Bynum 1-Year Standard Method for identifying Alzheimer’s Disease and Related Dementias (ADRD) in Medicare Claims data

Here, you will find resources to use the Bynum-Standard 1-Year Algorithm including a README file that accompanies SAS and Stata scripts for the 1-Year Standard Method for identifying Alzheimer’s Disease and Related Dementias (ADRD) in Medicare Claims data. <br> <br>There are seven script files (plus a parameters file for SAS [parm.sas]) for both SAS and Stata. The files are numbered in the order in which they should be run; the five “1” files may be run in any order.<br><br>The full algorithm requires access to a single year of Medicare Claims data for (1) MedPAR, (2) Home Health Agency (HHA) Claims File, (3) Hospice Claims File, (4) Carrier Claims and Line Files, and (5) Hospital Outpatient File (HOF) Claims and Revenue Files. All Medicare Claims files are expected to be in SAS format (.sas7bdat).<br>For each data source, the script will output three files*:<br><br>Diagnosis-level file: Lists individual ADRD diagnoses for each beneficiary for a given visit. This file allows researchers to identify which ICD-9-CM or ICD-10-CM codes are used in the claims data.Service Date-level file: Aggregated from the Diagnosis-level file, this file includes all beneficiaries with an ADRD diagnosis by Service Date (date of a claim with at least one ADRD diagnosis).Beneficiary-level file: Aggregated from the Service Date-level file, this file includes all beneficiaries with at least one* ADRD diagnosis at any point in the year within a specific file<br>* The algorithm combines the Carrier and HOF files at the Service Date-level. The final combined Carrier and HOF Beneficiary-level file includes those with at least two (2) claims that are seven (7) or more days apart.<br>​<br>A final combined file is created by merging all Beneficiary-level files. This file is used to identify beneficiaries with ADRD and can be merged onto other files by the Beneficiary ID (BENE_ID).<br><br><br>With appreciation &amp; acknowledgement to colleagues from a grant funded by the NIA for their involvement in development &amp; validation of the Bynum-Standard 1-Year Algorithm<br>

本数据集包含用于实现Bynum-Standard 1年算法（Bynum-Standard 1-Year Algorithm）的全部资源，其中附带README文件，以及用于在医疗保险索赔数据中识别阿尔茨海默病及相关痴呆（Alzheimer’s Disease and Related Dementias, ADRD）的1年标准方法对应的SAS与Stata脚本。

针对SAS与Stata平台，分别提供7个脚本文件（SAS平台额外附带参数文件parm.sas）。所有文件按执行顺序编号，其中编号以“1”开头的5个文件可按任意顺序运行。

完整算法需使用单年度的五类医疗保险索赔数据：(1) 医疗报销档案（MedPAR）、(2) 家庭健康机构（Home Health Agency, HHA）索赔文件、(3) 临终关怀索赔文件、(4) 经办机构索赔及明细文件、(5) 医院门诊文件（Hospital Outpatient File, HOF）的索赔与收入文件。所有医疗保险索赔文件需为SAS格式（.sas7bdat）。

针对每个数据源，脚本将生成三类输出文件*：
1. 诊断层面文件：列出每位受益人每次就诊对应的ADRD诊断记录，该文件可帮助研究者明确索赔数据中使用的国际疾病分类第9版临床修订版（ICD-9-CM）或国际疾病分类第10版临床修订版（ICD-10-CM）编码。
2. 服务日期层面文件：由诊断层面文件聚合得到，包含所有在某一服务日期（即包含至少一项ADRD诊断的索赔日期）存在ADRD诊断的受益人。
3. 受益人层面文件：由服务日期层面文件聚合得到，包含在对应数据源文件覆盖的年度内任意时间存在至少一项ADRD诊断的所有受益人。

* 本算法会在服务日期层面合并经办机构与HOF文件。最终合并后的经办机构与HOF受益人层面文件，将包含那些拥有至少两项间隔7天及以上的索赔的受益人。

通过合并所有受益人层面文件，可生成最终合并文件。该文件用于识别存在ADRD的受益人，并可通过受益人ID（BENE_ID）与其他文件进行关联。

特此感谢由美国国家衰老研究所（National Institute on Aging, NIA）资助的项目团队成员，他们参与了Bynum-Standard 1年算法的开发与验证工作。