Probabilistic Matching Approach to Link Deidentified Data from a Trauma Registry and a Traumatic Brain Injury Model System Center.
Probabilistic Matching Approach to Link Deidentified Data from a Trauma Registry and a Traumatic Brain Injury Model System Center.
复制标题
用于链接来自创伤登记处和创伤性脑损伤模型系统中心的去识别数据的概率匹配方法。
DOI：
10.1097/phm.0000000000000513
复制
发表时间：
2017
影响因子：
3
通讯作者：
Wagner,AmyKathleen
中科院分区：
文献类型：
--
作者：
Kesinger,MatthewRyan;Kumar,RajGopalan;Ritter,AnneConnelly;Sperry,JasonLee;Wagner,AmyKathleen
没有一个民用创伤性脑损伤数据库可以捕捉到所有护理连续体环境中的患者。这些数据库的链接将产生有价值的洞察可能的护理干预。因此，本文的目的是描述创建一个算法，用于连接创伤性脑损伤模型系统（TBIMS）的创伤数据在国家和国家trauma databases.DesignThe TBIMS数据从一个单一的中心被随机分为两组。一个子集用于生成概率链接算法，以将TBIMS数据链接到中心的创伤登记处。另一个子集用于验证算法。获得病历编号并将其用作唯一标识符来衡量链接的质量。新的方法被用来最大限度地提高阳性predictive value.ResultsThe算法生成子集有121例。其敏感性为88%，阳性预测值为99%。验证子集包括120例患者，敏感性为83%，阳性预测值为99%.ConclusionThe probabilistic linkage algorithm can accurately link TBIMS data across systems of trauma care.未来的研究可以使用这个数据库来回答有意义的研究问题的长期影响的急性创伤复杂的医疗保健利用和恢复整个护理连续创伤性脑损伤populations. EURGROUNDRecord连接是一个强大的工具，在公共卫生领域。通过计算手段，可以将两个大型独立数据集结合起来，以增加数据共享，并提供机会来回答单独使用单个数据集无法回答的研究问题。记录链接的两种形式是（1）确定性链接和（2）概率性链接。至关重要的是，决定使用哪种类型的记录链接是基于两个数据集之间是否存在共同的唯一标识符。在两个数据集之间存在唯一标识符的情况下，如名字和姓氏或社会安全号码，在链接变量上具有精确匹配的主题被定义为匹配。使用这种唯一标识符的精确匹配标准称为确定性记录链接。然而，在没有共同的唯一标识符的情况下，数据集仍有可能通过概率手段进行链接。在这种情况下，可以比较两个数据集中的共同数据元素，以评估两个患者相同的可能性，给定多个变量的相等值。
ObjectiveThere is no civilian traumatic brain injury database that captures patients in all settings of the care continuum. The linkage of such databases would yield valuable insight into possible care interventions. Thus, the objective of this article is to describe the creation of an algorithm used to link the Traumatic Brain Injury Model System (TBIMS) to trauma data in state and national trauma databases.DesignThe TBIMS data from a single center was randomly divided into two sets. One subset was used to generate a probabilistic linking algorithm to link the TBIMS data to the center's trauma registry. The other subset was used to validate the algorithm. Medical record numbers were obtained and used as unique identifiers to measure the quality of the linkage. Novel methods were used to maximize the positive predictive value.ResultsThe algorithm generation subset had 121 patients. It had a sensitivity of 88% and a positive predictive value of 99%. The validation subset consisted of 120 patients and had a sensitivity of 83% and a positive predictive value of 99%.ConclusionsThe probabilistic linkage algorithm can accurately link TBIMS data across systems of trauma care. Future studies can use this database to answer meaningful research questions regarding the long-term impact of the acute trauma complex on health care utilization and recovery across the care continuum in traumatic brain injury populations.BACKGROUNDRecord linkage is a powerful tool in the field of public health. 1, 2 Through computational means, two large independent data sets can be combined to increase data sharing and provide opportunities to answer research questions not possible with a single data set alone. The two forms of record linkage are (1) deterministic and (2) probabilistic linkage. Crucially, the decision on the type of record linkage to use is based on the presence, or absence, of a unique identifier common between the two data sets. In instances where a unique identifier exists between two data sets, like first and last name or social security number, subjects with exact matches on the linking variables are defined as matches. Using this exact matching criterion of a unique identifier is known as deterministic record linkage. However, in instances where a common unique identifier is not available, it is possible that data sets may still be linked through probabilistic means. In this case, common data elements in both data sets can be compared to assess the likelihood that two patients are the same, given equal values on a number of variables.