Risk factors for postoperative bleeding following breast cancer surgery: a systematic review and meta-analysis Authors: Zhi Yang, Mei Qing, Yina Zhang, Jian Wu Published: 2025 ABSTRACT ObjectiveSurgical operation serves as a primary approach for addressing cancer of the breast, but it also has potential complications, including hemorrhaging following surgery. Identifying the risk variables associated with this problem may assist individuals in stratifying and optimizing their risks. A meta-analysis and comprehensive review was conducted to pinpoint the risk variables associated with postoperative hemorrhaging in breast carcinoma.MethodsWe performed a systematic review and searched through Cochrane Library, PubMed, WoS, Embase, China National Knowledge Infrastructure, and Wanfang. The investigation spanned from the creation of the records until May 2025. A meta-analysis was performed to investigate the risk variables associated with postoperative hemorrhaging in breast carcinoma patients.Results8 studies were included, comprising 82,012 patients. A total of 10 risk factors were analyzed. They are mastectomy and axillary lymph node dissection, antithrombotic medication, age ≥ 60 years old, hypertension, and diabetes, chronic pulmonary illness, congestive heart failure, reconstruction and neoadjuvant chemotherapy. These results indicated that among the multiple risk factors included in the analysis, 3 had statistically significant effects on postoperative bleeding following breast cancer surgery: specifically, mastectomy (OR = 1.93,95%CI:1.20–3.09), hypertension(OR = 1.23,95%CI:1.05–1.44), CHF(OR = 2.18,95%CI:1.54–3.08).ConclusionBreast cancer risk factors for postoperative hemorrhage include mastectomy, hypertension and CHF. Understanding the risk factors closely related to postoperative bleeding in breast cancer is helpful for identifying high-risk patients with postoperative bleeding, who can benefit from enhanced prevention and treatment. In the future, more and larger sample size studies are needed to continue to investigate additional indicators of risk for hemorrhaging following surgery in breast carcinoma. Introduction The "2022 Global Cancer Report" published by the IARC indicates an occurrence of almost 2.3 million new instances of carcinoma of the breast, exceeding cancer of the lung and making it the most common cancer worldwide [1]. Surgical operations are the main methods for treating breast cancer, whether it is breast-conserving surgery (BCS) or mastectomy [2]. Although breast cancer surgery is known to be relatively safe, several complications can occur soon after surgery: for example, bleeding, wound infections, and seroma [3]. Data from literature indicate that the occurrence of hemorrhaging following breast carcinoma surgery varies between 2 and 10% [4–10]. If the bleeding is severe, an unplanned operation needs to be carried out immediately to control the bleeding to prevent endangering the patient’s life. Of course, such complications will prolong the hospital stay, increase the treatment costs, and simultaneously reduce the patient’s quality of life [4–6, 10].Therefore, identifying the relevant risk factors for early postoperative bleeding is of crucial significance for optimizing controllable risk factors before the operation and conducting individualized risk stratification (thereby enabling more intensive monitoring during the early postoperative period). The risk variables associated with hemorrhaging after breast carcinoma surgery are not well understood. A prior investigation utilizing a Danish record encompassed 19,919 individuals who had surgery for breast carcinoma. It examined the relationship between postoperative bleeding that necessitated re-operation in individuals > 80 years of age, those that went through mastectomy, and the administration of steroids [5]. An investigation conducted in the US involving 7097 individuals who had breast carcinoma surgery found that congestive heart failure was linked to postoperative hemorrhaging even though there were no statistically significant links with age, reconstruction, obesity, hypertension, pulmonary disease, or diabetes [10]. The evidence suggests that the severity of the patient’s disease, demographics and particular preoperative complications may elevate the risk of postoperative hemorrhaging; however, systematic proof for preoperative medical variables is lacking. Therefore, this meta-analysis and systematic review seeks to identify the risk variables associated with postoperative bleeding