The relation between body mass index (BMI) and bleeding after percutaneous coronary intervention (PCI) remains incompletely understood. This study aimed to assess the association between BMI and bleeding and mortality after PCI. The study included 14,178 patients with coronary artery disease treated by PCI. Bleeding within 30 days of PCI was defined using the Bleeding Academic Research Consortium criteria. The primary outcome was 1-year all-cause mortality. BMI quartiles were 14.1 to 24.8 kg/m 2 (first quartile [Q1]), >24.8 to 27.1 kg/m 2 (second quartile [Q2]), >27.1 to 29.8 kg/m 2 (third quartile [Q3]), and >29.8 to 56.3 kg/m 2 (fourth quartile [Q4]). In BMI Q1, Q2, Q3, and Q4, the frequency of bleeding was 13.8%, 10.1%, 10.8%, and 7.7%, respectively (odds ratio [OR] 1.90, 95% confidence interval [CI] 1.63 to 2.23, p <0.001, for Q1 vs Q4). Multiple logistic regression showed that BMI was independently associated with bleeding (adjusted OR 1.05, 95% CI 1.04 to 1.07, p <0.001, for any bleeding; adjusted OR 1.07, 95% CI 1.04 to 1.09, p <0.001, for access site bleeding; and adjusted OR 1.03, 95% CI 1.01 to 1.05, p = 0.039, for non–access site bleeding with all 3 risk estimates calculated per 1 kg/m 2 decrease in BMI). Analysis by sex showed an increase in the frequency of bleeding with the decrease in BMI for women and men (p for trend <0.001 for women and men) with no sex-by-BMI interaction (p = 0.90). The Cox proportional hazards model showed that bleeding (adjusted hazard ratio [HR] 2.17, 95% CI 1.67 to 2.82, p <0.001) and BMI (HR 1.03, 95% CI 1.01 to 1.06, p = 0.048, per 1 kg/m 2 decrease in the BMI) were independently associated with increased risk of 1-year mortality with no bleeding-by-BMI interaction (p = 0.81). In conclusion, BMI is inversely associated with the increased risk of bleeding and mortality after PCI.
Bleeding is a common complication of percutaneous coronary intervention (PCI) that is associated with a poor prognosis. Several studies have investigated the association between body mass index (BMI) and the risk of peri-PCI bleeding. Several characteristics of these studies deserve mentioning. First, many previous studies date back to the 1990s or early 2000s, and they do not reflect the current practice of PCI. Second, in several studies, the association between bleeding and BMI was assessed in the setting of vascular complications of PCI in general, and bleeding was poorly defined. Third, in many studies, the BMI-bleeding association was not risk adjusted. In studies that adjusted for potential confounders, being underweight was no longer associated with a higher risk of post-PCI bleeding compared with normal BMI. Fourth, none of the previous studies addressed the association between BMI and the risk of access site or non–access site bleeding. Fifth, with the exception of 1 study that analyzed the need for blood transfusion in men and women, sex differences of the BMI-bleeding risk association have not been explored. Sixth, although low BMI is suggested to be associated with increased risk of bleeding and death after PCI, bleeding-by-BMI interaction regarding mortality has not been investigated. Finally, the association between BMI and bleeding, defined according to the Bleeding Academic Research Consortium (BARC) criteria which are suggested to be more sensitive than other bleeding criteria, has not been investigated. The aim of this study was threefold: first, to assess the association between BMI and access or non–access site bleeding; second, to perform a sex-based analysis of the BMI-bleeding association; and third, to investigate whether there is a bleeding-by-BMI interaction in predicting the increased risk of mortality after PCI.
Methods
The study included 14,178 patients with stable coronary artery disease (n = 9,033) or non–ST-segment elevation acute coronary syndrome (n = 5,145) who underwent PCI from June 2000 to May 2011. By design, the study represents a retrospective analysis of prospectively collected data. The primary sample included 14,180 patients enrolled in the Intracoronary Stenting and Antithrombotic Regimen trials. The diagnosis of coronary artery disease was based on the clinical criteria and coronary angiography (angiographic documentation of coronary stenosis with ≥50% lumen narrowing in ≥1 major coronary artery or culprit lesions). More detailed inclusion/exclusion criteria are given in a previous publication from our group. Of the 14,180 patients, in 2 of them (both with stable coronary artery disease), information on BMI was not available. Thus, 14,178 patients with available BMI data are included in this study. Written informed consent was obtained in all patients in the setting of primary trials, and each study protocol was approved by the ethics committee in the respective recruitment centers. The study conforms to the Declaration of Helsinki.
