Berger Ariel, Edelsberg John, Inglese Gary, Bhattacharyya Samir, Oster Gerry
Policy Analysis, Inc., Brookline, Massachusetts 02445, USA.
Clin Ther. 2009 Jun;31(6):1321-34. doi: 10.1016/j.clinthera.2009.06.013.
The aim of this analysis was to assess alternative methods of identification of patients treated with peritoneal dialysis (PD) in health care claims databases for possible use in future analyses of costs of this treatment modality.
Using a US health insurance claims database spanning January 1, 2004, to December 31, 2006, we identified all patients with renal failure who satisfied a case-finding algorithm for PD anticipated to be highly specific, but not necessarily sensitive-namely, > or =2 claims for PD-related physician services (algorithm 1). All claims from these patients were assessed to identify additional PD-related codes, from which 6 additional algorithms were developed, each of which focused on specific categories of billing codes (eg, diagnostic, procedural/service, equipment). Patient selection was then reimplemented using these alternative algorithms. Concordance between the various algorithms and the extent to which resulting samples were similar in terms of patient characteristics, health care resource utilization, and costs were assessed.
We identified a total of 132,274 patients in the database with > or =1 claim for renal failure and valid enrollment data. Among these patients, a total of 2329 satisfied case-selection criteria for algorithm 1, and 4031 patients met criteria for at least 1 of the 7 algorithms for PD. The most sensitive algorithm identified 2859 patients who might have received PD; the least sensitive, 211. Concordance between algorithms was relatively poor. Patients identified using each algorithm were similar, however, with respect to mean age (45-50 years), sex (54%-56% male), and the prevalence of selected comorbidities. Annualized median health care costs were similar across the various algorithms (range, US $80,967-$118,668).
Based on the results from this analysis, it seems that health care providers bill insurers for PD-related care using a variety of codes. Investigators using health insurance claims data for analyses of patients treated with PD need to take this into account.
本分析的目的是评估在医疗保健理赔数据库中识别接受腹膜透析（PD）治疗患者的替代方法，以供未来对这种治疗方式的成本进行分析时使用。
利用一个涵盖2004年1月1日至2006年12月31日的美国医疗保险理赔数据库，我们识别出所有满足一个预计具有高度特异性但不一定敏感的PD病例查找算法的肾衰竭患者，即与PD相关的医生服务索赔≥2次（算法1）。对这些患者的所有索赔进行评估，以识别其他与PD相关的代码，据此开发了另外6种算法，每种算法都侧重于特定类别的计费代码（如诊断、程序/服务、设备）。然后使用这些替代算法重新进行患者选择。评估了各种算法之间的一致性以及所得样本在患者特征、医疗保健资源利用和成本方面的相似程度。
我们在数据库中总共识别出132,274名有≥1次肾衰竭索赔且有有效参保数据的患者。在这些患者中，共有2329名满足算法１的病例选择标准，4031名患者符合7种PD算法中至少1种的标准。最敏感的算法识别出2859名可能接受过PD治疗的患者；最不敏感 的算法识别出211名。算法之间的一致性相对较差。然而，使用每种算法识别出的患者在平均年龄（45 - 50岁）、性别（54% - 56%为男性）以及选定合并症的患病率方面相似。各种算法的年化中位数医疗保健成本相似（范围为80,967美元至118,668美元）。
基于该分析结果，医疗保健提供者似乎使用多种代码向保险公司开具与PD相关护理的账单。使用医疗保险理赔数据分析接受PD治疗患者的研究人员需要考虑到这一点。