among individuals diagnosed with breast cancer. Methods Register protocol This systematic review has been registered with PROSPERO (CRD420250651911), 13 February 2025, and adhered to the reporting requirements outlined in the PRISMA checklist [11]. Search strategy Two researchers independently conducted literature searches across several records: Wafang, Cochrane Library, WoS, Embase, CNKI, and PubMed. (search cutoff date: May 25, 2025). The core search terms included “breast neoplasms OR breast cancer OR breast tumor” AND “Hemorrhage OR Hemorrhages OR Bleeding”. The complete search techniques, optimized for every record, are provided in Supplementary Material 1. Furthermore, the bibliography of the investigations encompassed was carefully examined to uncover any further relevant articles. Criteria for inclusion and exclusion The criteria for inclusion in the investigation were outlined as: (1) The research subjects were patients who received surgical treatment due to a pathological diagnosis of breast carcinoma; (2) Investigations highlighting the related risk variables of postoperative hemorrhaging in individuals suffering from breast carcinoma; (3)The included studies generally defined postoperative bleeding as recorded in the ICD-10 code and used it as a complication diagnosis after admission; (4) Studies reported adequate data for pooling in the analysis;(5) observational studies (including cohort, case–control, randomized controlled trial studies, and cross-sectional designs). The subsequent exclusion standards were implemented: (1) Research involving a participant pool of less than 100 individuals; (2) Investigations with unavailable full texts or incomplete content; (3) Publications in the form of reviews, letters, or conference abstracts; (4) Studies lacking sufficient data to calculate risk factors. Selection of studies and extraction of data The files obtained were uploaded into the Endnote program for the purpose of managing references. Following the elimination of duplicates, two reviewers (Y.Z. and Z.Y.N.) conducted a separate screening of the abstracts and titles. The full texts of investigations considered for inclusion were subsequently obtained and evaluated by those exact editors. Studies failing to satisfy the conditions for inclusion were omitted. Disagreements that arose during the selection process were addressed through negotiations involving an additional examiner (Q.M.). The final included studies had their data separately extracted by the two writers (Y.Z. and Z.Y.N.). For duplicate reporting of data from the same population, only the most recent or the most complete study was retained. The process of data retrieval was carried out utilizing a meticulously crafted and pilot-tested form, which encompassed: the surname of the principal writer, the publishing year, the geographical location of the institution, the span of the investigation, the study population, the type of research design, the risk factors assessed, and the delineation of post-mastectomy hemorrhaging. Evaluation of quality The potential for bias in the investigations incorporated was evaluated separately by two reviewers, Y.Z. and Z.Y.N. Any inconsistencies were addressed through dialogue with an impartial additional reviewer (W.J.). The evaluation of the efficacy of cohort and case–control investigations was conducted utilizing the NOS, which examines three domains: selection (comprising 4 items), comparability (1 item, with a maximum of 2 stars), and outcome/exposure [12]. In the selection and outcome/exposure domains, each item can receive an upper limit of one star, whereas the comparability domain permits up to two stars. Studies achieving a total rating of ≥ 7 were categorized as excellent quality, those with ratings between 5 and 6 as medium quality, and those rating < 5 as low quality [13]. Data synthesis All acquired data were imported into Excel 2019 for preprocessing. The processed data were then imported into RevMan 5.3 for meta-analysis. When P ≥ 0.1 and I2 ≤ 50%, it suggests a high degree of homogeneity among the investigation results, warranting the use of a fixed-effect model for the meta-analysis. When I2 > 50%, significant heterogeneity exists between the research outcomes, and a random-effects model should be utilized for the meta-analysis [14]. Dichotomous parameters were represented as odds ratios (OR), while continuous parameters were represented as standardized mean differences (SMD), each accompanied by 95% CI. Heterogeneity was evaluated utilizing the