ABSTRACT
Background: The national spread of safety interventions has been slow and difficult. While it is widely known that hospital contextual features and implementation factors impact spread of evidence-based interventions, there is little prospective research on modifiable factors that impact implementation at multiple sites. Nephrotoxic Injury Negated by Just-in-Time Action (NINJA) is a system-level patient safety intervention that led to a sustained reduction in nephrotoxic medication–associated acute kidney injury (NTMx-AKI) at one hospital. Our objective was to prospectively characterize the association between context and implementation factors and reduction of NTMx-AKI at nine hospitals implementing NINJA, using mixed methods. Methods: Grounded in i-PARIHS, we used qualitative comparative analysis (QCA) to assess the association between context and implementation factors, measured quarterly by survey, and reduction of NTMx-AKI, measured using statistical process control and ARIMA modeling. Interviews provided an understanding of causal processes underlying associations identified in QCA. Qualitative and quantitative data were collected and analyzed concurrently and then merged. Results: Five hospitals reduced AKI, four did not. Overall, the collaborative reduced NTMx- AKI by 8 cases per 1000 patient-days per month (95% CI: 14.6-1.4; p=0.018). QCA analysis revealed that hospitals needed to have a baseline AKI rate > 1.0 to reduce NTMx-AKI (Ncon 1.0, Ncov 0.83). In addition, hospitals that reduced NTMx-AKI had either (a) a pharmacist champion and > 2 pharmacists working on NINJA (Scon 1.0, Scov 0.8) or (b) No other organizational priorities causing implementation delays (Scon 1.0, Scov 0.2). Involving quality improvement coordinators or data analysts did not influence success. Qualitative interviews supported these findings and underscored the importance of how the NINJA implementation team integrated with frontline staff. Conclusions: We identified two different pathways to successful reduction in NTMx-AKI when implementing NINJA. These findings have implications for the future spread of NINJA and suggest an approach to study spread and scale of safety interventions more broadly.