Vascular and bleeding complications are associated with significant morbidity and mortality, limited data are available on the relation between these complications and the platelet count in patients undergoing coronary angiography. We, therefore, performed a retrospective cohort analysis of 3,466 consecutive adults undergoing coronary angiography at a single institution during a 2-year period. The mean ± SD age was 60 ± 13 years; 58% of the patients were men, 67% were white, and 27% were black. The mean platelet count was 242 ± 86 × 10 9 /L. A total of 143 bleeding or vascular complications (4.2%), as defined using the American College of Cardiology CathPCI Registry criteria, a part of the American College of Cardiology National Cardiovascular Disease Registry, occurred. The risk of complications increased with increasing platelet counts. Compared to the quartile with the lowest platelet counts, the odds ratio (OR) for complications was 1.41 for the second quartile (95% confidence interval 0.83 to 2.37, p = 0.20), 1.74 for the third quartile (95% confidence interval 1.04 to 2.93, p = 0.04), and 2.04 for the fourth quartile (95% confidence interval 1.21 to 3.44, p = 0.008). On multivariate logistic regression analysis, age, percutaneous coronary intervention, and increasing platelet counts were independent predictors of bleeding and vascular complications. When the analysis was limited to the 2,220 patients undergoing diagnostic coronary angiography without percutaneous coronary intervention, an increasing platelet quartile was a significant predictor of vascular and/or bleeding complications (OR 1.88, p = 0.08; OR 2.48, p = 0.01; OR 2.52, p = 0.01 for the second, third, and fourth quartiles, respectively). In conclusion, the baseline platelet count was a strong and independent risk factor for bleeding and vascular complications in patients undergoing coronary angiography.
Data on the effect of increased platelet counts on vascular and bleeding complications are lacking. Data are lacking even though higher platelet counts have been correlated with adverse outcomes in patients with acute coronary syndromes, ST-segment elevation myocardial infarction treated with primary angioplasty, and in a pooled analysis of 4 randomized trials that examined the value of abciximab in patients undergoing percutaneous coronary intervention (PCI). Thus, the aim of the present study was to examine the relation between the platelet count and risk of bleeding and/or vascular complications during coronary angiography, both to increase the accuracy of risk stratification before invasive procedures and to provide insight into potential mechanisms contributing to the development of complications.
Methods
A retrospective cohort study of consecutive adults undergoing coronary angiography at the University of North Carolina Hospital from July 2, 2006 to June 30, 2008 was performed. Data on vascular and bleeding complications during inpatient hospitalization were collected prospectively as a part of a continuous quality improvement protocol. Vascular site and bleeding complications were defined according to the criteria of the American College of Cardiology CathPCI Registry criteria, a part of the American College of Cardiology National Cardiovascular Disease Registry (ACC-NCDR) a large, ongoing, national registry of diagnostic cardiac catheterization and PCI in the United States (available at: www.ncdr.com/WebNCDR/COMMON/RESEARCH.ASPX ).
Because the platelet count was not normally distributed, quartiles of the platelet count were used as explanatory variables in modeling to reduce the effects of outliers. Age, gender, PCI, baseline serum creatinine, baseline international normalized ratio, and body surface area were considered as candidate covariates. Univariate predictors of the composite outcome of bleeding and vascular complication were determined for all variables in the sample. A multivariate logistic regression model was constructed in which dummy variables were constructed for each platelet count quartile. A second model was constructed that used thrombocytopenia (platelet count <100 × 10 9 /L) as a dichotomous variable to determine the contribution of low platelet counts to the composite end point of bleeding and vascular complications. Multivariate analysis was performed using R2.9.1 software (R Project for Statistical Computing, http://www.r-project.org/ ).