χ2 test combined with the I2 statistic (significance level p = 0.1). If P > 0.1 and I2 < 50% heterogeneity was deemed low or negligible; otherwise, substantial heterogeneity was assumed, and sources were examined by excluding studies one by one. Results Literature and search findings An aggregate of 16,706 pertinent investigations were discovered from six electronic databases, and all references were uploaded into EndNote reference management program. Upon the elimination of duplicates, 14,260 titles and abstracts remained available for screening. By reading the titles and abstracts, and after excluding irrelevant reports and reviews, a total of 36 literatures were subject to full-text review. Of these 36 studies, 27 were omitted for failing to satisfy the inclusion requirements. Since some of the literature has not yet provided the original data in the original text, ultimately, an aggregate of 8 investigations were incorporated in the meta-analysis examining the risk variables of postoperative hemorrhaging after breast carcinoma surgical procedures. Figure 1 displays the PRISMA flow chart for the literature search.Fig. 1Flowchart of Literature Screening Flowchart of Literature Screening Features of the incorporated investigations The characteristics [5, 10, 15–20] of the 8 incorporated investigations are outlined in Table 1. This meta-analysis encompassed investigations published from 2012 to 2024, featuring a total sample size of 82,012 individuals. The majority of the investigations were retrospective cohort investigations (n = 6, 75%), while the others included a prospective observational investigation and a case–control investigation. The majority were published in English, with only one in Chinese [18]. Regarding study location, 37.5% of the studies (n = 3) [5, 15, 19] were conducted in Europe, 37.5% (n = 3) [16–18] in Asia, and 25.0%(n = 2) [10–20] in North America. This broad geographical representation enhances the generalizability and external validity of the findings from this meta-analysis. Table 1Characteristics of the included studies (n = 8)Author(year)LocationStudy designSample sizeRisk factorsL. Winther Lietzen (2012) [5]DenmarkRetrospective cohort study19,919①Anselm Tamminen (2024) [15]FinlandRetrospective cohort study3838②,③Iheoma Y. Nwaogu(2015) [10]The United StatesRetrospective cohort study7907①,④,⑤,⑥,⑦,⑧,⑨Hiroki Morikawa (2023) [16]JapanRetrospective cohort study584②,⑩Norio Emoto (2017) [17]JapanRetrospective cohort study193③,⑥,⑦Yunhong Hao(2019) [18]ChinaCase–control study360③,④,⑤,⑦,⑩Rikke Nørgaard Pedersen (2015) [19]DenmarkProspective cohort study30,711①,④,⑥,⑧,⑦,⑩Zahraa Al-Hilli (2015) [20]The United StatesRetrospective cohort study18,500①,②,④,⑤,⑨,⑩①Mastectomy;②ALND axillary lymph node dissection; ③Antithrombotic medication; ④Age ≥ 60 years old; ⑤Hypertension;⑥CHF: congestive heart failure; ⑦Diabetes; ⑧Chronic pulmonary disease; ⑨reconstruction; ⑩NAC Neoadjuvant chemotherapy; Table 1Characteristics of the included studies (n = 8)Author(year)LocationStudy designSample sizeRisk factorsL. Winther Lietzen (2012) [5]DenmarkRetrospective cohort study19,919①Anselm Tamminen (2024) [15]FinlandRetrospective cohort study3838②,③Iheoma Y. Nwaogu(2015) [10]The United StatesRetrospective cohort study7907①,④,⑤,⑥,⑦,⑧,⑨Hiroki Morikawa (2023) [16]JapanRetrospective cohort study584②,⑩Norio Emoto (2017) [17]JapanRetrospective cohort study193③,⑥,⑦Yunhong Hao(2019) [18]ChinaCase–control study360③,④,⑤,⑦,⑩Rikke Nørgaard Pedersen (2015) [19]DenmarkProspective cohort study30,711①,④,⑥,⑧,⑦,⑩Zahraa Al-Hilli (2015) [20]The United StatesRetrospective cohort study18,500①,②,④,⑤,⑨,⑩①Mastectomy;②ALND axillary lymph node dissection; ③Antithrombotic medication; ④Age ≥ 60 years old; ⑤Hypertension;⑥CHF: congestive heart failure; ⑦Diabetes; ⑧Chronic pulmonary disease; ⑨reconstruction; ⑩NAC Neoadjuvant chemotherapy; Characteristics of the included studies (n = 8) ①Mastectomy;②ALND axillary lymph node dissection; ③Antithrombotic medication; ④Age ≥ 60 years old; ⑤Hypertension;⑥CHF: congestive heart failure; ⑦Diabetes; ⑧Chronic pulmonary disease; ⑨reconstruction; ⑩NAC Neoadjuvant chemotherapy; Evaluation of quality The quality evaluation findings for every investigation are presented in Supplementary Tables 2 and 3. NOS reported that seven cohort investigations exhibited high quality, whereas one case–control investigation was rated as moderate quality. Table 2Results of methodological quality assessment of cohort studies (n = 7)Study inclusionSelection of study populationComparability