The Agency for Healthcare Research and Quality Inpatient Quality Indicator #11 overall mortality rate does not accurately assess mortality risk after abdominal aortic aneurysm repair

Abstract

Objective: The Agency for Healthcare Research and Quality (AHRQ) Inpatient Quality Indicator (IQI) #11, abdominal aortic aneurysm (AAA) repair mortality rate, is a measure of hospital quality that is publically reported but has not been externally validated. Because the IQI #11 overall mortality rate includes both intact and ruptured aneurysms and open and endovascular repair, we hypothesized that IQI #11 overall mortality rate does not provide accurate assessment of mortality risk after AAA repair and that AAA mortality cannot be accurately assessed by a single quality measure.

Methods: Using AHRQ IQI software version 4.2, we calculated observed (O) and expected (E) mortality rates for IQI #11 for all hospitals performing more than 10 AAA repairs per year in the Nationwide Inpatient Sample for the years 2007 to 2011. We used Spearman correlation coefficient to compare expected rates as determined by IQI #11 overall mortality rate risk adjustment methodology and observed rates for all AAA repairs in four cohorts stratified by aneurysm stability (ruptured vs intact) and method of repair (open vs endovascular).

Results: Among 187,773 AAA repairs performed at 1268 U.S. hospitals, hospitals' IQI #11 overall expected rates correlated poorly with their observed rates (E: 5.0% ± 4.4% vs O: 6.0% ± 9.8%; r = .49). For ruptured AAAs, IQI #11 overall mortality rate methodology underestimated the mortality risk of open repair (E: 34% ± 7.2% vs O: 40.1% ± 38.2%; r = 0.20) and endovascular repair (E: 24.8% ± 9% vs O: 27.3% ± 37.9%; r = 0.08). For intact AAA repair, IQI #11 overall mortality rate methodology underestimated the mortality risk of open repair (E: 4.3% ± 2.4% vs O: 6.3% ± 16.1%; r = .24) but overestimated the mortality risk of endovascular repair (E: 1.3% ± 0.8% vs O: 1.1% ± 3.7%; r = 0.25). Hospitals' observed mortality rates after intact AAA repair were not correlated with their mortality rates after ruptured AAA repair (r = 0.03).

Conclusions: IQI #11 overall mortality rate fails to provide accurate assessment of inpatient mortality risk after AAA repair. Thus, it is inappropriate to use IQI #11 overall mortality rate for quality reporting. The accuracy of separate quality measures that assess mortality risk after repair of ruptured and intact AAAs, stratified by the use of open or endovascular repair, should be examined.