The present study is a cross-sectional, retrospective, analytical investigation of patients submitted to surgical treatment of heart valve diseases in a cardiology reference center, intending to evaluate risk factors associated with postoperative bleeding in this population.

The population is constituted of patients submitted to heart valve surgery between January 2021 and December 2022. The immediate postoperative was defined as the first 12 hours after ICU admission, and early postoperative was defined as the period up to 48 hours.

Inclusion criteria were patients with acquired or congenital valve diseases; reoperations, combined procedures involving more than one valve; and patients aged equal to or above 18 years.

Exclusion criteria were combined valve surgery with coronary artery bypass graft or aorta, patients with incomplete information from the database, and those who refused to sign the informed consent form or were not recorded in the database.

Predictors of increased bleeding risk were defined as variables, according to the literature. The preoperative variables considered were age, weight, height, body surface area, gender, reoperation, systemic hypertension, coagulopathy and thrombocytopenia, diabetes mellitus, chronic kidney disease or acute kidney injury, use of anticoagulants or antiplatelets drugs, serum hemoglobin, and creatine levels. Intraoperative variables were total surgery, cardiopulmonary bypass and aortic cross-clamp durations, temperature during CBP, nature of the procedure (combined or not type of surgery), heparin dosage, activated clotting time (ACT) after protamine sulfate infusion, and the use of hemocomponents and autotransfusion system. Postoperative variables included the necessity of surgical hemostasis review, volume of blood drained hourly through a mediastinal drain in the first 6 hours after ICU admission, cumulative volume in the first 12, 24, and 48 hours; occurrence of intracranial hemorrhage, use of blood derivatives and their quantity, administration of antifibrinolytics, presence of metabolic acidosis, death, and death due to bleeding.

The data were systematically tabulated using the RedCap (Research Electronic Data Capture) software,²³ utilizing the database from the institute's valvular heart disease sector. An analysis of the risk factors was conducted to identify those most significantly associated with increased postoperative bleeding. The BARC¹³ and Bojar¹⁴ classifications were applied to differentiate between groups. For the BARC¹³ classification, criteria from item 1 were used to indicate the absence of bleeding, while item 04 from Figure 1 was used to identify cases of increased bleeding. This includes reoperation for hemostasis revision, intracranial hemorrhage within 48 hours, and blood drainage of 2000ml within the first 24 hours. In the Bojar classification,¹⁴ patients were assessed based on the volume of blood drained at each postoperative time interval, as detailed in Table 1.

The sample description was performed using absolute and relative frequency measures for categorical variables, and central tendency measures (mean and median) as well as dispersion measures (standard deviation, interquartile ranges) for continuous variables. The impact of demographic, intraoperative, and surgical variables on major bleeding was assessed using logistic regression models. Initially, univariate analyses were conducted. Variables that demonstrated statistical significance were subsequently included in the multivariate analysis, considering collinearity among variables, statistical significance, and theoretical plausibility for inclusion in the model. Statistical significance was determined at a threshold of p < 0.05, with a 95% confidence interval.

Based on the final model, a nomogram was developed, using the R software. This nomogram features a series of scales that represent patient characteristics (variables included in the multivariate analysis). Each scale assigns a score, which, when summed, yields a final score for each patient. This final score is utilized to evaluate the patient's risk of bleeding.

External validation using an alternative database is not within the scope of this study. The analyses were performed utilizing the R software, version 4.2.1.

The project was approved by the Research Ethics Committee of the Dante Pazzanese Institute of Cardiology. Number of CAAE: 59774422.9.0000.5462. Opinion number: 5.497.323. CEP number: 5268. The Informed Consent Form was waived by the committee.

From a total of 590 patients submitted to surgical treatment of valvular heart disease, 65 were excluded, remaining a cohort of 525 patients to be included in the analyses. The mean age of the population was 56 years old, with 57.3% being female. The average weight was 73 kg and the mean body surface area was 1.82 m². The most prevalent valve pathology was mitral regurgitation (51.2%), followed by mitral (37.7%) and aortic (33.9%) stenosis, and tricuspid regurgitation present in 19.6% of all cases. Reoperations represented 30.5% of total procedures.

Systemic arterial hypertension had a prevalence of 51%, while 16% of patients were diabetics, 9.7% had either chronic kidney disease or acute kidney injury and 18.7% had thrombocytopenia before surgery. Mean serum creatine and hemoglobin were 1.11g/dL and 12.87 g/dL, respectively, average activated partial thromboplastin time was 28.04 seconds, with a mean fibrinogen of 352 mg/dL. Out of the 525 patients, 5.5% used direct oral anticoagulants, and 24.6% used warfarin in the preoperative period.