Coronary angiography and PCI were performed as per standard practice. Femoral artery was used for vascular access. Before PCI procedure, all patients received 325 to 500 mg of aspirin and 600-mg loading dose of clopidogrel. Peri-PCI antithrombotic/anticoagulant regimen included one of the following options: a combination of unfractionated heparin (70 U/kg of weight) plus glycoprotein IIb/IIIa inhibitor abciximab (0.25 mg/kg of weight, as an intravenous bolus, followed by a continuous intravenous infusion of 0.125 μg/kg/min, but no more than a maximum of 10 μg/min, for 12 hours), unfractionated heparin (100 U/kg of weight), unfractionated heparin (140 U/kg of weight), and bivalirudin (0.75 mg/kg of weight, as an intravenous bolus, followed by an infusion of 1.75 mg/kg of weight per hour for the duration of the procedure). Patient’s weight and height were measured during the hospital course, and the BMI was calculated.
Postprocedural maintenance antithrombotic therapy consisted of 80 to 325 mg/day of aspirin and clopidogrel (75 to 150 mg/day until discharge but for no longer than 3 days, followed by 75 mg per day for ≥1 month in patients who received bare-metal stents or ≥6 months in patients who received drug-eluting stents). Prescription of other medications was left at the discretion of the patient’s physician.
Bleeding events within the first 30 days after PCI were defined using the BARC criteria. Class 5 bleeding was not analyzed as a distinct class (because the primary outcome of this investigation was mortality). Bleeding episodes potentially belonging to class 5 were allocated to other classes according to their initial assessment. Bleeding criteria were retrospectively applied. According to location, bleeding was classified as access-site (bleeding or hematoma arising at the site of vessel puncture or that spread from the access site to adjacent tissues) and non–access site (bleeding events occurring in other body parts and not related to the access site) bleeding. Bleeding episodes whose location remained obscure after imaging tests were classified as non–access site bleeding.
The primary outcome was all-cause mortality at 1 year after PCI. Nonfatal myocardial infarction or definite stent thrombosis at 1 year were secondary outcomes. Information on mortality was obtained from the hospital records, death certificates, insurance companies, and registration of address office and the relatives of the patient or referring physicians in the setting of follow-up protocols. Nonfatal myocardial infarction was diagnosed in case of development of new abnormal Q waves in ≥2 contiguous precordial leads or in ≥2 adjacent extremity leads or an elevation of the creatine kinase-myocardial band activity of twofold the upper limit of normal or more. For the first 48 hours after the PCI procedure and for patients who underwent coronary artery bypass surgery, a creatine kinase-myocardial band elevation of ≥3-fold and ≥10-fold the upper limit of normal, respectively, were required for the diagnosis of myocardial infarction. Definite stent thrombosis was defined based on the Academic Research Consortium criteria.
The follow-up protocol included a telephone communication at 1 month, 6 months, and 1 year after index PCI. If patients had complaints at these time points or at any time during the follow-up, they underwent a thorough clinical and laboratory examination. Personnel who performed the follow-up were blinded to clinical characteristics of the patients.
Data are presented as median with 25th to 75th percentiles, proportions (%), or Kaplan-Meier estimates (%). BMI quartiles and deciles were calculated. The distribution pattern of the continuous data was tested with the Kolmogorov-Smirnov test. Continuous data with skewed distribution were compared with the Kruskal-Wallis rank-sum test. Categorical data were compared with the chi-square test. The survival analysis was performed with the Kaplan-Meier method and the log-rank test. The association between BMI and the risk of bleeding was tested using the multivariable logistic regression. All variables of Table 1 , but the type of antithrombotic therapy (because of collinearity with bleeding) plus sex-by-BMI interaction term were entered into the model. The therapy-by-BMI interaction in predicting the risk of bleeding was tested by repeating the logistic regression model after inclusion of antithrombotic therapy alongside other variables. The Cox proportional hazards model was used to evidence independent associates of increased risk of 1-year mortality. All variables of Table 1 , but the type of antithrombotic therapy (because of collinearity with bleeding) plus bleeding-by-BMI interaction term, were entered into the model. To account for the type of the study, a random effect for study was included in the models alongside the pooled data. The proportional hazards assumption was checked using the method by Grambsch and Therneau. All analyses were performed using the R statistical package. A 2-sided p-value <0.05 was considered to indicate statistical significance.
