Evaluating the accuracy of an automatic counting system to detect dispensing of hand hygiene product

Abstract

Hand hygiene (HH) is an essential element of infection prevention and control programs. Direct observation of adherence to the 5 moments for HH is considered the gold standard in compliance monitoring. However, as direct observation introduces potential bias, other strategies have been proposed to supplement HH compliance data in healthcare facilities. This study evaluated the accuracy of an automatic counting system (MEZRIT™) to detect when a HH product (soap or alcohol-based hand rub) was dispensed, and thus measure product usage as opposed to compliance with the 5 moments for HH. A quasi-experimental study was conducted in a nursing simulation lab where seven participants undertook basic nursing tasks which included performing HH. Sensors were attached to soap and alcohol-based hand rub dispensers to record the time at which a product was dispensed. HH events were video recorded (time-stamped) and validated against timestamps from the automatic counting system. 260 HH events were detected by the automatic counting system and confirmed by video recordings. 5182 non-HH events were calculated from analysis of the video recordings. The automatic counting system had 90 % sensitivity (95%CI 85.8-93.1 %), and 100 % specificity (95%CI 99.9-100 %). This model generated a positive predictive value of 100 % (95%Cl 98.4-100 %), and a negative predictive value of 99.5 % (95%CI 99.3-99.7 %). The MEZRIT™ system accurately identified 90 % of HH events and excluded 100 % of non-HH events. The real-time monitoring of HH product usage may be beneficial in responding quickly to changes in product usage.