The purpose of our study was to evaluate the use of bleeding-avoidance strategies (BAS) and risk-adjusted bleeding over time in patients ≥80 years of age undergoing primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction. We analyzed data from the CathPCI Registry from July 1, 2006 through June 30, 2009. Patients were included if they were ≥80 years old, presented with ST-segment elevation myocardial infarction, and underwent primary PCI. We evaluated trends in use of BAS (direct thrombin inhibitors, vascular closure devices, and radial access) and risk-adjusted bleeding over time. Of 10,469 patients ≥80 years old undergoing primary PCI, 1,002, (9.6%) developed a bleeding complication. Use of direct thrombin inhibitors and vascular closure devices increased over time (12.8% to 24.9% and 29.2% to 32.7%, p <0.01 and <0.05 for trends, respectively). Radial access was extremely uncommon (<1%) and did not change over the course of the study. In multivariable analyses, use of BAS was associated with lower bleeding. However, over the course of the study period, overall risk-adjusted bleeding did not decrease significantly (9.9% to 9.4%, p = 0.14 for trend). In conclusion, patients ≥80 years old undergoing primary PCI are at high risk of bleeding, and despite significant increases in use of BAS, the overall rate of bleeding complications remains high.
Patients ≥80 years of age with ST-segment elevation myocardial infarction (STEMI) undergoing primary percutaneous coronary intervention (PCI) are at high risk for bleeding complications. However, the use and effectiveness of strategies to decrease bleeding are not well understood. The aim of the present study therefore was to describe the use of bleeding-avoidance strategies (BAS) and associated in-hospital bleeding rates in patients ≥80 years of age with STEMI undergoing primary PCI. Furthermore, we sought to assess whether changes in BAS over time have been associated with decreases in risk-adjusted bleeding. To accomplish these goals, we analyzed data from the National Cardiovascular Data Registry (NCDR) Catheter Percutaneous Coronary Intervention (CathPCI) Registry, which provides detailed clinical information on patients undergoing PCI at >1,000 United States hospitals. A major strength of the NCDR is that it captures patients seen in routine clinical practice, many of whom may have been excluded from randomized trials.
Methods
The NCDR CathPCI Registry is a large voluntary quality improvement program cosponsored by the American College of Cardiology and the Society for Cardiovascular Angiography and Interventions. The registry includes clinical and in-hospital outcome data on patients undergoing cardiac catheterization and PCI procedures including baseline clinical and sociodemographic characteristics, adjunctive therapies, complications, and mortality. Definitions for data elements within the registry are available on the NCDR Web site ( http://www.ncdr.com ).
For the present study we identified patients who presented with STEMI and underwent PCI (emergency or salvage) within 6 hours of arrival. Patients with a door-to-balloon time of <15 minutes were excluded to eliminate possible coding errors. We restricted our analysis to PCIs performed at hospitals that participated in the CathPCI Registry continuously from July 1, 2006 through June 30, 2009 (i.e., reported ≥1 PCI in each consecutive year). Patients transferred from another acute-care hospital for PCI were excluded because details of their initial presentation and management were not captured. In addition, we considered information only from the first PCI performed during a hospital stay.
Anticoagulant strategies were divided into approaches commonly used in clinical practice: (1) unfractionated heparin (UFH) alone, (2) UFH plus glycoprotein IIb/IIIa inhibitor, (3) low-molecular weight heparin plus glycoprotein IIb/IIIa, and (4) direct thrombin inhibitor. Arterial access site was characterized as femoral, radial, or brachial. Hemostasis techniques were dichotomized into vascular closure device (suture, staple, sealant, or other) or manual compression (which included mechanical compression devices). Given the uncertainty of the devices included under “other,” we performed a sensitivity analysis for vascular closure devices that excluded this classification. Because there was no significant change in results, the designation of other was retained in the final analysis. We used the term “bleeding-avoidance strategies” to refer to the use of thrombin inhibitors, vascular closure devices, or radial access based on previous studies showing lower bleeding rates with each intervention compared to other therapies.
The primary outcome was in-hospital bleeding, defined as any bleeding event (percutaneous entry site, retroperitoneal, gastrointestinal, genitourinary, or other) causing a decrease in hemoglobin >3.0 g/dl, requiring a transfusion, or prolonging hospital stay. Secondary outcomes included thrombocytopenia, transfusion of blood products, vascular complications (including access site occlusion, peripheral embolization, dissection, pseudoaneurysm, and arteriovenous fistula), and in-hospital mortality.
