Variations in US hospital performance on imaging-use measures

Abstract

Background: The Hospital Outpatient Quality Reporting Program (HOQR) publicly reports measures of US hospitals' use of 4 imaging studies that may be problematic if overused: magnetic resonance imaging (MRI) for low back, follow-up imaging after screening mammography, and abdominal and thoracic computed tomography (CT) with and without contrast.

Objectives: To characterize performance on these measures, determine whether performance was consistent across measures, and identify hospital characteristics associated with highest-decile imaging use.

Research design: Cross-sectional analysis.

Measurements: Correlation across measures was assessed using Spearman rank order tests. We linked 2008 HOQR data to the 2009 American Hospital Association Survey and used multivariable logistic regression to examine associations between hospital characteristics and the likelihood of highest-decile imaging use.

Results: Imaging use varied widely. Imaging use was weakly correlated (ρ<0.10) across most measures. Compared with hospitals with moderate imaging volume (25th to 75th percentile), hospitals with low volume (<25th percentile) were more likely to report highest-decile imaging use on all measures [adjusted odds ratios (95% confidence interval) range from 1.38 (1.05-1.80) for CT Abdomen to 4.22 (3.04-5.84) for MRI Back]. Rural hospitals were more likely to report highest-decile use on most measures [MRI: 1.42 (1.21-1.68), CT Abdomen: 1.46 (1.28-1.66), and CT Thorax: 1.32 (1.16-1.51)]. For-profit hospitals were more likely to report highest-decile use on mammography [1.47 (1.10-1.98)] and CT Thorax measures [1.71 (1.28-2.27)].

Conclusions: Wide variations in imaging use and extraordinarily high use at some hospitals may indicate that imaging overuse occurs at US hospitals. The effectiveness of the HOQR measures to decrease imaging overuse remains to be seen.