Chou Chiahung, McDaniel Cassidi C, Lai Tim C, McDonald Courtney P, Rockwell Devan, Loh Feng-Hua
Department of Health Outcomes Research and Policy Harrison College of Pharmacy, Auburn University Auburn Alabama USA.
Department of Medical Research China Medical University Hospital Taichung City Taiwan.
Health Sci Rep. 2023 Apr 13;6(4):e1193. doi: 10.1002/hsr2.1193. eCollection 2023 Apr.
The Medicare Current Beneficiary Survey (MCBS) limited-access data provides the unique opportunity to utilize administrative claims and adjusted survey data to investigate trends in utilization and medical expenditure across time. The adjusted survey data is a synthesized, matched version of the original survey data and claims. Researchers may choose adjusted survey data or original claims for cost evaluations according to their research purpose. However, limited research has examined methodological issues when estimating medical cost using different MCBS data sources.
The study objective was to examine the reproducibility of individual-level medical cost using both MCBS data sources: adjusted survey and claims data.
This serial cross-sectional study design analyzed 2006-2012 MCBS data. The sample included noninstitutionalized older Medicare beneficiaries (≥65 years old), with a cancer diagnosis and annually enrolled in Medicare Parts A, B, and D. The population was stratified by diabetes diagnosis. The primary outcome was annual medical cost. We investigated the discrepancies of medical cost estimated from the adjusted survey and original claims data. The agreement between cost estimates from the two sources in each year was determined using the Wilcoxon signed-rank test.
A total of 4918 eligible Medicare beneficiaries were included in this study, and 26% of beneficiaries also had diabetes ( = 1275). Significant disagreements in cost estimates between adjusted survey and claims data were present regardless of disease complexity (with or without diabetes). Significant disagreements in medical cost estimates were present in most years, except in 2010 ( = 0.467) and 2011 ( = 0.098), for beneficiaries with cancer and diabetes ( < 0.001 for all). Significant disagreements in medical cost estimates were present in all years for beneficiaries with cancer without diabetes ( < 0.001 for all).
Based on discrepant cost estimates across data sources, researchers using MCBS to estimate costs should be cautious when using claims or adjusted survey data alone.
医疗保险当前受益调查（MCBS）的受限访问数据提供了独特的机会，可利用行政索赔数据和经过调整的调查数据来研究不同时间的医疗服务利用趋势和医疗支出情况。经过调整的调查数据是原始调查数据和索赔数据的综合匹配版本。研究人员可根据研究目的选择经过调整的调查数据或原始索赔数据进行成本评估。然而，在使用不同的MCBS数据源估计医疗成本时，很少有研究探讨方法学问题。
本研究旨在使用MCBS的两种数据源（经过调整的调查数据和索赔数据）检验个体层面医疗成本的可重复性。
本系列横断面研究设计分析了2006 - 2012年的MCBS数据。样本包括未入住机构的老年医疗保险受益人（≥65岁），患有癌症且每年参加医疗保险A、B和D部分。研究人群按糖尿病诊断情况进行分层。主要结局指标为年度医疗成本。我们调查了从经过调整的调查数据和原始索赔数据估计的医疗成本之间的差异。每年使用Wilcoxon符号秩检验确定两种数据源成本估计之间的一致性。
本研究共纳入4918名符合条件的医疗保险受益人，其中26%的受益人也患有糖尿病（n = 1275）。无论疾病复杂性如何（有或无糖尿病），经过调整的调查数据和索赔数据在成本估计方面均存在显著差异。除2010年（P = 0.467）和2011年（P = 0.098）外，大多数年份癌症和糖尿病患者的医疗成本估计存在显著差异（所有P < 0.001）。无糖尿病的癌症患者在所有年份的医疗成本估计均存在显著差异（所有P < 0.001）。
基于不同数据源的成本估计存在差异，使用MCBS估计成本的研究人员在单独使用索赔数据或经过调整的调查数据时应谨慎。
