Optimizing Validation of Central-Line-Associated Bloodstream Infections (CLABSIs)— Oregon, 2014

BACKGROUND: Central-line-associated bloodstream infections (CLABSIs) are preventable healthcare-associated infections reportable by intensive care units (ICUs) to the National Healthcare Safety Network (NHSN) since 2009. External validation is essential to ensure data accuracy.

METHODS: The Oregon Health Authority (OHA) validated 2012 Oregon ICU CLABSI NHSN data to identify unreported CLABSIs. Using sampling and validation methods provided in the 2012 CDC CLABSI Validation Toolkit, we reviewed up to 50 charts at facilities reporting a CLABSI standardized infection ratio ≥1. OHA reviewed each chart’s selected blood culture, corresponding infectious event, and complete admission with a standardized abstraction tool. Discrepancies between cases identified by OHA and those reported by hospitals to NHSN were adjudicated by case review with hospitals and CDC NHSN consultation. We interviewed hospital Infection Prevention staff about denominator collection methods. We measured frequency of under-reporting and denominator collection methods, determined sensitivity and specificity of NHSN reporting and ICU CLABSI rates, and compared validated with non-validated NHSN rates.

RESULTS: OHA reviewed 23 (56%) of 41 eligible hospitals: they had a median of 16 ICU beds (range: 4–52); 6 were academic and 3 were critical access. We analyzed 621 medical charts with ≥1 positive blood culture (range: 5–50 per facility) and 633 infectious events. Following adjudication of 27 cases, 12 (23%) of 53 true positive CLABSIs had not been reported to NHSN. NHSN data transfer issues prevented OHA from “seeing” 5 CLABSIs. Patient days were counted electronically by 17 (74%) of hospitals; 14 (61%) received accounting numbers. Nine (39%) counted central-line days electronically; most counted manually because of data-validation difficulties. Compared to complete admission review, NHSN reporting was 77% (95% confidence interval [95%CI]: 64–87%) sensitive and event review was 92% (95%CI: 82–97%) sensitive, consistent with previous studies. ICU CLABSI rates ranged from 0–4.2 per 1,000 central-line days. Compared to 2009 validated rates, 13 (56%) of 23 facilities decreased their rates by ≥45%, 6 (26%) increased, and 4 (2%) remained at zero. The 2012 CLABSI rate in validated facilities was 1.24 per 1,000 central line days; the unvalidated NHSN rate was 0.61.

CONCLUSIONS: Sensitivity of unvalidated NHSN data was fair. We recommend periodic external validation of NHSN data using complete admission review with adjudication to maximize sensitivity. States that validate NHSN data can have infection rates artifactually higher than other states.