between groupsResultsTotal score (points)(1)(2)(3)(4)(5)(6)(7)L. Winther Lietzen [5]111111118Anselm Tamminen [15]111111017Iheoma Y. Nwaogu [10]111111118Hiroki Morikawa [16]111111118Norio Emoto [17]111111107Rikke Nørgaard Pedersen [19]111111107Zahraa Al-Hilli [20]111111118(1) Representativeness of exposure group; (2) selection of non-exposure group; (3) determination of exposure factors; (4) occurrence of outcome events before the start of the study; (5) Assessment of Outcome; (6) Was Follow-Up Long Enough for Outcomes to Occur; (7) Adequacy of Follow Up of CohortsTable 3Results of methodological quality assessment of case–control study (n = 1)Study inclusionSelection of study populationComparability between groupsExposureTotal score (points)(1)(2)(3)(4)(5)(6)(7)Yunhong Hao(2019) [18]111011106(1) Is the Case Definition Adequate;(2) Representativeness of the Cases;(3) Selection of Controls;(4) Definition of Controls;(5) Ascertainment of exposure;(6) Same method of ascertainment for cases and controls;(7) Non-response rate; Selection of study population; Comparability between groups; Results Table 2Results of methodological quality assessment of cohort studies (n = 7)Study inclusionSelection of study populationComparability between groupsResultsTotal score (points)(1)(2)(3)(4)(5)(6)(7)L. Winther Lietzen [5]111111118Anselm Tamminen [15]111111017Iheoma Y. Nwaogu [10]111111118Hiroki Morikawa [16]111111118Norio Emoto [17]111111107Rikke Nørgaard Pedersen [19]111111107Zahraa Al-Hilli [20]111111118(1) Representativeness of exposure group; (2) selection of non-exposure group; (3) determination of exposure factors; (4) occurrence of outcome events before the start of the study; (5) Assessment of Outcome; (6) Was Follow-Up Long Enough for Outcomes to Occur; (7) Adequacy of Follow Up of Cohorts Results of methodological quality assessment of cohort studies (n = 7) (1) Representativeness of exposure group; (2) selection of non-exposure group; (3) determination of exposure factors; (4) occurrence of outcome events before the start of the study; (5) Assessment of Outcome; (6) Was Follow-Up Long Enough for Outcomes to Occur; (7) Adequacy of Follow Up of Cohorts Table 3Results of methodological quality assessment of case–control study (n = 1)Study inclusionSelection of study populationComparability between groupsExposureTotal score (points)(1)(2)(3)(4)(5)(6)(7)Yunhong Hao(2019) [18]111011106(1) Is the Case Definition Adequate;(2) Representativeness of the Cases;(3) Selection of Controls;(4) Definition of Controls;(5) Ascertainment of exposure;(6) Same method of ascertainment for cases and controls;(7) Non-response rate; Selection of study population; Comparability between groups; Results Results of methodological quality assessment of case–control study (n = 1) (1) Is the Case Definition Adequate;(2) Representativeness of the Cases;(3) Selection of Controls;(4) Definition of Controls;(5) Ascertainment of exposure;(6) Same method of ascertainment for cases and controls;(7) Non-response rate; Selection of study population; Comparability between groups; Results Meta-analysis of risk variables associated with postoperative hemorrhaging after breast carcinoma operation Utilizing RevMan 5.3 software, we conducted a pooled analysis of risk factors reported in two or more investigations. The findings revealed that among the various risk factors considered in the analysis, 3 had statistically significant effects on postoperative bleeding following breast cancer surgery: specifically, mastectomy (OR = 1.93, 95% CI: 1.20–3.09), hypertension(OR = 1.23,95%CI:1.05–1.44), CHF(OR = 2.18,95%CI:1.54–3.08).However, ALND, antithrombotic medication, age ≥ 60 years old, diabetes, chronic pulmonary disease, reconstruction and NAC were not important risk factors. Detailed information is presented in Table 4. Table 4Meta-analysis of risk factors of postoperative bleeding following breast cancer surgeryCorrelation factorsNumber of studiesStatistical methodPooled OR (95%CI)P-valueHeterogeneity Test(I2) (%)Mastectomy4Random1.93(1.20,3.09)0.00694ALND3Random0.43(0.12,1.15)0.1988Antithrombotic medication3Random3.91(0.79,19.40)0.0996Age ≥ 60 years old4Random1.09(0.91,1.30)0.3657Hypertension3Fixed1.23(1.05,1.44)0.010CHF3Fixed2.18(1.54,3.08)0.00010Diabetes4Random1.30(0.89,1.90)0.1855Chronic pulmonary disease2Random1.24(0.86,1.80)0.2556Reconstruction2Random1.16(0.52,2.59)0.7294NAC4Random0.83(0.57,1.22)0.3554 Table 4Meta-analysis of risk factors of postoperative bleeding following breast cancer surgeryCorrelation