Background: A substantial proportion of peripherally inserted central catheters (PICCs) are placed for inappropriate indications. We performed a multi-hospital intervention to improve the appropriateness of PICC use.

Methods: We conducted a collaborative cohort study of hospitals participating in the Hospital Medicine Safety Consortium. An evidence-based intervention centered on the Michigan Appropriateness Guide for Intravenous Catheters (MAGIC) was implemented in all sites. Following MAGIC recommendations, sites focused on reducing short term PICC use (< 5 days), increasing use of single lumen devices, and avoiding PICC placement in patients with stage IIIb or greater (eGFR<45ml/min) chronic kidney disease (CKD). As part of a “pay for performance” program, participating hospitals were provided with resources including educational materials, sample order-sets, and expert support to facilitate implementation. Multilevel mixed effects Poisson regression was used to assess the impact of MAGIC by comparing rates of inappropriate PICC use before vs. after launch of the intervention.

Results: Between April 2015 to August 2018, data from 31,783 PICCs placed in 29,628 patients in 42 hospitals was available for analysis. Median duration of PICC use at each hospital ranged from 10 to 61 days. The baseline rate of inappropriate PICC use prior to the intervention (composite of <5 days, multi-lumen use, and/or PICC use in CKD) was 69.8% (13007/18648). Following the intervention, inappropriate PICC use decreased to 57.6% (7565/13135). Contemporaneously, appropriateness of PICC use increased from 30.2% to 42.4% (p<0.001). The greatest metric to experience PICC improvement was use of single lumen devices (40.1% to 53.4%, p<0.001) while the metric that improved the least was PICC avoidance in CKD (80.8% to 84.1%, p<0.001).
After accounting for the multi-level design (patients nested within hospitals), MAGIC improved PICC appropriateness by 24.7% (95% CI: 22.9%, 26.5%; odds ratio=2.30 p<0.001). Median hospital rate of improvement in appropriateness was 9.7% (IQR, 4.1% to 17.5%). Hospitals that most reduced inappropriate PICC use were more likely to have implemented electronic order entry sets that incorporated MAGIC recommendations into computerized decision support. In these facilities, much of this change was spurred by PICC teams and vascular access device inserters.

Conclusions: An intervention based on MAGIC substantially reduced inappropriate use of PICCs in multiple hospitals across Michigan. Order sets that incorporated MAGIC were key elements of this change.