1 Introduction

The utilization of total shoulder arthroplasty (TSA) has increased significantly in recent years.^1,2 Specifically, in the United States there was a 103 % increase in patients undergoing primary TSA between 2011 and 2017. TSA has a high success rate, however it is associated with several well-documented complications.^3–5 An uncommon, yet potentially significant postoperative adverse event following a total joint arthroplasty (TJA) is gastrointestinal (GI) bleeding.⁶ GI bleeding has been experienced across cardiac, gastric, and colorectal procedures, but recently this complication has been noted following total joint arthroplasty procedures.^6–9 Mortality rates of acute GI bleeds have been estimated between 5 % and 10 %, which causes concern for dire postoperative outcomes in patients, as well as the healthcare resources that are expended.^10–12 While GI bleeding has been recognized as a rare complication after TJA, there is limited literature addressing predisposing risk factors and complications following this event after anatomic and reverse TSA.⁶

Multiple patient-specific characteristics have been implicated in elevating the risk of postoperative GI bleeding, including advanced age, male sex, and the presence of significant medical comorbidities.^6,13,14 Conditions such as congestive heart failure, malignancy, and peptic ulcer disease have demonstrated a strong association with perioperative GI hemorrhage.¹⁵ In the context of TSA, transfusion-requiring hemorrhage has been reported as a frequent postoperative complication, with an estimated incidence of approximately 4.3 %.¹⁶ Physiologically, surgical stress is believed to trigger elevations in endogenous cortisol levels, which may potentiate mucosal inflammation, upregulate gastric acid secretion, and suppress protective mucosal barrier mechanisms.⁶ This heightened vulnerability, when compounded by reduced splanchnic perfusion during surgery and cumulative blood loss in the perioperative period, may predispose the gastrointestinal lining to erosive injury and subsequent bleeding.⁶

Pharmacologic venous thromboembolism (VTE) prophylaxis represents another well-established contributor to bleeding risk. Agents such as aspirin and direct factor Xa inhibitors are routinely employed for both thromboprophylaxis and analgesia in orthopedic patients, with alternative regimens considered for individuals at heightened thrombotic risk or with aspirin intolerance.^17–22 Recent retrospective analyses have further identified elevated preoperative international normalized ratio (INR) values as independent predictors of postoperative bleeding following TSA.^23,24 Currently, consensus guidelines have not been reached on VTE prophylaxis following TSA.²⁵ However, the American Academy of Orthopedic Surgeons’ clinical practice guidelines recommend that, given the lack of strong evidence, physicians should consider using mechanical and/or chemical prophylaxis to prevent perioperative venous thromboembolism in patients undergoing shoulder arthroplasty. Though no specific pharmacologic recommendations are presented.

As the volume of shoulder arthroplasty continues to increase nationally, there is a growing need to better delineate patient- and treatment-specific factors that predispose to GI bleeding in the perioperative setting.^1,2 Improved recognition of these risk profiles may facilitate tailored perioperative protocols considering GI prophylaxis, with the dual goals of optimizing patient safety and minimizing healthcare resource utilization. Accordingly, this study aims to determine the incidence of GI bleeding following anatomic and reverse TSA, identify associated patient and procedural risk factors, and assess postoperative clinical outcomes at 90 days and 1 year, with corresponding cost analyses at the 90-day postoperative period.

2 Methods

A retrospective cohort study was conducted using the Mariner database (PearlDiver Technologies, Colorado Springs, CO), a large, nationally representative administrative claims dataset containing records from over 150 million patients. Patients who underwent primary anatomic and reverse TSA between 2010 and 2020 were identified using CPT code 23472. To ensure adequate longitudinal data, patients were required to have continuous enrollment for at least six months before and one to two years after the index TSA procedure. Only the first instance of TSA per patient was included. This study was deemed exempt from institutional review board (IRB) oversight as it utilized de-identified, retrospective claims data.

GI bleeding events were defined using a comprehensive set of ICD-9 and ICD-10 diagnosis codes representing upper and lower GI hemorrhages, peptic ulcer disease with hemorrhage, esophageal varices, and angiodysplasia (Supplementary Table 1). Patients were stratified into two cohorts based on the presence or absence of a GI bleed within 90 days of the index TSA. A secondary classification assessed prior GI bleeding within the year before surgery.

Patients in the GI bleed and no-bleed cohorts were 1:1 propensity score matched by age, gender, Charlson Comorbidity Index (CCI), obesity, tobacco use, and long-term anticoagulant or antiplatelet use. To further investigate potential risk factors for postoperative GI bleeding, covariates were defined using ICD codes within one year prior to surgery, including cardiovascular comorbidities (congestive heart failure [CHF], atrial fibrillation), chronic kidney disease (CKD), dementia, alcohol use, inflammatory bowel disease, gastritis, GERD, malignancy, and medication exposures (NSAIDs, corticosteroids, and drug-induced GI injury). All variables found to be significant on univariable analysis were subsequently included in a multivariable logistic regression model to assess their independent association with postoperative GI bleeding.

