In:
Circulation: Cardiovascular Quality and Outcomes, Ovid Technologies (Wolters Kluwer Health), Vol. 7, No. suppl_1 ( 2014-07)
Abstract:
Background: Nationally, the diagnostic yield of coronary angiography is low. Little is understood about how referring providers vary in propensity to refer for coronary angiography. Understanding these sources of variation and improving the yield of coronary angiography has the potential to reduce costs and enhance value. Methods: We identified all cases of diagnostic coronary angiography performed at the Massachusetts General Hospital from January 1, 2012 until June 30, 2013 (Analysis A). We then identified all positive angiograms, as defined as at least one epicardial coronary stenosis greater or equal than 50%. We excluded angiograms for STEMI, NSTEMI and excluded angiograms in patients with cardiomyopathy, LVEF 〈 50%, valve disease, history of cardiac transplantation, prior PCI, or prior CABG (Analysis B). For both Analysis A and Analysis B, we calculated proportions of positive angiograms for each referring provider and calculated variance, medians, and standard deviations of those proportions among providers. Referring providers with fewer than 10 included angiograms (Analysis B) over the 18 month period were excluded for both analyses. We compared the variances of the two analyses with an F-test. Logistic regression models were developed using both characteristics of referring physician and characteristics of patients. Results: 5186 total coronary angiograms were performed, of which 1334 met inclusion criteria. 117 physicians ordered at least one angiogram, and 34 physicians ordered at least 10 angiograms. The positivity rate for Analysis A was greater than the positivity rate for Analysis B (64.3% versus 50.9%, p 〈 0.0001). The variance of the positivity rates for Analysis A was 0.0099, and the variance of the positivity rates for Analysis B was 0.0201 (F-test = 0.045). Preliminary results suggest that more variance is due to patient factors (R-squared = 0.46) than physician factors (R-squared = 0.04). The distribution of positivity rates by referring physician for Analysis B is shown in Figure 1. Conclusions: Variation among referring providers increases after cases with clearer guidelines are excluded. Preliminary results suggest this variance is related to different patient characteristics (case mix) rather than referring physician characteristics.
Type of Medium:
Online Resource
ISSN:
1941-7713
,
1941-7705
DOI:
10.1161/circoutcomes.7.suppl_1.228
Language:
English
Publisher:
Ovid Technologies (Wolters Kluwer Health)
Publication Date:
2014
detail.hit.zdb_id:
2453882-6
Bookmarklink