To characterize bleeding risk associated with primary PCI in elderly patients, we compared unadjusted outcomes observed between patients ≥80 and patients <80 years of age using the chi-square test. We focused the remainder of the analyses on patients ≥80 years old because they represented the primary subgroup of interest. In patients ≥80 years old we compared characteristics of patients who developed a bleeding event to those of patients with no bleeding using a 2-sided t test for continuous variables and chi-square test for categorical variables. We then examined the independent association of use of BAS with bleeding events. We also stratified the population by use of 0 BAS, 1 BAS, or ≥2 BAS and examined the effects of the number of BAS on the outcome of bleeding. Unadjusted and adjusted odds ratios were reported.
To account for differences in patients’ risk of bleeding, we used a previously developed NCDR bleeding risk model. The model adjusts for age, female gender, weight, cardiogenic shock, previous congestive heart failure, previous valvular surgery, cerebrovascular disease, peripheral vascular disease, hypertension, previous PCI, intra-aortic balloon pump, estimated glomerular filtration rate, New York Heart Association class, and PCI status of salvage/emergency. We also investigated the presence of a risk–treatment paradox (where patients at higher risk are less likely to receive beneficial strategies) by comparing use of BAS across strata of predicted bleeding risk (lower <6%, intermediate 6% to 12%, and higher ≥12%).
We examined trends in BAS and risk-adjusted bleeding rates from 2006 through 2009. Risk-adjusted bleeding rates were calculated using the NCDR bleeding risk model. Changes in BAS were examined using the Cochran-Amitage trend test. Changes in risk-adjusted bleeding rates over time were evaluated using linear regression.
Results
From an initial sample of 218,935 cases with primary PCI for STEMI from July 2006 through June 2009, the following were excluded: 1,050 for not being the first PCI, 16,175 for PCI not coded as emergency or salvage, 54,511 for transfer from another hospital, 4,083 with door-to-balloon time <15 minutes or >6 hours, and 39,540 for PCI performed at hospitals that did not participate continuously in the CathPCI Registry from 2006 through 2009. This left a final sample of 103,576 procedures for analysis. Of these cases, 10.1% were performed in patients ≥80 years of age (n = 10,469). In these patients, unadjusted bleeding was nearly 2 times as common as in patients <80 years old (9.6% vs 4.9%, p <0.01). Similarly, rates of other adverse outcomes were higher in older patients including transfusion (16.0% vs 7.7%, p <0.01), thrombocytopenia (2.0% vs 1.1%, p <0.01), vascular complications (1.3% vs 0.8%, p <0.001), and crude in-hospital mortality (13.5% vs 3.9%, p <0.01).
A comparison of characteristics in patients ≥80 with and without bleeding is presented in Table 1 . Patients with bleeding were more likely to be women (63.6% vs 54.7%, p <0.01), have a history of hypertension (77.6% vs 73.7%, p <0.01) and chronic lung disease (17.2% vs 13.9%, p <0.01), and to present in congestive heart failure (23.9% vs 16.0%, p <0.01) or cardiogenic shock (20.5% vs 13.6%, p <0.01). There were no significant differences between groups in rates of renal insufficiency, diabetes, or previous MI. Bleeding at remote sites (gastrointestinal, genitourinary, or other) was more common in these patients than bleeding related to arterial access (percutaneous entry site or retroperitoneal, 67.7% vs 34.9%) and increased as a proportion of total bleeding over time (61.1% in 2006 vs 78.5% in 2009).