The primary outcomes included 90-day and 1-year postoperative complications, such as hospital readmission, blood transfusion, acute kidney injury (AKI), pneumonia, urinary tract infection (UTI), myocardial infarction, surgical site infection (SSI), hematoma, nerve injury, wound disruption, venous thromboembolism (VTE), reoperation, and mortality. Complication rates were compared using chi-square tests for categorical variables and t-tests or Wilcoxon rank-sum tests for continuous variables as appropriate. Multivariable logistic regression was performed to evaluate the association between postoperative GI bleeding and each complication, adjusting for age, gender, and CCI.

Cost analysis was conducted by comparing total outpatient and inpatient charges within 90 days postoperatively using breakdown commands for standard deviation estimation. Kaplan-Meier survival analyses were conducted to assess cumulative incidence of reoperation and mortality up to two years postoperatively, with log-rank testing between matched GI bleed and no-bleed cohorts (Supplementary Figs. 1 and 2).

3 Results

4 Discussion

This large, matched cohort study evaluated the risk factors, clinical outcomes, and cost burden associated with GI bleeding following TSA. Although the incidence of postoperative GI bleeding was low (0.8 %), its clinical and economic impact was substantial. Among the risk factors evaluated, a prior history of GI bleeding emerged as the only independent predictor of postoperative hemorrhage on multivariable analysis, underscoring its clinical significance and utility in preoperative risk stratification. Patients who experienced GI bleeding had significantly greater odds of 90-day readmission, transfusion, AKI, pneumonia, UTI, myocardial infarction (MI), and VTE, with elevated readmission, transfusion, and DVT risk persisting at one year. Cost analysis revealed that the GI bleed cohort incurred substantially higher outpatient costs postoperatively, whereas inpatient costs were not significantly different. These findings suggest that even though GI bleeding is uncommon after TSA, it may carry a substantial clinical and financial burden, and a discussion about placing these patients on DVT prophylaxis after surgery needs to be considered.

The strong association between prior GI bleeding and postoperative hemorrhage is consistent with existing literature across both medical and surgical disciplines. Prior GI bleeding reflects both mucosal vulnerability and cumulative risk from chronic comorbidities, such as peptic ulcer disease, liver pathology, or anticoagulant use, all of which may be exacerbated during the perioperative period.^26–28 While several other factors, including CHF, atrial fibrillation, alcohol use, and esophageal varices, were significantly associated with GI bleeding on univariable analysis, they lost significance in multivariable modeling. This suggests that these variables may act as indirect proxies for systemic vulnerability or may be collinear with prior bleeding history.^29,

The clinical consequences of GI bleeding after TSA were considerable. At 90 days postoperatively, affected patients had significantly higher rates of hospital readmission, transfusion, AKI, pneumonia, UTI, MI, and DVT. These findings reflect the physiologic vulnerability of elderly or medically complex patients, for whom acute blood loss can precipitate multisystem complications, including cardiovascular, renal, and infectious sequelae.^30,31At one year, patients with GI bleeds continued to experience significantly elevated rates of readmission, transfusion, and thromboembolic events, indicating a sustained downstream risk beyond the immediate recovery period, which may reflect ongoing physiological strain or incomplete recovery in high-risk patients. Importantly, however, reoperation and mortality rates did not differ significantly between cohorts, suggesting that while GI bleeding is associated with medical morbidity, it does not appear to compromise the mechanical success or long-term survival of the arthroplasty itself.^32,33

The economic burden of GI bleeding was also substantial. Outpatient 90-day reimbursement in the GI bleed cohort was nearly double that of matched controls, potentially driven by increased resource utilization such as follow-up visits, imaging, laboratory monitoring, and medical management of complications. These findings are in line with prior studies demonstrating the financial impact of medical complications in orthopedic surgery, which can often exceed the cost of the index procedure itself.^34–36 In contrast, inpatient day-of-surgery costs did not differ significantly between cohorts. This is likely attributable to the bundled nature of surgical billing and the standardized reimbursement structures for index procedures under diagnosis-related group (DRG) payment models, which often do not account for complications that arise after discharge.³⁷

From a clinical perspective, these findings support a more nuanced approach to preoperative risk stratification and perioperative planning. Patients with a known history of GI bleeding should be evaluated for their use of anticoagulants and NSAIDs preoperatively and postoperatively and the potential need for gastroprotective strategies such as proton pump inhibitors or preoperative gastroenterology consultation.^38,39 Early identification and proactive management of high-risk patients may reduce the incidence and severity of bleeding events, potentially avoiding costly downstream complications.

5 Limitations

This study has several limitations inherent to the use of a national administrative claims database and its retrospective design. All diagnoses, procedures, comorbidities, and outcomes were identified using CPT and ICD-9/ICD-10 billing codes. As such, the accuracy of outcome identification depends on provider coding and may be subject to misclassification or underreporting. The database does not include clinical variables such as laboratory values, intraoperative blood loss, endoscopy results, or medication dosages, which limits the ability to assess bleeding severity or confirm etiology.

