BackgroundNational surveillance data should be validated to identify data quality issues. This study tested the validity of healthcare-associated infection (HAI) data in the Korean National Healthcare-associated Infections Surveillance System (KONIS), intensive care unit (ICU) module.MethodsThe validation process consisted of external (EV) and internal (IV) validation phases. For the 10 hospitals that were selected based on the HAI rate, among the 193 participating hospitals between July 2016 and June 2017, both EV and IV were performed. For the EV, the validation team reviewed 295 medical records of 60 patients with reported HAIs, including 20 urinary tract infections (UTIs), 27 bloodstream infections (BSIs), and 13 cases of pneumonia (PNEU), and 235 patients with no reported HAI during 1-day visits conducted in November and December 2017. The reviewer’s diagnosis of HAI was regarded as the reference standard. IV was conducted by the staff of each hospital and evaluated whether UTI or BSI were present. Primary IV was performed for 279 patients who were subject to EV. Secondary IV was performed on 203 patients in another 11 selected participating hospitals that did not report HAIs to KONIS during the 1-year study period.ResultsIn the EV, the diagnosis of UTI in the participating hospitals had a sensitivity of 72.0% and specificity of 99.3%. The sensitivity of BSI and PNEU was 63.2% and 70.6%, respectively, and specificity was 98.8% and 99.6%. The agreement (kappa) between the EV and primary IV was significant, with κ = 0.754 for UTI and κ = 0.674 for BSI. The results of the secondary IV showed that the hospitals that had no reports of HAI had few hospital beds and performed few blood or urine culture tests. In the secondary IV, eight UTIs and three BSIs were newly diagnosed in three hospitals, respectively. The reasons for not reporting the HAIs were presumed to be a lack of understanding of the surveillance standards and fear of the disadvantages of disclosing the HAI.ConclusionThis study shows the need for ongoing validation and continuous training of surveillance personnel to maintain the accuracy of surveillance data. We also confirmed that IV can be used as an alternative monitoring method to examine validity and accuracy.Disclosures All authors: No reported disclosures.