| Variable | Bleeding | p Value | |
|---|---|---|---|
| Yes | No | ||
| (n = 1,002) | (n = 9,467) | ||
| Age (years), mean ± SD | 84.5 ± 3.7 | 84.7 ± 4.3 | 0.27 |
| Women | 63.6% | 54.7% | <0.01 |
| Race | 0.27 | ||
| Caucasian | 89.9% | 87.9% | |
| Black | 3.6% | 3.7% | |
| Hispanic | 1.7% | 2.7% | |
| Other ⁎ | 4.7% | 5.6% | |
| Current smoker | 6.8% | 7.2% | 0.65 |
| Chronic renal insufficiency | 8.6% | 7.0% | 0.06 |
| Diabetes mellitus | 20.7% | 19.4% | 0.34 |
| Previous myocardial infarction | 17.2% | 19.2% | 0.12 |
| Hypertension † | 77.6% | 73.7% | <0.01 |
| Dyslipidemia ‡ | 53.7% | 53.7% | 0.99 |
| Previous congestive heart failure | 11.7% | 9.9% | 0.08 |
| Chronic lung disease | 17.2% | 13.9% | <0.01 |
| Cerebrovascular disease | 15.7% | 15.5% | 0.89 |
| Previous percutaneous coronary intervention | 13.3% | 18.0% | <0.01 |
| Previous coronary bypass | 6.2% | 8.6% | <0.01 |
| Clinical presentation | |||
| Congestive heart failure | 23.9% | 16.0% | <0.01 |
| Cardiogenic shock | 20.5% | 13.6% | <0.01 |
| Symptom onset >12 hours | 10.4% | 9.6% | 0.40 |
| Right heart catheterization | 7.5% | 3.7% | <0.01 |
⁎ Includes Asian, Native-American, and “other.”
† Blood pressure >140/90 mm Hg or on antihypertensive therapy.
‡ Total cholesterol >200 mg/dl or low-density lipoprotein ≤130 mg/dl or high-density lipoprotein <30 mg/dl or triglycerides >150 mg/dl or on dyslipidemia therapy.
Thrombin inhibitors were used in 18.3% of cases. Vascular closure devices were used in 31.7% of cases, and radial arterial access was uncommon (<1%). A comparison of characteristics in patients who received 0 BAS, 1 BAS, and ≥2 BAS is presented in Table 2 . Baseline co-morbidities were similar in the 2 groups. Patients presenting in congestive heart failure were more common in the group receiving 0 BAS (17.6%) compared to 1 BAS or ≥2 BAS (16.1% and 13.6% respectively, p <0.01). A similar trend was seen in patients who presented in cardiogenic shock (17.3% 0 BAS, 10.8% 1 BAS, 7.5% ≥2 BAS, p <0.01) and in those who received an intra-aortic balloon pump (16.9% 0 BAS, 7.6% 1 BAS, 2.9% ≥2 BAS, p <0.01).
| Variable | Total | 0 | 1 | ≥2 | p Value |
|---|---|---|---|---|---|
| (n = 10,469) | (n = 5,849) | (n = 3,942) | (n = 678) | ||
| Age (years), mean ± SD | 84.6 ± 4.2 | 84.6 ± 4.1 | 84.7 ± 4.3 | 85.0 ± 4.2 | <0.05 |
| Women | 55.6% | 55.5% | 55.9% | 54.3% | 0.71 |
| Race | <0.01 | ||||
| Caucasian | 88.1% | 88.6% | 87.1% | 88.9% | |
| Black | 3.7% | 4.0% | 3.3% | 3.0% | |
| Hispanic | 2.6% | 2.6% | 2.6% | 2.8% | |
| Other ⁎ | 5.5% | 4.8% | 6.8% | 5.2% | |
| Body mass index (kg/m 2 ), mean ± SD | 25.7 ± 5.0 | 25.7 ± 4.9 | 25.8 ± 5.1 | 25.5 ± 5.1 | 0.33 |
| Current smoker | 7.1% | 7.6% | 6.6% | 6.1% | 0.08 |
| Chronic renal insufficiency | 7.1% | 7.0% | 7.4% | 6.2% | 0.51 |
| Diabetes mellitus | 19.5% | 19.5% | 19.8% | 18.1% | 0.61 |
| Previous myocardial infarction | 19.0% | 19.3% | 18.9% | 16.5% | 0.21 |
| Hypertension † | 74.1% | 74.2% | 74.3% | 72.0% | 0.42 |
| Dyslipidemia ‡ | 53.7% | 53.2% | 54.6% | 53.0% | 0.36 |
| Previous congestive heart failure | 10.1% | 10.4% | 9.8% | 8.9% | 0.39 |
| Chronic lung disease | 14.2% | 14.5% | 13.7% | 14.9% | 0.45 |
| Cerebrovascular disease | 15.5% | 15.6% | 15.5% | 15.0% | 0.93 |
| Previous percutaneous coronary intervention | 17.6% | 17.3% | 18.1% | 16.5% | 0.42 |
| Previous coronary bypass | 8.4% | 8.5% | 8.5% | 6.3% | 0.15 |
| Clinical presentation | |||||
| Congestive heart failure | 16.8% | 17.6% | 16.1% | 13.6% | <0.01 |
| Cardiogenic shock | 14.2% | 17.3% | 10.8% | 7.5% | <0.01 |
| Symptom onset >12 hours | 9.6% | 9.4% | 10.0% | 9.6% | 0.65 |
| Intra-aortic balloon pump | 12.5% | 16.9% | 7.6% | 2.9% | <0.01 |
| Right heart catheterization | 4.1% | 5.3% | 2.9% | 1.9% | <0.01 |
| Bleeding | 9.6% | 11.1% | 8.1% | 5.0% | <0.01 |
⁎ Includes Asian, Native-American, and “other.”