While 1:1 propensity score matching was performed to control for measurable confounders, residual confounding from unmeasured variables remains possible. Factors such as frailty, nutritional status, over-the-counter NSAID use, and patient adherence to prescribed medications were not captured and may influence both bleeding risk and postoperative outcomes. In addition, mortality may be underestimated in this dataset. Death must be documented through provider-submitted billing codes, and not all deaths occur in hospitals or reimbursed care settings. As a result, deaths that occur at home, in hospice, or in other non-billed environments may not be reflected, limiting the accuracy of survival analyses.

The incidence of GI bleeding following TSA was low, which, while consistent with real-world data, reduces statistical power for detecting associations with rare complications such as cardiac arrest or surgical site infection. Additionally, although we required continuous enrollment for sufficient follow-up, care delivered outside of the billing network or during lapses in insurance coverage may not be captured, leading to potential underestimation of postoperative complications and cost burden. Additionally, given the large sample size afforded by a national claims database, statistically significant p-values should be interpreted with caution, as even small absolute differences may achieve significance without corresponding clinical relevance.

A notable limitation of this study is the inability to differentiate between anatomic and reverse total shoulder arthroplasty, as both procedures are encompassed under the same procedural code (CPT 23472) within the administrative claims database. While this constraint limits the granularity of procedure-specific risk stratification, the overall findings remain applicable to the general TSA population. Given that reverse TSA is more frequently performed in older patients and those with rotator cuff deficiency who may carry distinct perioperative risk profiles, future studies incorporating implant-level data are warranted to delineate whether the observed associations with postoperative gastrointestinal bleeding differ by implant type.

Finally, while the database is broadly representative of insured patients in the United States, it does not include uninsured individuals or those exclusively enrolled in Medicaid, potentially limiting generalizability to more socioeconomically diverse or underinsured populations. Despite these limitations, the use of a large national claims database enabled a large-scale, matched cohort analysis, offering insight into the risk factors, outcomes, and healthcare burden associated with GI bleeding following TSA.

6 Conclusions

As the volume of TSA continues to rise nationally in response to expanding indications and a growing elderly population, rare but high-impact complications such as GI bleeding merit greater attention.^,40 The findings of this study suggest that while uncommon, GI bleeding represents a marker of systemic vulnerability and is associated with increased morbidity and cost. Enhanced perioperative vigilance, targeted prophylaxis, and integrated care pathways may help mitigate the impact of this complication and improve outcomes in high-risk shoulder arthroplasty patients. Future research should explore risk mitigation strategies in this population, including prehabilitation, medication optimization, and enhanced recovery protocols tailored to patients with a history of GI pathology.

Author contribution

Contributorship: Jared Sasaki: Conceptualization, Software, Validation, Investigation, Writing - Original Draft, Writing - Review & Editing, Visualization. Catherine Hand: Investigation, Methodology, Resources, Writing - Original Draft, Writing - Review & Editing, Visualization. Victor Koltenyuk: Conceptualization, Methodology, Writing - Original Draft, Writing - Review & Editing, Visualization. Hirmand Salehi: Writing - Original Draft, Writing - Review & Editing. Henry Eilen: Writing - Original Draft, Writing - Review & Editing. Tanmaya D. Sambare: Investigation, Writing - Review & Editing, Visualization, Supervision. John Andrawis: Writing - Original Draft, Writing - Review & Editing, Conceptualization, Visualization, Supervision.Supplementary Table 1List of ICD-9 and ICD-10 codes used for patient inclusion of postoperative GI bleeding. CPT codes for primary TSA.Supplementary Table 1Type of CodeCodesICD-9ICD-9-D-4560, ICD-9-D-45620, ICD-9-D-53100, ICD-9-D-53120, ICD-9-D-53140, ICD-9-D-53160, ICD-9-D-53200, ICD-9-D-53220, ICD-9-D-53240, ICD-9-D-53260, ICD-9-D-53300, ICD-9-D-53320, ICD-9-D-53340, ICD-9-D-53360, ICD-9-D-53400, ICD-9-D-53420, ICD-9-D-53440, ICD-9-D-53460, ICD-9-D-5693, ICD-9-D-5780, ICD-9-D-5781, ICD-9-D-5789ICD-10ICD-10-D-I8501, ICD-10-D-I8511, ICD-10-D-K250, ICD-10-D-K252, ICD-10-D-K254, ICD-10-D-K256, ICD-10-D-K260, ICD-10-D-K262, ICD-10-D-K264, ICD-10-D-K266, ICD-10-D-K270, ICD-10-D-K272, ICD-10-D-K274, ICD-10-D-K276, ICD-10-D-K280, ICD-10-D-K282, ICD-10-D-K284, ICD-10-D-K286, ICD-10-D-K625, ICD-10-D-K920, ICD-10-D-K921, ICD-10-D-K922CPT23472

Ethical

Institutional Ethical Committee Approval was not needed for this study. However, we upheld publishing ethics as our duty as authors.

Consent

Guardian/Patient consent was not needed for this study.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.