Impact of aspirin on bleeding and blood product usage in off‐pump and on‐pump coronary artery bypass graft surgery

Abstract Major bleeding is linked to poorer outcomes following cardiac surgery. Current guidelines recommend continuation of aspirin prior to coronary artery by‐pass graft (CABG) but the effect of continuing aspirin in patients with prior indication for aspirin, in particular during off‐pump CABG (OPCABG), has not been systematically assessed. In this study, we analysed the effect of continuing aspirin prior to OPCABG and on‐pump CABG with respect to bleeding and blood product usage. We compared propensity‐matched cohorts of patients who continued aspirin until the day of OPCABG or CABG to controls (no antiplatelet) and to patients discontinuing aspirin 5–7 days prior. Length of hospital stay, 30‐day mortality and thromboembolism rates were similar for both OPCABG and CABG. During OPCABG, aspirin‐continued patients received more intraoperative red cell units compared to controls without difference in bleeding. Aspirin‐continued patients received more blood products perioperatively and bled more than aspirin‐discontinued patients undergoing OPCABG. The only difference during CABG was a small increase in the volume of cells salvaged among aspirin‐continued patients compared to controls. Current guidelines on the continuation of aspirin prior to CABG and OPCABG are safe. Continuation of aspirin prior to OPCABG may result in more bleeding and blood product usage.

to surgery [6] as antiplatelet therapy is a patient-related factor that increases risk should inadequate perioperative discontinuation occur [4]. Meta-analyses however, found no association between bleeding (estimated via chest drain volume) and aspirin doses of < 160 mg taken on the day of CABG [7]. In 2016, a large randomised controlled trial, with the use of tranexamic acid (antifibrinolytic), demonstrated no difference between patients who received aspirin (100 mg) or placebo on the day of CABG in respect of thromboembolic events, re-operation rates and bleeding [6]. Following this study, it became a standard practice that aspirin be continued prior to cardiac surgeries with discontinuation of therapy reserved for those with high bleeding risk, very low thrombotic risk or refusing transfusion [8]. Neither Myles et al. [6] nor Hastings et al. [7], however, included patients with prior indication for aspirin therapy [6,7]. Both studies also lacked data on continuing aspirin prior to off-pump CABG (OPCABG): only 3% of cases in Myles et al. were OPCABG [6] with no OPCABG sub-group testing by Hastings et al. due to small numbers [7]. Other studies have reported the effect of continuing aspirin until the day of OPCABG, but these too have important limitations including small sample sizes, ambiguous or lack of comparator groups and lack of information on bleeding and blood product use [9][10][11][12].
This study addresses these limitations [9][10][11][12] and better assesses the true in vivo effects of continuing aspirin having eliminated the confounding effects of cardiopulmonary by-pass (CPB) including high-dose heparin, loss of large von-Willebrand Factor (VWF) multimers [13], hypothermia-induced coagulopathy [14][15][16][17] and protamine-induced platelet dysfunction [18] via OPCABG analyses. As all included aspirin patients had an indication for therapy prior to surgery, our study population better represents patients taking aspirin prior to CABG as patients included in Myles et al. [6] and Hastings et al. [7] had no history of aspirin use prior to surgery. Comparisons between patients continuing and discontinuing aspirin also enabled analyses of the clinical decision to discontinue aspirin prior to both CABG and OPCABG since the change in British Society of Haematology (BSH) guidelines [8].
The objective of this study was to determine the effect of continuing aspirin on bleeding and blood product use in adult patients (> 18 years) undergoing OPCABG or CABG. We therefore compared propensity-matched cohorts of patients on long-term aspirin monotherapy who continued therapy until the day of surgery to controls (not on antiplatelet or anticoagulant) and to patients who discontinued therapy 5-7 days prior, with separate analyses for CABG and OPCABG.
Primary outcome measures were the intraoperative (during Secondary outcome measures were the initial location of postoperative care, length of hospital stay (LOHS), thromboembolic (TE) rate and 30-day mortality. TE is defined as objectively confirmed venous or arterial thrombosis within 30 days of the surgery. All patients had clinic review at 6 weeks following the surgery; therefore, even if the thrombosis was diagnosed in another hospital, this information was available.

PATIENTS AND METHODS
This was a single-centre, retrospective study in a major tertiary refer- Data were cleaned for outliers, and patients with unresolved data quality issues were identified. In both instances, the EPR were checked to correct any errors in the data. Patients with missing data were identified with attempts made to fill in missing data using EPR before exclusion.

Patients
Patients on long-term aspirin therapy (continued and discontinued 5-7 days prior) comprised the study groups. Patients not on anticoagulant or with no antiplatelet effect at the time of surgery (either because they had never taken it or had discontinued > 7 days prior) comprised the control group.

