Compliant or Biased? Comparing Central Line Maintenance Bundle Compliance Data Between Unit-based and Non-unit Based Observers in an Academic Hospital

Abstract

<h3>Background</h3> Appropriate maintenance of central lines (CL) is important to prevent central line-associated bloodstream infections (CLABSI). Daily assessment of CL maintenance with feedback was instituted as a CLABSI prevention effort. The objective of this study was to determine if observer bias influenced CL maintenance bundle observation compliance data. <h3>Methods</h3> We developed and implemented a maintenance bundle based on CLABSI prevention guidelines at a 1266-bed academic hospital. Observers were trained to audit CL for compliance with maintenance bundle components, including: alcohol-impregnated disinfection caps on unused CL ports and infusion tubing, dressings clean and dry, no oozing at insertion site greater than the size of a quarter, dressings occlusive and intact, transparent dressing change recorded within 7-day timeframe, and within 48 hours for gauze dressings. Observations were performed by unit based observers and non-unit based nurse observers house-wide. Bundle compliance between unit based observer and non-unit based observer was compared using Mantel-Haenszel chi-square analysis. <h3>Results</h3> Between 8/1/2018 and 11/1/2019 there were 91,068 CL observations [63,945 (70.2%) by unit-based and 27,123 (29.8%) by non-unit based observers]. Overall, 9,455 (10.4%) CL observations were non-compliant with 1 or more bundle components. 3,019 (3.3%) (14.11%)]. Unit-based observers were less likely to report CL observations that were non-compliant with 1 or more bundle components compared to non-unit based observers [3,808 (6%) vs. 5,647 (20.8%), p < 0.001]. Unit-based observers were less likely to document dressing non-compliance [2,163 (3.4%) vs. 3,586 (13.3%), p < 0.001] or alcohol-impregnated disinfection cap non-compliance [943 (1.6%) vs. 2,076 (8.5%), p < 0.001] compared to non-unit based observers. <h3>Conclusions</h3> Observer bias may impact reported compliance from unit-based observations, compared to trained external observer data. Unit-based observers may be less likely to report non-compliance with bundle components. The use of external observers should be considered to reduce potential bias, and increase confidence in the accuracy of compliance data.