factorsNumber of studiesStatistical methodPooled OR (95%CI)P-valueHeterogeneity Test(I2) (%)Mastectomy4Random1.93(1.20,3.09)0.00694ALND3Random0.43(0.12,1.15)0.1988Antithrombotic medication3Random3.91(0.79,19.40)0.0996Age ≥ 60 years old4Random1.09(0.91,1.30)0.3657Hypertension3Fixed1.23(1.05,1.44)0.010CHF3Fixed2.18(1.54,3.08)0.00010Diabetes4Random1.30(0.89,1.90)0.1855Chronic pulmonary disease2Random1.24(0.86,1.80)0.2556Reconstruction2Random1.16(0.52,2.59)0.7294NAC4Random0.83(0.57,1.22)0.3554 Meta-analysis of risk factors of postoperative bleeding following breast cancer surgery Discussion Our study examined the risk variables associated with postoperative hemorrhaging following breast carcinoma operation using a systematic review approach. The results showed that mastectomy, hypertension and CHF were identified as risk variables for postoperative hemorrhaging following breast carcinoma surgical procedures. Our analysis indicates that breast carcinoma individuals opting for mastectomy as their surgical method faced a higher danger of postoperative hemorrhaging. This conclusion aligns with the findings of a study carried out in Denmark regarding predictive variables for re-operation stemming from postoperative bleeding in breast carcinoma patients [4]. An investigation has confirmed that the danger of postoperative hemorrhaging is greatest among individuals undergoing mastectomy, subsequent to those who have autologous tissue reconstruction surgery [21]. Breast carcinoma, a prevalent malignant tumor among women, primarily relies on surgery as its main treatment approach. Whether it is breast-conserving surgery, total mastectomy or breast reconstruction surgical procedures, all of the procedures are employed. The treatment guidelines for breast cancer specify that patients choosing the breast-conserving procedure will receive radiotherapy. Therefore, some patients may choose total mastectomy to avoid radiotherapy [22, 23]. A guideline from Denmark also suggests that if the patient is older or has underlying conditions, total mastectomy should be chosen to avoid radiotherapy [24]. In breast-conserving procedures, the pectoralis major muscle is more extensively exposed compared to other surgical methods, and the wider surgical site may increase the risk of postoperative hemorrhage and exudation [25]. This could be one of the reasons why patients undergoing total mastectomy have a higher risk of postoperative bleeding. To avoid this risk, doctors usually place a drainage tube after total mastectomy to reduce the formation of hematoma and also to better monitor the postoperative bleeding situation [26]. The current investigation has not identified the administration of antithrombotic medications as a risk variable for postoperative hemorrhaging following breast carcinoma surgical procedures,which is inconsistent with the research results of some scholars [15, 18, 27]. Possible reasons for this discrepancy include the limited number of original studies incorporating anticoagulant medications, substantial heterogeneity among these studies, and significant differences in the study populations. However, antithrombotic therapy remains a crucial component of cancer treatment, and cancer-associated venous thromboembolism (VTE) is the second leading cause of death in patients with cancer after progression [28]. Therefore, we have still included a discussion on the impact of antithrombotic therapy for post-operative hemorrhaging following breast carcinoma surgical procedures. Several studies indicate that antithrombotic medications may elevate the risk of postoperative hemorrhaging in non-cancer surgical procedures, including hip replacement surgery, tonsillectomy, and transurethral prostatectomy, among others on [29–31]. The same is true in breast cancer surgeries. This is in line with the research conducted by scholars such as Tamminen [15] and Konishi [27]. While the use of antithrombotic medications may elevate the danger of hemorrhaging following breast carcinoma surgical procedures, the overall occurrence of bleeding complications is still quite low [15, 17]. We must consider this question: Is antithrombotic treatment more important than reducing the risk of bleeding? Some guidelines suggest that preventing thromboembolism is more crucial, as it can lead to serious complications or even death [32–34]. In individuals who have received coronary artery stent implantation, discontinuation of antithrombotic medications markedly elevates the danger of thrombosis, potentially resulting in