The average surgery duration was 309.5 minutes, 123.8, and 86.91 minutes of CBP and aortic cross-clamp time, respectively. The mitral valve was addressed in 67.4% of all procedures, the aortic valve in 43.9%, and the tricuspid and pulmonary valves were involved in 15.3% and 0.6% of surgeries. During the intraoperative period, 42.3% of patients received red blood cell concentrate (RBC), with a mean quantity of 1.94 units. Blood derivatives transfusion was necessary for 31.6% of procedures, and 33.3% required the use of an autotransfusion system.

Intracranial hemorrhage in the first 48 hours had an occurrence of 0.2%, and 4.8% of patients required a surgical approach to the increased bleeding, being a thoracic diffuse unspecific hemorrhage as the cause. Patients presented a cumulative mean drained blood volume of 257 ml in 12 hours, 380 ml in 24 hours, and 475 ml in 48 hours. In the postoperative period, around 11.8% of patients required RBC concentrate transfusion, averaging 2.56 units of RBC concentrate per patient. With an overall mortality rate of 8.8%, 10.9% of those were attributable to bleeding.

Using the BARC definition, a prevalence of major bleeding was identified as 5.3% (n=28), whereas the Bojar classification identified 40 cases (7.6%). A total of 22 patients met both the Bojar and BARC criteria to qualify as postoperative increased bleeding. The overall prevalence according to both definitions was 8.8% (n=46), with the BARC classification accounting for 61% of patients, and the Bojar classification accounting for 87%. Table 2 summarizes the variables according to the presence of major bleeding.

Among the preoperative variables analyzed, those that achieved statistical significance were tricuspid insufficiency with an odds ratio (OR) of 3.31 (95% confidence interval [CI] 1.74 - 6.20, p < 0.001), the presence of chronic kidney disease or acute kidney injury with an OR of 2.97 (95% CI 1.31 - 6.23, p = 0.006), preoperative hemoglobin with an OR of 0.73 (95% CI 0.62 - 0.85, p < 0.001), and reoperation with an OR of 2.5 (95% CI 1.35 - 4.62, p = 0.003).

The analyzed surgical variables included the median duration of the procedure, which was 327.5 minutes, with an OR of 1.05 (95% CI 1.03 - 1.10, p < 0.001); CBP time of 130 minutes, OR 1.12 (95% CI 1.06 - 1.18, p < 0.001); aortic cross-clamp of 90 minutes, OR 1.13 (95% CI 1.05 - 1.21, p < 0.001. The effects were evaluated considering 10-minute increments for each variable. For example, each additional 10 minutes of cardiopulmonary bypass is associated with a 12% increase in the likelihood of major bleeding.

The involvement of two valves requiring repair/replacement during the surgical procedure was significantly associated with an increased risk of bleeding, with an OR of 2.34 (95% CI 1.17 - 4.53, p = 0.013). A triple valve approach showed a heightened risk with an OR of 3.7 (95% CI 1.00 - 11.1, p – 0.028). Additionally, different valve combinations demonstrated significant risk increases, such as mitral valve replacement and tricuspid valve repair, OR 3.4 (95% CI 1.33 - 8.66, p = 0.01), and the combined approach of mitral, aortic, and tricuspid interventions, OR 3,79 (95% CI 0.96 0 12.8, p = 0.04).

RBC concentrate transfusion demonstrated a statistically significant association with major bleeding, with an OR of 2.8 (95% CI 1.51 - 5.4, p = 0.001). Additionally, the administration of two or more units increased the likelihood of bleeding, OR 1.52 (CI 95% 1.06 - 2.16, p = 0.021). Furthermore, the use of an autotransfusion system was also associated with a higher risk of bleeding, OR 2.61 (95% CI 1.42 - 4.85, p = 0.002).

The variables that retained statistical significance in the multivariate model were preoperative hemoglobin, cardiopulmonary bypass time, and the presence of tricuspid regurgitation, as detailed in Table 3. Tables 4 and 5 evaluate other relevant variables, such as chronic kidney disease, reoperation, and the use of RBC concentrate. Despite their clinical relevance, these variables did not achieve statistical significance in the multivariate analysis.

Figures 2 and 3 present the nomograms related to the analyses shown in Tables 3 and 5, which provide a scoring system to estimate the risk of postoperative bleeding.

This study is a single-center, retrospective analysis based on a database, which may limit the representativeness of the population. Additionally, external validation was not conducted, though plans for such validation are anticipated in future research.