| Characteristic | Body mass index quartiles | P value | |||
|---|---|---|---|---|---|
| 1 (n=3592) | 2 (n=3472) | 3 (n=3663) | 4 (n=3451) | ||
| Age (years) | 69 [61.5; 77] | 68 [61; 75] | 67 [60; 74] | 64 [58; 72] | <0.001 |
| Women | 1147 (32%) | 664 (19%) | 709 (19%) | 832 (24%) | <0.001 |
| Diabetes mellitus | 805 (22%) | 893 (26%) | 1144 (31%) | 1485 (43%) | <0.001 |
| Requiring insulin | 188 (5%) | 240 (7%) | 296 (8%) | 479 (14%) | <0.001 |
| Body mass index (kg/m 2 ) | 23 [22; 24] | 26 [25; 27] | 28 [27; 29] | 32 [31; 35] | <0.001 |
| Arterial hypertension | 2592 (72%) | 2633 (76%) | 2934 (80%) | 2905 (84%) | <0.001 |
| Total cholesterol (≥220 mg/dl) | 2243 (62%) | 2340 (67%) | 2577 (70%) | 2498 (72%) | <0.001 |
| Current smoker | 714 (20%) | 545 (16%) | 606 (16%) | 591 (17%) | <0.001 |
| Prior myocardial infarction | 1093 (30%) | 968 (28%) | 1130 (31%) | 1055 (31%) | 0.02 |
| Prior coronary artery bypass surgery | 367 (10%) | 425 (12%) | 429 (12%) | 390 (11%) | 0.05 |
| Acute coronary syndrome | 1324 (37%) | 1271 (37%) | 1298 (35%) | 1252 (36%) | 0.61 |
| Elevated cardiac troponin | 718 (20%) | 665 (19%) | 701 (19%) | 686 (20%) | 0.70 |
| Creatinine clearance (ml/min) | 68 [53; 86] | 79 [62; 97] | 87 [69; 107] | 102 [80; 126] | <0.001 |
| C-reactive protein (mg/L) | 1.3 [0.0; 5.1] | 1.4 [0.0; 5.2] | 1.6 [0.0; 5.0] | 2.7 [0.7; 7.0] | <0.001 |
| Number of narrowed coronary arteries | 0.13 | ||||
| 1 | 743 (21%) | 653 (19%) | 742 (20%) | 719 (21%) | |
| 2 | 949 (26%) | 1001 (29%) | 1040 (28%) | 944 (27%) | |
| 3 | 1900 (53%) | 1818 (52%) | 1881 (52%) | 1788 (52%) | |
| Multivessel coronary disease | 2849 (79%) | 2819 (81%) | 2921 (80%) | 2732 (79%) | 0.14 |
| Left ventricular ejection fraction (%) | 59 [49; 64] | 59 [50; 64] | 60 [40; 65] | 58 [50; 65] | 0.04 |
| Platelet count (x 109/L) | 221 [185; 264] | 214 [183; 251] | 215 [181; 253] | 215 [182; 254] | <0.001 |
| Periprocedural anticoagulant therapy | 0.03 | ||||
| Unfractionated heparin (100 U/kg) | 594 (17%) | 587 (17%) | 668 (18%) | 656 (19%) | |
| Unfractionated heparin (140 U/kg ) | 1273 (35%) | 1245 (36%) | 1274 (35%) | 1180 (34%) | |
| Bivalirudin | 764 (21%) | 790 (23%) | 807 (22%) | 788 (23%) | |
| Abciximab plus UFH (70 U/kg) | 961 (27%) | 850 (24%) | 914 (25%) | 827 (24%) | |
Results
The study included 14,178 patients. BMI quartiles were 14.1 to 24.8 kg/m 2 (first quartile [Q1]), >24.8 to 27.1 kg/m 2 (second quartile [Q2]), >27.1 to 29.8 kg/m 2 (third quartile [Q3]), and >29.8 to 56.3 kg/m 2 (fourth quartile [Q4]). Baseline data according to BMI quartiles are listed in Table 1 . Overall, 13,043 patients (92.0%) received coronary stenting (3,306 patients [92%] in the Q1, 3,197 patients [92%] in the Q2, 3,372 patients [92%] in the Q3, and 3,168 patients [92%] in the Q4; p = 0.90). Drug-eluting stents were used in 8,752 patients (62%): 2,153 patients (60%) in the Q1, 2,145 patients (62%) in the Q2, 2,264 patients (62%) in the Q3, and 2,190 patients (63.5%) in the Q4 (p = 0.026).