myocardial infarction and mortality [35]. Consequently, certain guidelines recommend the continuation of antiplatelet medications for individuals at a greater danger of thromboembolism throughout the perioperative phase [33, 36, 37]. Overall, some studies suggest that postoperative bleeding complications following breast cancer surgery can be regarded as a non-fatal complication. However, if the treatment of antithrombotic drugs is suspended due to the surgery, it may pose a threat to the patient’s life [38].Therefore, whether to discontinue antithrombotic drugs still needs to be chosen based on the patient’s condition. Research findings indicate that individuals with breast carcinoma who also have hypertension or congestive heart failure are at an increased risk of postoperative hemorrhaging. This outcome aligns with that of a study in Japan that investigated the risk variables of postoperative hemorrhaging for breast carcinoma in nearly 480,000 people [27]. Patients with CHF often receive long-term antithrombotic therapy. Even if the recommended drug withdrawal time is followed, the anticoagulant effect of this treatment still persists [39]. If we choose to suspend the use of anticoagulants or replace them with low-molecular-weight heparin before the operation, the patient’s clotting ability will also be poor. Moreover, some investigations indicate that low-molecular-weight heparin may be a risk variable for hematoma formation following breast carcinoma surgical procedures [40]. Conversely, individuals with CHF frequently exhibit hypertension. During the surgery, the condition of hypertension can be well controlled and may be relatively low. This situation may make it difficult to identify and control small bleeding sources during the operation. Postoperatively, due to stress factors such as pain, these bleeding sources will bleed heavily, thereby affecting the patient’s postoperative recovery [10]. Therefore, the control of blood pressure during the operation and the management of blood pressure after surgery are extremely important. Of course, our research has its limitations. Firstly, this study exhibits considerable heterogeneity. Our analysis might be due to the fact that the studies included different types of breast cancer, and the study populations came from all over the world. Furthermore, our analysis focused solely on mastectomy, despite the existence of various classifications of breast cancer surgery, which may contribute to the observed heterogeneity. Secondly, since the quantity of incorporated literature was fewer than 10, after literature review and group discussion, we did not use funnel plots for heterogeneity analysis. The funnel plot test is ineffective and inadequate for detecting true asymmetry if the quantity of incorporated investigations is limited. Finally, since we chose to extract the original data from the literature, four of the studies failed to provide the original data and were thus excluded. One of these studies had a sample size of nearly 500,000. Even after we emailed the authors, we were unable to obtain the original data. Nevertheless, this investigation represents the first quantitative evaluation of risk variables for postoperative hemorrhaging in breast carcinoma patients, synthesizing global research findings. The present meta-analysis has the potential to offer insights into the early detection and treatment of postoperative hemorrhaging in individuals diagnosed with breast carcinoma. Conclusion Our research results indicate that mastectomy, hypertension and CHF are risk variables contributing to the likelihood of postoperative hemorrhage in cases of breast carcinoma. To our understanding, this represents the inaugural instance of a meta-analysis and systematic review addressing the risk variables linked to postoperative hemorrhaging in breast carcinoma. For individuals with the aforementioned conditions, our medical staff need to pay special attention and take appropriate preventive measures to reduce the likelihood of postoperative bleeding. This not only accelerates the recovery of patients but also reduces the additional treatment costs caused by postoperative bleeding. In our opinion, the assessment and intervention of bleeding risk factors before the surgery and the intervention of patients’ basic conditions during the preoperative assessment are important steps. In the future, it is conceivable that researchers will develop a model or medical instrument capable of forecasting the likelihood of postoperative hemorrhaging.