† Blood pressure >140/90 mm Hg or on antihypertensive therapy.
‡ Total cholesterol >200 mg/dl or low-density lipoprotein ≤130 mg/dl or high-density lipoprotein <30 mg/dl or triglycerides >150 mg/dl or on dyslipidemia therapy.
Patients at higher risk of bleeding were modestly more likely to receive thrombin inhibitors than lower-risk patients (18.7% vs 16.9%, p <0.01; Table 3 ) but less likely to receive vascular closure devices (25.8% vs 37.3%, p <0.01). The combination strategy of thrombin inhibitor plus vascular closure device was used less frequently in higher-risk compared to lower-risk patients (5.3% vs 7.0%, p <0.05; Table 3 ). In patients who received 1 BAS the adjusted odds ratio for in-hospital bleeding was 0.76 (95% confidence interval 0.66 to 0.88), and in patients receiving ≥2 BAS the odds ratio was 0.45 (95% confidence interval 0.32 to 0.63; Table 4 ).
| Strategy | Lower Risk (<6%) | Average Risk (6–12%) | Higher Risk (≥12%) | p Value |
|---|---|---|---|---|
| (n = 2,152) | (n = 5,281) | (n = 3,036) | ||
| Anticoagulants | <0.01 | |||
| Unfractionated heparin alone | 12.8% | 15.2% | 18.5% | |
| Unfractionated heparin + glycoprotein IIb/IIIa inhibitor | 58.6% | 56.8% | 54.2% | |
| Low-molecular-weight heparin + glycoprotein IIb/IIIa inhibitor | 3.3% | 3.1% | 2.4% | |
| Thrombin inhibitor | 16.9% | 18.6% | 18.7% | |
| Vascular closure device | 37.3% | 32.6% | 25.8% | <0.01 |
| Thrombin inhibitor + vascular closure device | 7.0% | 6.8% | 5.3% | <0.05 |
| Outcome | 1 BAS | ≥2 BAS | p Value |
|---|---|---|---|
| Unadjusted bleeding | 0.73 (0.63–0.84) | 0.41 (0.29–0.58) | <0.01 |
| Risk-adjusted bleeding | 0.76 (0.66–0.88) | 0.45 (0.32–0.63) | <0.01 |
There was a steady increase in use of thrombin inhibitors and decrease in use of UFH plus glycoprotein IIb/IIIa in patients ≥80 years old over time (p <0.001 for trend; Figure 1 ). Of note, in patients receiving a thrombin inhibitor, many also received heparins and/or IIb/IIIa inhibitors, although this decreased modestly over time (thrombin inhibitor plus UFH 43.8% in 2006, 40.8% in 2007, 41.0% in 2008, 39.4% in 2009; thrombin inhibitor plus IIb/IIIa 53.9% in 2006, 46.3% in 2007, 42.3% in 2008, 34.8% in 2009, p <0.001 for trends). Use of vascular closure devices increased from 29.2% to 32.7% (p <0.05 for trend; Figure 2 ). In-hospital bleeding rate in patients ≥80 years of age was 9.6% during the entire study period. After risk adjustment using the NCDR bleeding model, there was a nonsignificant decrease in risk-adjusted bleeding over the 4 years (9.9% in 2006, 9.5% in 2007, 9.6% in 2008, 9.4% in 2009, p = 0.14 for trend). There was no significant decrease in risk-adjusted mortality over the same period (12.8% in 2006, 12.4% in 2007, 12.8% in 2008; 12.0% in 2009, p = 0.33 for trend).
