a time period of 3 years. To do so, we will retrospectively evaluate administrative claims data from the BARMER/GEK, a large German sickness fund with 8.6 million insured. Patients will be included into the study (a) if with at least one inpatient ICD-10 main diagnosis for COPD of J41.-, J42.-, J43.- or J44.-, (b) if with at least one inpatient main diagnosis of J12.- to J18.- and J20.- to J22.- in combination with a secondary diagnosis for COPD of J41.-, J42.-, J43.-, or J44.-, or (c) if with at least two outpatient diagnoses for COPD of J41.-, J42.-, J43.-, or J44.- within 180 days in the year 2010. The patient cohort will then be divided into an intervention group (i.e., the patient is a member in a DMP COPD) and into a control group. We will follow-up patients for 3 years until 2013 while using data from 2008 to 2010 for risk-adjustment. Effectiveness will be measured in terms of mortality, number of (emergency) hospital visits and ability to work. Costs will be measured from the payer perspective.As the use of observational data may result in treatment selection bias, a two-step risk-adjustment will be applied. First, propensity score matching will reduce heterogeneity between the intervention and control group including socio-demographic variables, COPD specific variables and generic comorbidity measures (Elixhauser-Score). Second, based on the matched sample, regression models for each outcome are estimated, i.e. a Cox proportional hazard model for mortality, count data models for the number of (emergency) hospital visits and for the days incapable to work, and a gamma models for costs.The results might demonstrate the success of structured care programs which can be used to encourage active participation and enrolment by patients and to stimulate the participation of providers. Using a large dataset, the study might also be able to identify patient subgroups that benefit most from DMP COPD.