There were 1,510 bleeding complications within the first 30 days after PCI: 495 bleeding events (13.8%) in the BMI Q1, 351 events (10.1%) in the BMI Q2, 397 events (10.8%) in the BMI Q3, and 267 events (7.7%) in the BMI Q4 (odds ratio [OR] 1.90, 95% confidence interval [CI] 1.63 to 2.23, p <0.001, for Q1 vs Q4). With the exception of BARC bleeding class 4 (coronary artery bypass-related bleeding), for all other BARC bleeding classes, the frequency of bleeding was reduced from lower to upper BMI quartiles. Of note, the frequency of bleeding class ≥2 decreased from 8.7% in patients in the BMI Q1 to 4.1% in patients in the BMI Q4 (p <0.001; Table 2 ).
| BARC bleeding class | Body mass index quartiles | P value | |||
|---|---|---|---|---|---|
| 1 (n=3592) | 2 (n=3472) | 3 (n=3663) | 4 (n=3451) | ||
| 1 | 184 (5.1%) | 139 (4.0%) | 184 (5.1%) | 125 (3.6%) | <0.001 |
| 2 | 66 (1.8%) | 60 (1.7%) | 58 (1.6%) | 46 (1.3%) | |
| 3a | 181 (5.0%) | 97 (2.8%) | 123 (3.4%) | 67 1.9%) | |
| 3b | 56 (1.6%) | 47 (1.4%) | 31 (0.8%) | 25 (0.7%) | |
| 3c | 5 (0.14%) | 5 (0.14%) | 1 (0.03%) | 1 (0.03%) | |
| 4 | 3 (0.08%) | 3 (0.09%) | 0 (0.0%) | 3 (0.9%) | |
| Any bleeding | 495 (13.8%) | 351 (10.1%) | 397 (10.8%) | 267 (7.7%) | <0.001 |
| Access-site bleeding | 311 (8.7%) | 196 (5.6%) | 243 (6.6%) | 155 (4.5%) | <0.001 |
| Non-access site bleeding | 184 (5.1%) | 155 (4.5%) | 154 (4.2%) | 112 (3.2%) | 0.001 |
| Bleeding class ≥2 | 311 (8.7%) | 212 (6.1%) | 213 (5.7%) | 142 (4.1%) | <0.001 |
Analysis by bleeding location showed that the frequency and the risk of access site (OR 2.01, 95% CI 1.65 to 2.46, p <0.001, for Q1 vs Q4) and non–access site (OR 1.60, 95% CI 1.27 to 2.05, p <0.001, for Q1 vs Q4) increased with the decrease in the BMI quartiles ( Table 2 ).
To obtain a more detailed view of the relation between the BMI (particularly the low BMI) and the risk of bleeding, a decile-based analysis was performed. The analysis showed that the frequency of bleeding decreased with the increase in the BMI decile scale (p for trend <0.001 for any bleeding, access site bleeding, and non–access site bleeding; Table 3 and Figure 1 ).
| BMI decile | BMI cutoff (kg/m 2 ) | Number | Any bleeding | Access site bleeding | Non-access site bleeding |
|---|---|---|---|---|---|
| 1st | 14.2 – <22.9 | 1423 | 217 (15.2%) | 139 (9.8%) | 78 (5.4%) |
| 2nd | 22.9 – <24.2 | 1416 | 192 (13.6%) | 117 (8.2%) | 75 (5.4%) |
| 3rd | 24.2 – <25.3 | 1418 | 158 (11.1%) | 80 (5.6%) | 78 (5.5%) |
| 4th | 25.3 – <26.2 | 1417 | 147 (10.4%) | 99 (7.0%) | 48 (3.4%) |
| 5th | 26.2 – <27.1 | 1422 | 137 (9.6%) | 76 (5.3%) | 61 (4.3%) |
| 6th | 27.1 – <28.0 | 1411 | 154 (10.9%) | 109 (7.7%) | 45 (3.2%) |
| 7th | 28.0 – <29.1 | 1453 | 159 (10.9%) | 84 (5.8%) | 75 (5.1%) |
| 8th | 29.1 – < 30.5 | 1387 | 122 (8.8%) | 71 (5.1%) | 51 (3.7%) |
| 9th | 30.5 – < 32.8 | 1414 | 113 (8.0%) | 60 (4.2%) | 53 (3.8%) |
| 10th | 32.8 – 56.3 | 1417 | 111 (7.8%) | 70 (4.9%) | 41 (2.9%) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
