Wojcik, ExxonMobil Biomedical Sciences, Inc., 1545 US Highway 22 East, Annandale, NJ 08801-3059, USA.
Am J Epidemiol. 2010 Aug 15;172(4):469-77. doi: 10.1093/aje/kwq130. Epub 2010 Jul 19.
This paper describes a multistage process to improve the completeness and time-/cost-effectiveness of ascertaining deaths in large employee cohorts. The process uses the vital status data service of the Social Security Administration (SSA) to identify people who can be confirmed as living, in order to reduce the number of records submitted for a National Death Index (NDI) search. The accuracy of SSA results is verified by submitting a sample of known-living and known-deceased people. For the NDI search results, an algorithm based on a discrete combination of matching variables is applied to distinguish NDI records as true, false, or questionable matches to reduce the number of death certificate requests to state offices. Subsequent decision making is based on manual reviews at various stages. In a cohort of over 200,000 employees, an SSA vital status search reduced the size of the NDI death record search by 85%. The algorithm sorted thousands of NDI records into 15 distinct strata and reduced the number of death certificate requests by 76%. The authors believe that the matching process is enhanced by obtaining paper copies of death certificates from the states, because death certificates often provide additional information and aid in determining true matches to company employees.
本文描述了一个多阶段的过程,以提高在大型员工队列中确定死亡的完整性和时间/成本效益。该过程利用社会保障管理局(SSA)的生命状态数据服务来识别可以确认存活的人,以减少提交国家死亡索引(NDI)搜索的记录数量。SSA 结果的准确性通过提交已知存活和已知死亡的人的样本进行验证。对于 NDI 搜索结果,应用基于匹配变量离散组合的算法来区分 NDI 记录是真实、虚假还是可疑匹配,以减少向州办公室请求死亡证明的数量。随后的决策基于各个阶段的手动审查。在一个超过 200,000 名员工的队列中,SSA 生命状态搜索将 NDI 死亡记录搜索的规模缩小了 85%。该算法将数千条 NDI 记录分类到 15 个不同的层次,并将死亡证明请求的数量减少了 76%。作者认为,通过从各州获取死亡证明的纸质副本,可以增强匹配过程,因为死亡证明通常提供额外的信息,并有助于确定与公司员工的真实匹配。