Data errors in the National Hip Fracture Database: a local validation study

• Published in: The bone & joint journal Vol. 98-B; no. 10; pp. 1406 - 1409
• Main Authors: Cundall-Curry, D J, Lawrence, J E, Fountain, D M, Gooding, C R
• Format: Journal Article
• Language: English
• Published: England 01.10.2016
• Subjects: Aged; Aged, 80 and over; Arthroplasty, Replacement, Hip - statistics & numerical data; Databases, Factual; Female; Hip Fractures - epidemiology; Hip Fractures - surgery; Humans; Incidence; Male; Middle Aged; Registries - statistics & numerical data; Reproducibility of Results; Retrospective Studies; ROC Curve; United Kingdom - epidemiology; Validation; Accuracy; Administrative database; National Hip Fracture Database
• ISSN: 2049-4394; 2049-4408
• DOI: 10.1302/0301-620X.98B10.37089

We present an audit comparing our level I major trauma centre's data for a cohort of patients with hip fractures in the National Hip Fracture Database (NHFD) with locally held data on these patients. A total of 2036 records for episodes between July 2009 and June 2014 were reviewed. The demographics of nine patients were recorded incorrectly. The rate of incorrect data in operation codes was most significant with overall accuracy of 0.637 (95% CI 0.615 to 0.658). The sensitivity of NHFD coding ranged from 0.250 to 1.000 and the specificity 0.879 to 0.999. The recording of cementation had a sensitivity of 0.932 and specificity of 0.713. The recording of total hip arthroplasty had a sensitivity of 0.739 and specificity of 0.983. The overall accuracy of mortality data was 0.942 (95% CI 0.931 to 0.952), with sensitivity of 0.967 and specificity of 0.419. This paper highlights the need for local audit of the integrity of data uploaded to the NHFD. Cite this article: Bone Joint J 2016;98-B:1406-9.