Safety-Engineered Device Implementation: Does It Introduce Bias in Percutaneous Injury Reporting?

Authors:
  • JEJanet Eagan
    Janet Eagan
  • KSKent A. Sepkowitz
    Kent A. Sepkowitz
Publisher website
Google Scholar
Abstract
Objective: To examine whether implementation of safety-engineered devices in 2001 had an effect on rates of percutaneous injury (PI) reported by HCWs.Design: Before-and-after intervention trial comparing 3-year preintervention (1998–2001) and 2-year postintervention (2001–2002) periods. PI data from anonymous, self-administered surveys were prospectively entered into CDC NaSH software.Setting: A 427-bed, tertiary-care hospital in Manhattan.Participants: HCWs who attended state-mandated training sessions and completed the survey (1,132 preintervention; 821 postintervention).Intervention: Implementation of a “safer-needle system” composed of various safety-engineered devices for needle-safe TV delivery-insertion, blood collection, and intramuscular-subcutaneous injection.Results: Preintervention, the overall annual rate of PIs self-reported on the survey was 36.5 per 100 respondents, compared with 13.9 per 100 respondents postintervention (P < .01). The annual rate of formally reported PIs decreased from 8.3 to 3.1 per 100 respondents (P < .01). Report rates varied by occupational group (P ≤ .02). The overall rate did not change between study periods (22.7% to 22.3%), although reporting improved among nurses (23.6% to 44.4%, P = .03) and worsened among building services staff (90.5% to 50%, P = .03). HCWs with greater numbers of Pis self-reported on the survey were less likely to formally report injuries (P < .01). The two most common reasons for nonreport (ie, thought injury was low risk or believed patient was low risk for blood-borne disease) did not vary from preintervention to postintervention.Conclusions: Safety-engineered device implementation decreased rates of Pis formally reported and self-reported on the survey. However, this intervention, with concomitant intensive education, had varying effects on reporting behavior by occupation and a minimal effect on overall reporting rates.

This publication has 14 references indexed in Scilit: