Implementation of a Multidimensional Strategy to Reduce Post-PCI Bleeding Risk

Abstract

BACKGROUND:

The American College of Cardiology Reduce the Risk: PCI Bleed Campaign was a hospital-based quality improvement campaign designed to reduce post-percutaneous coronary intervention (PCI) bleeding events. The aim of the campaign was to provide actionable evidence-based tools for participants to review, adapt, and adopt, depending upon hospital resources and engagement.

METHODS:

We used data from 8 757 737 procedures in the National Cardiovascular Data Registry between 2015 and 2021 to compare patient and hospital characteristics and bleeding outcomes among campaign participants (n=195 hospitals) and noncampaign participants (n=1384). Post-PCI bleeding risk was compared before and after campaign participation. Multivariable hierarchical logistic regression was used to determine the adjusted association between campaign participation and post-PCI bleeding events. Prespecified subgroups were examined.

RESULTS:

Campaign hospitals were more often higher volume teaching facilities located in urban or suburban locations. After adjustment, campaign participation was associated with a significant reduction in the rate of bleeding (bleeding: adjusted odds ratio, 0.61 [95% CI, 0.53–0.71]). Campaign hospitals had a greater decrease in bleeding events than noncampaign hospitals. In a subgroup analysis, the reduction in bleeding was noted in non–ST-segment–elevation acute coronary syndrome and ST-segment–elevation myocardial infarction patients, but no significant reduction was seen in patients without acute coronary syndrome.

CONCLUSIONS:

Participation in the American College of Cardiology Reduce the Risk: PCI Bleed Campaign was associated with a significant reduction in post-PCI bleeding. Our results underscore that national quality improvement efforts can be associated with a significant impact on PCI outcomes.

WHAT IS KNOWN

• Postpercutaneous coronary intervention bleeding is associated with adverse patient outcomes and increased health care costs.

• The definition of postpercutaneous coronary intervention bleeding expands beyond procedural access site complications.

• There are a number of evidence-based strategies to reduce postpercutaneous coronary intervention bleeding events.

WHAT THE STUDY ADDS

• This article provides a quality improvement framework to improve postpercutaneous coronary intervention bleeding rates on a large scale.

• This article demonstrates the value of choice for campaign participants to choose their improvement strategies from the toolkit for their improved outcomes.

In 2011, the National Cardiovascular Data Registry CathPCI Registry released an in-hospital risk-adjusted quality metric to capture observed bleeding events post-percutaneous coronary intervention (PCI). Due to the association between bleeding events and adverse patient outcomes, the metric expanded beyond vascular access complications to include additional bleeding events, such as hemorrhagic stroke, pericardial tamponade, and red blood cell transfusions. The metric was designed to inform clinical decision-making and direct the use of bleeding avoidance strategies (BAS) to improve the safety of PCI procedures.^1–5 It has been endorsed by the National Quality Forum, Measure 2459. The CathPCI Registry Q4 2017 Report demonstrated a national average of 3.89%, for the rolling 4 quarters of data, with significant hospital-level variation observed in the post-PCI in-hospital risk-adjusted quality metric. The American College of Cardiology (ACC) Reduce the Risk: PCI Bleed Quality Improvement (QI) Campaign was developed to reduce bleeding at low performing sites. The campaign used existing data collected in the CathPCI Registry and required no additional data elements or cost for participation. It focused on the adoption of BAS such as increased radial artery access and vascular closure device use.^6,7 This report summarizes the details of the campaign and assesses the association between campaign participation and post-PCI bleeding rates.

METHODS

ACC Reduces the Risk: PCI Bleeding QI Campaign

Because of the sensitive nature of the data collected for this study, the ACC National Cardiovascular Data Registry will not accept access to the data set used in this study. The QI Campaign launched in 2018 and remained active through December 2020. The framework included 4 main areas of focus: a change package, developing a learning network, recognition, and evaluation. The change package consisted of a self-assessment, 17 evidence-based clinical tools, 11 project management resources, and a dashboard designed for participants to track the progress of their local improvement efforts (Supplemental Material). The learning network was created on the ACC QI for Institutions website (www.cvquality.acc.org), supported through webinars, conference presentations, best practice sharing, and an active listserv. Press kit releases, credit as a high activity weight for MACRA (The Medicare Access and CHIP Reauthorization Act of 2015) MIPS (Merit-Based Incentive Payment System) Improvement Activities, and hospital name publication in Cardiology and US News & World Report Best Hospitals Edition recognized participants in the campaign. The study was approved by the institutional review board of Duke University Medical Center, which determined that the study met the definition of research not requiring informed consent. Hospital participation in the QI Campaign was voluntary.

Six metrics were developed and monitored within the campaign dashboard. The primary end point was in-hospital risk-standardized rate of bleeding events for all patients with PCI (metric 1). The additional 5 metrics focused on process and outcome drivers to monitor the adoption of BAS. They included the following: metric 2: whole blood/red blood cell transfusion; metric 3: procedures with an observed bleeding event; metric 4: anticoagulation utilization; metric 5: access site utilization; and metric 6: method for closure for arterial access site. The campaign’s clinical tool kit was comprised of a broad selection of evidence-based tools. Protocol examples provided the participants an opportunity to review or adopt protocol enhancements for obtaining arterial access and sheath removal. Examples of preprocedural and postprocedural order sets rendered an opportunity for participants to compare their existing orders with those provided in the campaign. The CathPCI Bleeding Risk Calculator App and a risk-concordant framework for BAS included patient-centered approaches for BAS utilization.⁸ Nursing-specific tools were included in the forms of competencies, education, and bedside tools. To support successful adoption at the local level, the tool kit was packaged into groups that supported transparent assessment of hospital resources with implementation requirements. The tools cross-walked to ≥1 campaign metric and were placed into the following groupings: preprocedural, intraprocedural, postprocedural, and pharmacotherapy groupings.

Study Population

The CathPCI Registry is an initiative of the ACC and the Society for Cardiovascular Angiography and Intervention and has been described previously.⁹ This registry records data on patient and hospital characteristics, clinical presentations, hospital length of stay, treatment, and in-hospital outcomes from PCI procedures from >1800 sites from the United States and internationally. For this study, only the US sites were included. The National Cardiovascular Data Registry has a comprehensive data quality program, including both data quality reporting specifications for data capture and transmission, as well as an auditing program.^10,11 Data set variables are determined and defined by clinician work groups. For this study, all PCIs performed between May 11, 2015, and December 31, 2021, that fulfilled the Data Quality Report requirements, were included. Campaign participants were given a 30-day rollout period once the hospital opted in for participation. Procedure- and patient-level exclusions included the following: nonindex PCIs, procedures with mechanical hemodynamic support, and procedures having index PCI during the rollout period.

Definitions and Outcomes

The Reduce the Risk: PCI Bleed Campaign hospitals were defined as hospitals that opted into the campaign, had active participation in the CathPCI registry, submitted a team roster, and performed at least 25 PCIs during the study period. The primary outcome for this analysis was post-PCI bleeding, as defined by the CathPCI Registry (metric 40) PCI in-hospital risk-standardized bleeding. An observed bleeding event was defined as any of the following: arterial access site bleeding, gastrointestinal, genitourinary, hematoma at access site, retroperitoneal, or other location bleeding event that occurred between the start of the procedure and 72 hours after PCI or discharge (whichever occurred first). Additional bleeding events assessed included hemorrhagic stroke, pericardial tamponade, red blood cell transfusion with a preprocedure Hgb >8 h/dL, or an absolute Hgb decrease from pre-PCI of ≥4 g/dL with a preprocedure Hgb ≤16 g/dL. The rate of the primary outcome was compared among all hospitals before and after participation. Secondary outcomes included 2 process measures: (1) rates of radial access and (2) use of vascular closure devices among patients undergoing transfemoral PCI and 1 outcome measure: red blood cell transfusion, before and after campaign participation among campaign hospitals.

Statistical Analysis

Patient and hospital characteristics were compared in patients before and after the campaign in participating hospitals and in patients seen at campaign participating hospitals versus noncampaign participating hospitals. Continuous variables are reported as median (Q1, Q3) while categorical variables are reported as standardized differences for all comparisons (Tables 1 through 3). The odds of bleeding and, separately, each secondary bleeding outcome, after versus before the intervention, were calculated using logistic regression, in campaign hospitals only. Both adjusted and unadjusted models were performed. To access the relationship between secular trends and outcomes, the odds ratio (95% CI) of all outcomes was calculated in hospitals before and after the campaign in campaign hospitals and time-matched noncampaign hospitals. A modified difference-in-differences (DiD) analysis was conducted, where the outcome was bleeding. As all outcomes are dichotomous, a modification of the DiD technique was used to allow for modeling on the logit scale. A mixed-effects logistic regression model was used to account for nonindependence of patients within hospitals. There were independent terms for campaign status pre- and post-campaign and the interaction between campaign and pre/post. The P value of the interaction term is presented as the DiD P value. The exponent of the interaction term was presented as the DiD odds ratio (95% CI). A term for the duration between the intervention rollout and hospital admission was included in all DiD models. The equal trends assumptions were verified using predicted probability of bleeding from a linear probability model. All statistical analyses were conducted in accordance with a prespecified statistical analysis plan and programmed in SAS, version 9.4.

Table 1.

Comparison of Patient Characteristics Between Campaign and Noncampaign Hospitals

Table 3.

Comparison of Patient Characteristics at Campaign Hospitals Before and After Implementation

RESULTS

Study Sample

Between May 11, 2015, and December 31, 2021, 8 757 737 PCI procedures were performed at 1890 hospitals. After applying procedural/patient-level exclusion criteria 4 089 828 index PCI procedures were included in the procedural analysis (Figure 1). One hundred ninety-five hospitals were identified as campaign participants, and 1384 hospitals comprised the noncampaign cohort. There were small differences between patients treated at campaign hospitals compared with those treated at noncampaign hospitals (Table 1). Due to the large sample size, many of these differences reached statistical significance. Hospital characteristics included more urban and teaching hospitals in the campaign group compared with the noncampaign group (Table 2). There were no significant differences in patient characteristics at campaign hospitals before and during campaign participation (Table 3).

Figure 1.

Study population CONSORT diagram (Consolidated Standards of Reporting Trials). This figure depicts the initial study population through the final study population after applying the exclusion criteria. PCI indicates percutaneous coronary intervention; and Pts, patients.

Table 2.

Comparison of Hospital Characteristics Between Campaign and Noncampaign Hospitals

Primary and Secondary Outcomes

The unadjusted rates of post-PCI bleeding before and after campaign participation are shown in Table 4. The primary and secondary outcomes were improved after campaign participation (Figure 2). After adjustment, there was a statistically significant association between campaign participation and lower odds for bleeding risks (adjusted, 0.61 [0.53–0.71]; unadjusted, 0.63 [0.58–0.69]) and a higher odds of radial access (1.81 [1.68–1.95]) and vascular closure device use (1.44 [1.31–1.59]; Figure 3). There was an increased odds of transfusion in campaign participants. In a subgroup analysis, the reduction in bleeding was noted in non–ST-segment–elevation acute coronary syndrome and ST-segment–elevation myocardial infarction patients, but no significant reduction was seen in patients without acute coronary syndrome.

Table 4.

Before and After Outcomes for Campaign Hospitals

Figure 2.

Comparison of bleeding outcomes by campaign status. This figure depicts the incidence of bleeding outcomes before and after the campaign, by campaign status. The primary observed patient bleeding outcomes and secondary process outcomes were included. The models show the covariate-adjusted predicted probability of each outcome. The bars represent the 95% CI of the adjusted mean probability.

Figure 3.

Study population odds ratio plot diagram. This figure depicts unadjusted and adjusted outcomes in patients receiving a percutaneous coronary intervention before and after the campaign for campaign participants. The primary observed patient bleeding outcomes and secondary process outcomes were included.

DiD Analysis

To assess the relationship between possible secular trends in the primary and secondary outcomes, we compared campaign hospitals with noncampaign hospitals in a DiD analysis. Patient characteristics between 2 groups are shown in Table 1. Table 4 shows the DiD and the unadjusted and adjusted odds ratios and corresponding 95% CIs for the primary and secondary outcomes. There was a decrease in the primary end point among campaign participants (bleeding: adjusted odds ratio, 0.70 [95% CI, 0.66–0.73]; unadjusted odds ratio, 0.72 (95% CI, 0.65–0.78]). The reduction in the odds of bleeding in noncampaign hospitals was significant (DiD adjusted odds ratio, 0.83 [95% CI, 0.80–0.87]) but less than the reduction in bleeding risk in campaign hospitals (adjusted DiD analysis, P<0.0001; Table 4).

The adjusted odds of transfusion was not significantly different between campaign and noncampaign hospitals (P=0.2001). There was a significant increase in the odds of radial access in both campaign and noncampaign hospitals. The increase in radial access was higher in campaign hospitals than noncampaign hospitals (DiD adjusted odds ratio, 1.07 [95% CI, 1.06–1.08]), P=<0.0001). The DiD analysis shows no significant change in increase of manual closure between campaign and noncampaign hospitals (DiD adjusted odds ratio, 0.98 [95% CI, 0.89–1.08]; P=0.669; Table 4).

DISCUSSION

The PCI Bleeding Campaign was designed to provide participating hospitals evidence-based tools and resources to define and implement hospital-specific BAS. Our study was a real-world evaluation of QI tools and their individualized adoption by hospitals. Patient characteristics at campaign hospitals were similar before and during the study period, and there was significant uptake of BAS at participating hospitals. This study shows that this QI Campaign, with voluntary participation, was associated with a statistically significant reduction in adjusted and unadjusted post-PCI bleeding events compared with noncampaign hospitals.

Bleeding following PCI remains a largely preventable complication associated with increased patient morbidity and mortality, as well as health care costs. Previously published studies have demonstrated the adoption of BAS varies within a health care system and is inconsistently utilized.¹² Other studies have shown the number of BAS may be observed less as a patient’s age increases, a population that is likely to benefit the most.¹³ Additional studies have proven the effectiveness of known BAS strategies^12,14,15 and a lack of correlation between clinical bleeding and transfusion practices.⁵ Notably, the campaign produced a single resource for known BAS strategies. The Campaign Toolkit and learning community facilitated actionable knowledge, change management, and team-based approaches to reduce bleeding events in patients with PCI. Hospitals participating in this study were able to successfully adopt BAS for improved patient outcomes at their hospitals.

There was a significant decrease in bleeding among campaign participants and also a decrease among nonparticipants. However, a statistically stronger decline in bleeding was seen in participating hospitals. Of note, the rate of bleeding was lower among nonparticipants compared with participants at the start of the campaign. These results may reflect several issues. First, it is possible that hospitals that participated in the campaign did so because of awareness that their bleeding rates were higher, and they were motivated to reduce them. Second, there may be floor effect of bleeding reduction whereby bleeding complications cannot be reduced further regardless of any intervention or QI efforts. Finally, it appears that the Reduce the Risk PCI Bleed Campaign sites had higher rates of bleeding before the campaign, and the campaign improved their rates commensurate with leading sites.

Our study complements other cardiovascular QI campaigns. In May 2001, the Centers for Medicare and Medicaid Services and the Joint Commission created Core Quality Measures, which included acute myocardial infarction and heart failure. Shortly after, the Door-to-Balloon: An Alliance for Quality was developed as a novel campaign to accelerate the adoption of key strategies to reduce the door-to-balloon time. This QI effort was focused on identifying hospital operations to improve a process measure.¹⁶ Despite the sharing of evidence-based strategies, barriers persisted leading to only 30.4% of the participants incorporating at least 4 of the 5 key strategies at the end of the campaign.¹⁷

The construct differences in these 2 campaigns highlight how the design of QI campaigns may facilitate accelerated adoption of proven improvement strategies. Future QI campaigns should include actionable evidence-based tools, which support existing team-based workflows for optional adoption, as well as including both process and outcome measures supported by the campaign for comparison.

Limitations

While there is widespread adoption of the ACC CathPCI Registry, hospital participation remains optional. Hospitals that opted into the campaign may survey and report bleeding complications more comprehensively than hospitals that did not opt into participation. Adoption of BAS, such as radial use, may not completely account for the results. Transfusion is a subcomponent of the National Cardiovascular Data Registry bleeding metric and thus is challenging to separate from bleeding. As with any observational study, our findings indicate association, not causation. Whether the reductions in bleeding are sustainable beyond campaign participation are unknown and should be the focus of future analysis. Finally, it is possible that the greater reduction in bleeding complications among campaign hospitals was due to the fact that there is a floor effect for bleeding among noncampaign hospitals whereby they are unable to improve further.

Conclusions

Participation in the ACC Reduce the Risk: PCI Bleed Campaign increased the use of BAS and reduced bleeding complications among patients undergoing PCI.

ARTICLE INFORMATION

Acknowledgments

The authors thank the American College of Cardiology staff who supported the framework of the campaign. We greatly appreciate hospitals that participated in the American College of Cardiology Reduce the Risk: PCI Bleed Campaign and their CV teams that facilitate change within their hospitals.

Sources of Funding

This research was supported by the American College of Cardiology National Cardiovascular Data Registry (NCDR). The views expressed in this article represent those of the authors and do not necessarily represent the official views of the NCDR or its associated professional societies identified at www.ncdr.com.

Disclosures

Dr Amin has institutional grant support (modest) from GE Healthcare and Chiesi. Dr Abbott has the following relationships with industry: research: Boston Scientific and Microport; consulting: Abbott, Medtronic, Penumbra, Shockwave, and Philips. Dr Masoudi had a contract with the American College of Cardiology for his role as the Chief Scientific Advisor, National Cardiovascular Data Registry. The other authors report no conflicts.

Supplemental Material

Table S1