Discussion

Through the use of a second sign process, we achieved a 40% reduction in the use of plasma mNGS without detected patient harm. This intervention’s success was likely driven by its impact on multiple stages of the process—it served both as a deterrent to initial placement of inappropriate orders and as a hard stop to ensure that any placed order underwent expert review.

On the Continuum of Factors Influencing De-Implementation Process, plasma mNGS is a classic example of a “mixed” intervention.^{Reference Norton, Chambers and Kramer5} The concept behind the technology is promising, especially in specific clinical scenarios such as endocarditis, in which fastidious organisms in the bloodstream routinely cause disease, traditional microbiologic testing is frequently negative, and therapy could be optimized if a causative agent were identified.^{Reference Fowler, Durack and Selton-Suty6} However, data on test performance characteristics are limited, and these data frequently come from single center studies that use change in clinical management as the outcome of interest, which can be confounded by provider interpretation of the test result.^{Reference Niles, Lee, Lamb, Dhaheri and Boguniewicz1} Many hope this technology can replace invasive procedures, and some data suggest it can supplement the sensitivity of traditional microbiologic testing. However, limited reports suggest it does not have adequate sensitivity to replace sampling such as BAL and cannot independently “rule out” infection.^{Reference Bergin, Chemaly and Dadwal2} The unknown specificity also raises questions, as the test is often positive and the clinical significance of these findings can be unclear.^{Reference Niles, Lee, Lamb, Dhaheri and Boguniewicz1,Reference Blauwkamp, Thair and Rosen7}

Following the reduction in use, we successfully sustained our improvement for multiple years. This is likely due to the robustness of the intervention—hard stops are a recognized diagnostic stewardship tool and a higher-level intervention on the Institute for Healthcare Improvement’s effectiveness hierarchy.^{Reference Fabre, Davis and Diekema8}

This work has multiple limitations. Our hospital sees a complex patient population. This, combined with our high utilization at baseline, allowed more reduction opportunities. Our hospital is also fortunate to have a robust ID division, which allowed our services to take on the additional second sign process burden. Recognizing that ID workforce shortages are only likely to worsen,^{Reference Reece and Beckwith9} sites with limited ID faculty may find this intervention harder to implement. Additionally, despite efforts to update billing codes to more appropriately recognize work done by ID providers,^{Reference Naureckas Li, Murtagh, Claussen, Malakooti and Jhaveri10} it is important to acknowledge that this intervention has the potential to increase consult burden without a tangible increase in compensation, which could create tensions if ID providers are performing additional work without direct benefit. We were unable to capture total time spent by ID providers pre- and postintervention, so we cannot measure the degree to which this burden was offset by a reduction in time spent interpreting clinically insignificant results. Unfortunately, data on provider perceptions of this intervention are not available. We also recognize that plasma mNGS cost consequences are far-reaching, and further work is necessary to understand the potential impact of this testing on other costs of care such as antimicrobial use, diagnostic procedures, and length of stay. However, our findings demonstrate that a relatively simple intervention can safely lead to dramatic direct cost savings.