Department of Healthcare Policy and Research, Weill Cornell Medical College, New York, NY.
Surgical Outcomes and Quality Improvement Center, Feinberg School of Medicine, Northwestern University, Chicago, IL.
J Arthroplasty. 2019 Jun;34(6):1076-1081.e0. doi: 10.1016/j.arth.2019.01.063. Epub 2019 Feb 2.
Registries and administrative databases have unique and complementary strengths in device epidemiologic studies. We sought to develop, validate, and assess a sequential algorithm using indirect identifiers to link registry and administrative data.
Hip and knee arthroplasty procedures performed at 6 New York State hospitals enrolled in American Joint Replacement Registry in 2014 were included. After conducting a direct linkage using patient identifiers including name and social security numbers, we validated the methodology of indirect linkage using facility ID, patients' year and month of birth, sex, and zip code, and procedure date and site (hip/knee). We further evaluated the influence of absent indirect identifier(s) and compromised data quality on linkage success.
Using our sequential algorithm, 3739 of the 4063 directly linked records (92.03%) were matched with indirect identifiers, with an accuracy of >99.9%. Main reasons for nonmatching included discrepancies in procedure codes and dates. When one of the indirect identifiers was not available, the linkage algorithm still achieved over 90% sensitivity and 99.8% accuracy. Analyses showed that the algorithm was robust when quality of data was moderately compromised.
This study demonstrated high sensitivity and accuracy of an algorithm to create linkages between a registry and an administrative database using indirect identifiers. The methodology will enable long-term surveillance and outcome assessment of a wide variety of devices and procedures. Variations in the coding of procedures, availability of indirect identifiers, and their quality have limited impact on this algorithm.
注册和管理数据库在设备流行病学研究中具有独特且互补的优势。我们旨在开发、验证和评估一种使用间接标识符将注册和管理数据进行链接的顺序算法。
纳入了在纽约州 6 家医院接受 2014 年美国关节置换登记处登记的髋膝关节置换手术。在使用包括姓名和社会安全号码在内的患者标识符进行直接链接后，我们使用设施 ID、患者的出生日期年月、性别和邮政编码以及手术日期和部位（髋/膝）验证了间接链接方法的准确性。我们进一步评估了缺失间接标识符和数据质量受损对链接成功率的影响。
使用我们的顺序算法，在 4063 条直接链接记录中，有 3739 条（92.03%）与间接标识符匹配，准确率>99.9%。不匹配的主要原因包括手术代码和日期的差异。当一个间接标识符不可用时，链接算法仍能达到 90%以上的灵敏度和 99.8%的准确率。分析表明，当数据质量中度受损时，该算法具有稳健性。
本研究证明了一种使用间接标识符在注册和管理数据库之间创建链接的算法具有较高的灵敏度和准确性。该方法将能够对各种设备和手术进行长期监测和结果评估。手术编码的差异、间接标识符的可用性及其质量对该算法的影响有限。