Inclusion criteria
Patients ≥ 18 years old undergoing OPCABG or CABG alone from 1 March 2016 to 31 December 2018, with full blood count within 14 days prior to the surgery.

Exclusion criteria
The

Statistical analysis
Nearest-neighbour propensity matching (for demographics and comorbidities) was used to account for differences in baseline patient characteristics. Independent variables used in the propensity matching (  102 aspirin-discontinued). Minor imbalances (standard mean difference > 0.1) were seen between OPCABG cohorts (Table 1), but differences seen in any of these characteristics did not reach significance (Table 2). Both cohorts for CABG analyses were well-matched (Tables 1   and 2).

Off-pump coronary artery bypass graft
Compared to controls, patients who continued aspirin had a lower percentage of elective cases with consultants being the lead surgeon more frequently (Table 2). Patients who continued aspirin also had higher EuroSCOREs and lower platelet counts than controls. Aspirincontinued patients had higher pre-op activated partial thromboplastin time (APTT), mean 48-h prothrombin time (PT) and APTT and 24-h fibrinogen levels than both controls and aspirin-discontinued patients (Table 2). Operating times were longer among aspirin-continued patients with a lower percentage of elective cases in comparison to discontinued patients ( Table 2).

Indicators of bleeding
No differences were seen between patients who continued aspirin prior to OPCABG and controls with respect to indicators of bleeding.
Patients who continued aspirin bled more than aspirin-discontinued patients with a greater drop in Hb and PCV, a larger 24-h drain volume and an increased volume of cells salvaged intraoperatively (Table 3).

Intraoperative blood product administration
Aspirin-continued patients received more units of PRC and were more likely to receive PRCs during surgery than controls ( Figure 2). Compared to discontinued patients, aspirin-continued patients received more units of cryoprecipitate, FFP, PRC and platelets and a higher F I G U R E 1 Flow chart of the inclusion and exclusion of the study participants volume of salvaged cells than those discontinuing aspirin intraoperatively. Aspirin-continued patients were also more likely to receive each product during surgery ( Figure 3).

Postoperative blood product administration
There were no differences in postoperative blood product administration for either OPCABG comparison.

Perioperative blood product administration
Despite differences intraoperatively, there were no overall differences in blood product administration during the perioperative period between controls and those who continued aspirin prior to OPCABG.
However, in comparison to discontinued patients, aspirin-continued patients received more units of platelets and a higher volume of salvaged cells with a higher percentage of patients receiving each of these products ( Figure 4).

Postoperative outcomes
There were no differences in the initial location of postoperative care, LOHS, 30-day mortality or TE rate for either OPCABG comparison (Table 4).

Control versus aspirin-continued
Baseline characteristics following propensity matching for controls and aspirin-continued patients are summarised in Table S1. Aspirin- and APTT (p = 0.001) and mean 24-h fibrinogen levels (p < 0.001) were also higher in those who continued aspirin. The only difference in bleeding was a small increase in cells salvaged intraoperatively among controls (p = 0.01; Table S2). Intraoperative, postoperative and perioperative blood product administration was similar between controls and aspirin-continued patients. There were no differences in the initial location of postoperative care, LOHS, 30-day mortality or TE rate between controls and those who continued aspirin prior to CABG (Table S3).

3.2.2
Aspirin-continued versus discontinued   S2). During the perioperative period, only the volume of salvaged cells returned differed with a higher volume returned to discontinued patients (p = 0.005) who were also more likely to receive salvaged cells (p = 0.002; Figure S1). There were no differences in location of postoperative care, LOHS, 30-day mortality or TE rate (Table S3).