Results
The present study consisted of 3,466 consecutive adult patients undergoing coronary angiography at a single tertiary medical center. Of these patients, 54 were excluded because of incomplete data, leaving a cohort of 3,412 patients. The demographic data are listed in Table 1 . The most common indications for coronary angiography were myocardial infarction in 28% (ST-segment elevation in 9% and non–ST-segment elevation in 19%), angina and/or ischemia on a noninvasive evaluation in 51%, heart failure in 9%, and valvular heart disease in 6%. The artery accessed was femoral, radial, and brachial in 93%, 6.9%, and 0.1% of the patients, respectively. All patients received 325 mg of oral aspirin before the procedure. Hemostasis after sheath removal was obtained by manual compression in >99% of patients.
| Demographic | Total Cohort (n = 3,412) | Complications (n = 143) | Angiography Without PCI (n = 2,220) | Angiography With PCI (n = 1,192) |
|---|---|---|---|---|
| Age (years) | 60 ± 13 | 64 ± 12 ⁎ | 59 ± 12 | 60 ± 12 |
| Women | 42% | 42% | 46% ⁎ | 36% ⁎ |
| Body surface area (m 2 ) | 2.0 ± 0.3 | 1.9 ± 0.2 | 2.0 ± 0.3 | 2.0 ± 0.3 |
| Creatinine (mg/dl) | 1.4 ± 1.6 | 1.4 ± 1.9 | 1.4 ± 1.6 | 1.3 ± 1.4 |
| Hematocrit (%) | 38.3 ± 5.3 | 38.4 ± 5.2 | 38.1 ± 5.3 | 38.7 ± 5.1 |
| Percutaneous coronary intervention | 35% | 45% † | 0% | 100% |
| Protime (seconds) | 12.4 ± 3.0 | 12.3 ± 1.9 | 12.5 ± 3.2 | 12.1 ± 2.8 |
| International normalized ratio | 1.09 ± 0.19 | 1.09 ± 0.13 | 1.10 ± 0.21 | 1.08 ± 0.17 |
| Platelet count (×10 9 /L) | 242 ± 86 | 256 ± 83 ⁎ | 240 ± 88 | 244 ± 83 |
† p <0.0001; statistical analysis not performed on number of PCIs performed in angiography without PCI or angiography with PCI cohorts.
A total of 143 bleeding and vascular complications (4.2%) occurred, including 29 large hematomas (0.85%), 14 pseudoaneurysms (0.41%), 8 retroperitoneal bleeding episodes (0.23%), 6 gastrointestinal bleeding episodes (0.18%), 9 arteriovenous fistulas (0.26%), and 4 vascular occlusions (0.12%). In addition, 73 patients (2.14%) had a 3 g/dl decrease in hemoglobin without an obvious source. No episodes of intracranial or genitourinary hemorrhage occurred. The incidence of bleeding and vascular complications was 4.4% with femoral artery access (140 of 3,170 patients) and 1.3% with radial access (3 of 237). No complications developed in the 4 patients who had had brachial access or in the 1 patient with ulnar access.
The median platelet count in this population was 231 × 10 9 /L (range 187 to 283), with the distribution skewed slightly to the right ( Figure 1 ). Of the 3,412 patients, 41 (1.2%) had thrombocytosis (≥500 × 10 9 /L) and 82 (2.4%) had thrombocytopenia (≤100 × 10 9 /L). The cause of thrombocytosis was reactive in 18 patients (systemic inflammation in 7, iron deficiency anemia in 3, recent surgery in 3, and ankylosing spondylitis, malignancy, gout, POEMS syndrome [polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes], and sickle cell crisis in 1 each), autonomous in 2 patients (both with polycythemia vera), and undiagnosed in 21 patients.
Dividing the patients into quartiles according to the platelet count showed that the risk of complications increased with an increasing platelet count ( Figure 2 ). Compared to the quartile with the lowest platelet count, the odds ratio (OR) for bleeding or vascular complications was 1.41 for the second quartile (95% confidence interval 0.83 to 2.37, p = 0.20), 1.74 for the third quartile (95% confidence interval 1.04 to 2.93, p = 0.04), and 2.04 for the fourth quartile (95% CI 1.21 to 3.44, p = 0.008). Patients in quartile 4 were younger, more likely to be women, and had a lower creatinine than patients in quartile 1. However, no difference was found in the performance of PCI in the different quartiles (PCI was performed in 32%, 37%, 37%, and 34% of patients in quartile 1, 2, 3, and 4, respectively). Variables, including age, gender, body surface area, serum creatinine, fluoroscopy time, international normalized ratio, and PCI, were controlled using a multivariate logistic regression model. In the present analysis, an increasing platelet count, as well as age and PCI, were independent predictors of bleeding or vascular complications ( Figure 3 ).
