Abstract 342: Variation In Contrast Volumes For Percutaneous Interventions Across Hospitals In Northern New England: It Is Not Just Patient And Procedural Characteristics

Background: Contrast volume (CV) is an important risk factor for acute kidney injury following PCI. Hypothesis: Hospitals participating in a regional PCI registry differ in their use of CV and this difference is a consequence not only of patient characteristics but of modifiable site-specific practices. Methods: Data from 4110 consecutive PCIs performed during 2011 at 7 member sites of the Northern New England Cardiovascular Disease Study Group PCI Registry. Contrast volume was categorized into 100ml increments. Ordered multivariate logistic regression identified patient characteristics associated with CV and compared site use of CV after controlling for them. Then, procedural characteristics significantly associated with CV were added to the model and the influence of site reassessed. Results: CV differed across sites (Figure 1). Patient characteristics significantly associated with CV included age, sex, body mass index, prior CABG or PCI, ejection fraction, and pre-procedure estimated GFR. After adjusting for patient characteristics there remained a significant difference in CV use with Site 7 (OR 0.45, p<0.01) and Site 3 (OR 0.25, p<0.01) using less contrast volume than referent Site 4. Available procedural characteristics significantly associated with CV included access site (OR 0.73), Ad Hoc PCI (OR 3.65), CTO PCI (OR 2.1), number of vessels intervened on (OR 0.67-0.99) and number of lesions attempted (OR 1.59-9.40). After controlling for patient and procedural characteristics, 4 of 7 sites (Sites 3, 5, 7, and 10) were found to use significantly less contrast volume than referent Site 4. A ’Chi-Pie’ model showed that 16% of the variation was explained by patient characteristics, 45% by procedure characteristics, and 39% by site-specific factors. Conclusions: Even after controlling for patient and procedural characteristics, there remains significant variation in CV across northern New England PCI sites. Some of the known procedural characteristics may be modifiable. In addition, as yet unmeasured site-specific characteristics (e.g., method of assessing EF, catheter size, automatic vs hand injections) contribute to 39% of the variation. A better understanding of what drives the use of CV represents an opportunity for some sites in northern New England to decrease CV and improve PCI outcomes. ![Graphic][1] [1]: /embed/inline-graphic-1.gif