DISCUSSION
Since the change in aspirin guidelines prior to cardiac surgery in 2016 [8], this is the first study to compare the effects of continuing aspirin prior to OPCABG and CABG to controls and to patients that discontinued aspirin 5-7 days prior to surgery. By analysing OPCABG patients separately, we have addressed limitations of notable studies lacking sufficient OPCABG analyses [6,7] and better elicit the true in vivo effects of continuing aspirin with respect to bleeding and blood product usage by eliminating the confounding effects of CPB [6,7,[13][14][15][16][17][18]. As all included aspirin patients had an indication for monotherapy and through comparisons between aspirin-continued and discontinued patients, we could also analyse the clinical decision to discontinue aspirin since it became a practice to continue aspirin prior to surgery [8]. Such comparisons will better inform surgeons about the effects of continuing or discontinuing aspirin prior to OPCABG or CABG.
The main finding was the safety of continuing aspirin until the day of OPCABG or CABG as judged by equivalent LOHS, 30-day mortality and TE rates to both controls and discontinued comparators. Because this is a retrospective study and so unrandomised, it is not possible to match all confounding risk factors. However, these findings suggest that the clinicians' application of the 2016 criteria for discontinuation (very high bleeding risk, a very low thrombosis risk or declining transfusion) [8] prior to OPCABG or CABG are safe.
An increase in PRC usage intraoperatively during OPCABG was observed among patients who continued aspirin in comparison to controls. However, when looking at only the patients that received blood products, the mean number of units given to each group was similar Patients who continued aspirin prior to OPCABG also received more products than discontinued patients. This was accompanied by marginal, yet significant increases in bleeding in the continued group. These findings suggest that when confounding effects of CPB [13][14][15][16][17][18] are removed and in vivo effects of aspirin are more noticeable, continuing aspirin prior to OPCABG may increase the requirement for blood products. These differences are more pronounced in comparison to aspirin-discontinued patients than in comparison to controls given the greater intensity and frequency of blood product usage and increased bleeding not seen in control comparisons. Such findings in OPCABG contradict published works [9][10][11][12], but these studies have significant limitations including: small study sizes (< 25 per group) [9], comparator groups being confounded by patients taking clopidogrel and warfarin within < 7 days of surgery [9], ambiguous group definitions with it being unclear if the comparators were acting more like an aspirin discontinuation group (stopping 5-7 days prior) or a control group (stopping antiplatelet > 7 days prior) [10,11], failure to assess administration of platelets or cryoprecipitate, intraoperative product administration or changes in haematological parameters following surgery [10], no measurement of mean differences in product administration [11], lack of outcome measures assessing bleeding or blood product usage [12] and lack of comparison to previous guideline discontinuation protocols (stopping 5-7 days prior) [12]. We believe our study includes a more robust assessment of bleeding and blood product usage and better defines comparator groups therefore better describing the effects of continuing aspirin prior to OPCABG. Our comparisons also suggest that clinical decisions to discontinue aspirin were not associated with an increase in thromboembolic events. The small number of events seen, however, means further study is required to better understand the balance between bleeding and TE when deciding to discontinue aspirin prior to OPCABG.
Although aspirin-continued patients had prolonged mean 48-h PT and APTT, compared to controls, they were within normal ranges (Table 2). It is therefore unlikely that clotting times impacted the observed results. Not surprisingly, control patients were more likely to undergo an elective OPCABG and were less likely to have a consultant as the lead surgeon. The in vivo effects of these differences are unknown and are a limitation of the retrospective nature of this study. The differences seen between aspirin-continued and discontinued patients during OPCABG with respect to operation duration and percentage of elective cases may also have confounded results. However, it is possible that the prolonged operation duration was secondary to increased bleeding that, despite increased blood product administration intraoperatively, was still significantly higher in the aspirincontinued group.
We report no clinically significant differences in bleeding or blood product usage between controls and aspirin-continued patients undergoing CABG as the difference between the two groups is very small.
These results support previous works [6,7] but add to the literature as we can now suggest that the intensity of blood product transfusion, not just the likelihood of transfusion [6], is similar between controls and those continuing aspirin prior to CABG. We can also now suggest that such conclusions can be applied to patients on long-term aspirin who continue therapy prior to CABG. As with the OPCABG comparison between controls and aspirin-continued patients, differences were seen in clotting times, percentage of elective cases and lead surgeon grade despite propensity matching with the same conclusions about the effects of these differences applying here.
Surprisingly, patients who discontinued aspirin prior to CABG bled more than those who continued aspirin. Differences in clotting times were observed but with higher PT and APTT in the aspirin-continued group, the difference were unlikely to be significant. Operation duration was longer in those discontinuing aspirin prior to CABG (20 min longer), but the percentage of elective cases was higher (87.3% vs. 68.6%). It is possible, however, that the increased operation time may have resulted from the increased bleeding in the discontinuation group.
With TE and bleeding risk factors being well-matched at baseline and similar transfusion rates between groups (refusing of transfusion being a possible indication for discontinuation and a confounding factor), we conclude that other unidentified factors were responsible for the increased bleeding seen.
The main limitations of this study are its retrospective, observational nature and lack of pre-specified transfusion criteria and power calculations for the outcomes. Clinician bias could therefore not be excluded as an explanation for differences seen although professionals at the centre have great experience of haemostatic practices during high bleeding risk procedures, and so we anticipated minimal bias.
However, the bias airing from the decision to continue or discontinue aspirin by the clinician cannot be excluded. Propensity matching analyses were performed to minimise differences in patient cohorts and better facilitate comparison. Whilst these were successful in significantly reducing differences between groups, some differences were still present. These differences in patient characteristics were small and believed to have minimal impact on results; a confounding effect could not be completely excluded.

CONCLUSION
Current BSH guidelines and practices at the study centre are safe and effective, including the decision to discontinue aspirin 5-7 days prior to OPCABG. However, given low TE rates, this study was likely underpowered to definitively comment on TE. Any increase in blood product administration associated with continuing aspirin monotherapy prior to OPCAB, if required, is small and any increase in bleeding is not associated with adverse outcomes.