Association Between Hospital Recognition for Resuscitation Guideline Adherence and Rates of Survival for In-Hospital Cardiac Arrest

Abstract

Background Hospitals participating in the national Get With The Guidelines-Resuscitation registry receive an award for high rates of adherence to quality metrics for in-hospital cardiac arrest. We sought to evaluate whether awards based on these quality metrics can be considered a proxy for performance on cardiac arrest survival. Methods and Results Among 195 hospitals with continuous participation in Get With The Guidelines-Resuscitation between 2012 and 2015, we identified 78 that received an award (Gold or Silver) for ≥85% compliance for all 4 metrics for in-hospital cardiac arrest-time to chest compressions, ≤1 minute; time to defibrillation, ≤2 minutes; device confirmation of endotracheal tube placement; and a monitored/witnessed arrest-for at least 12 consecutive months during 2014 to 2015. Award hospitals had higher cardiac arrest volumes than nonaward hospitals but otherwise had similar site characteristics. During 2014 to 2015, award hospitals had higher rates of return of spontaneous circulation for in-hospital cardiac arrest than nonaward hospitals (median [interquartile range], 71% [64%-77%] versus 66% [59%-74%]; Spearman ρ, 0.19; P=0.009). However, rates of risk-standardized survival to discharge at award hospitals (median, 25% [interquartile range, 22%-30%]) were similar to nonaward hospitals (median, 24% [interquartile range, 12%-27%]; Spearman ρ, 0.13; P=0.06). Among hospitals in the best tertile for survival to discharge in 2014 to 2015, 55.4% (36/65) did not receive an award, with poor discrimination of high-performing hospitals by award status (C statistic, 0.53). Similarly, there was only a weak association between hospitals' award status in 2014 to 2015 and their rates of survival to discharge in the preceding 2-year period (Spearman ρ, 0.16; P=0.03). Conclusions The current recognition mechanism within a national registry for in-hospital cardiac arrest captures hospital performance on return of spontaneous circulation but is not well correlated with survival to discharge. This suggests that current awards for resuscitation quality may not adequately capture hospital performance on overall survival-the outcome of greatest interest to patients.

Keywords: heart arrest; outcome and process assessment (health care).

Figure 1:. Risk-Standardized Rates of Survival to Discharge at Award and Non-Award Hospitals.

Median risk-standardized survival (interquartile range) at award and non-award hospitals, concurrently during the recognition period (2014–2015) (A), and during a preceding 2-year period, 2012–2013 (B). Correlation coefficient represents Spearman’s point-biserial correlation.

Figure 2:. Receiver Operator Characteristic (ROC) Curve for Relationship Between Hospital Award Status and Risk-Standardized Survival Tertile.

The ROC for award status in identifying hospitals in the highest tertile for risk-standardized survival during the award period (2014–2015), and during a preceding 2-year period (2012–2013), are displayed. Abbreviations: RSSR, risk-standardized survival rate

Figure 3:. Award and Non-Award Hospitals by Tertile of Hospital Risk-Standardized Survival.

The proportion of hospitals in the highest tertile for risk-standardized survival which were non-award hospitals, as well as the proportion of hospitals in the lowest tertile for risk-standardized survival which were award hospitals, are displayed. Rates are presented for both the award period (2014–2015), and for a preceding 2-year period (2012–2013). Abbreviations: RSSR, risk-standardized survival rate

Figure 4:. Risk-Adjusted Rates of Return of Spontaneous Circulation at Award and Non-Award Hospitals.

Median risk-standardized rates of return of spontaneous circulation (ROSC) and interquartile ranges at award and non-award hospitals, concurrently during the recognition period (2014–2015) (A), and during a preceding 2-year period, 2012–2013 (B). Correlation coefficient represents Spearman’s point-biserial correlation.