Safety of abdominal paracentesis in hospitalised patients receiving uninterrupted therapeutic or prophylactic anticoagulants

Abstract

Background

Abdominal paracentesis is a frequently performed procedure in hospitalised patients with ascites. Concurrently, most hospitalised adult patients receive pharmacologic anticoagulation, either for therapeutic purposes or prophylactically to prevent venous thromboembolism. Despite this, minimal evidence exists to guide management of anticoagulant therapy pre‐ and post‐paracentesis.

Aims

The authors aimed to investigate the safety of abdominal paracentesis in hospitalised patients receiving therapeutic or prophylactic anticoagulation, including in patients for whom these medications were withheld periprocedurally.

Methods

TriNetX, an electronic health record data set, was queried to identify patients between the ages of 18 and 80 years who received an abdominal paracentesis while hospitalised at the authors' institution between September 2017 and June 2022. Patients receiving prophylactic anticoagulation (137), therapeutic anticoagulation (74) and no anticoagulation because of coagulopathy or thrombocytopenia (15) were compared. Rates of withholding anticoagulation, performing service, pre‐ and post‐paracentesis haemoglobin, bleeding complications, thrombotic complications and need for red blood cell transfusion were analysed.

Results

Procedure‐related bleeding complications occurred in two (1.4%) patients in the prophylactic group and 0 (0%) patients in the therapeutic group (P = 0.54). No thrombotic complications occurred. Rates of red blood cell transfusions post‐paracentesis were similar between groups. Analysis of secondary end‐points identified significant differences in rates of withholding anticoagulation and mean change in haemoglobin between performing services.

Conclusion

Performance of abdominal paracentesis in patients receiving therapeutic or prophylactic anticoagulation appears to be safe regardless of whether anticoagulation was interrupted periprocedurally, with low rates of bleeding complications, thrombotic complications or need for red blood cell transfusions post‐paracentesis.

Introduction

Abdominal paracentesis is a frequently performed invasive procedure in hospitalised patients, most often for those with diagnosed or suspected liver disease, such as cirrhosis. ¹ This procedure, involving drainage of ascitic fluid from the peritoneal cavity with a needle through a percutaneous approach, is generally performed outside of the operating room and may be safely performed at the patient's bedside with ultrasound guidance. ¹ , ² , ³ Sampling of ascitic fluid assists providers in determining the cause of ascites, evaluating for malignancy or infection, and relieving pain and discomfort. ¹ Considerable evidence exists in support of the general safety of this procedure, though the most common risks include iatrogenic infection, bowel perforation, bleeding and persistent ascitic leak. ¹ , ² To date, there is scant evidence available to guide the management of patients receiving periprocedural anticoagulation, whether it be for prophylaxis of venous thromboembolism or for a therapeutic indication such as atrial fibrillation or a mechanical cardiac valve. The Society of Interventional Radiology consensus guidelines categorise abdominal paracentesis as a low‐risk procedure and generally recommend that it may be performed without interruption of anticoagulant therapy, although the data guiding this recommendation are characterised as low quality. ⁴ However, despite the general consensus that abdominal paracentesis is safe to be performed without interruption of anticoagulant medications, it remains common that these medications are withheld prior to and often following paracentesis in hospitalised patients.

The authors of this investigation hypothesised that abdominal paracentesis may be safely performed in patients continued on therapeutic or prophylactic anticoagulation without interruption periprocedurally. Further, the authors hypothesised that because of a lack of high‐quality evidence to guide decision‐making, there would be significant variance in rates of administration or withholding of anticoagulant therapies for hospitalised patients undergoing abdominal paracentesis at a large, tertiary care academic medical facility.

Methods

Following institutional review board approval, TriNetX, an electronic health record (EHR) data set, was queried to identify all patients between the ages of 18 and 80 years who received an abdominal paracentesis while hospitalised at a large, tertiary care, academic institution between September 2017 and June 2022. The query was further stratified to include only patients who also received one of the following anticoagulants before or after paracentesis: warfarin, apixaban, rivaroxaban, dabigatran, argatroban or enoxaparin. Heparin was not included as a search term because of technical difficulties with the EHR data set software. TriNetX utilises diagnosis and billing codes to generate patient samples, so it is possible that some procedures of interest were not captured if provider billing was erroneous or omitted.

Patients were identified through the above query. Each patient's EHR was retrospectively analysed by two, independent investigators. Patients were excluded from review and analysis if any of the following were identified during index hospitalisation: presence or concern for gastrointestinal or alternative acute blood loss within 3 days before paracentesis precluding anticoagulation, performance of paracentesis did not occur while hospitalised, anticoagulation was not received because of patient refusal, inadequate data within the EHR for comprehensive review or performance of an additional procedure within 2 days before or after paracentesis affecting management of anticoagulation.

Patients meeting inclusion criteria were categorised into three groups: those receiving prophylactic anticoagulation periprocedurally (prophylactic group), those receiving therapeutic anticoagulation periprocedurally (therapeutic group) and those who received no periprocedural anticoagulation with documentation indicating that this decision was because of thrombocytopenia or coagulopathy (coagulopathy group). Prophylactic anticoagulation was defined as subcutaneous heparin (most often at a dose of 5000 units two to three times daily) or subcutaneous enoxaparin at a dose of <1 mg/kg daily (most often 30–40 mg once daily). Therapeutic anticoagulation was defined as continuous intravenous heparin infusion, subcutaneous enoxaparin at a dose of 1 mg/kg twice daily or 1.5 mg/kg daily, apixaban 2.5 mg for atrial fibrillation or 5 mg twice daily regardless of indication, rivaroxaban 15 mg daily, 15 mg twice daily or 20 mg daily, warfarin use with an international normalised ratio (INR) of >1.5 or any use of dabigatran or argatroban.

Among the 297 patient records generated through TriNetX query, 71 patients were excluded for the reasons listed above. Two hundred twenty‐six patients were included in the final analysis, with 137 patients in the prophylactic group, 74 patients in the therapeutic group and 15 patients in the coagulopathy group (Fig. 1). It is worth noting that if patients had intrinsic coagulopathy or thrombocytopenia but still received any periprocedural anticoagulation, they were included in their respective anticoagulation group.

Figure 1.

Protocol design. AC, anticoagulation; EHR, electronic health record.

Baseline characteristics were obtained by manual review by the authors. Outcomes of interest were defined as follows. Bleeding complications were defined as documented haemorrhage at the puncture site, abdominal wall hematoma and/or hemoperitoneum. Thrombotic complications were defined as the identification of a newly diagnosed pulmonary embolism (PE) or venous thrombosis (VT) within 72 h post‐paracentesis in patients who had any doses of anticoagulation withheld periprocedurally. Red blood cell transfusion was defined as a packed red blood cell (PRBC) transfusion within 48 h post‐paracentesis regardless of indication. Categorical variables were analysed with chi‐square testing and continuous variables with one‐way analysis of variance analysis. Comparison of change in haemoglobin concentration between the prophylactic, therapeutic and coagulopathy groups was performed with Wilcoxon testing. Further analysis between prophylactic and therapeutic groups was performed with Fisher exact testing. All additional significance testing between all groups was performed with linear and logistic regression analysis controlled for age and sex.

Results

Two hundred twenty‐six patients were included in the final analysis. The median age of all included patients was 58 years; 61% of patients were male and 39% were female. Median pre‐procedure haemoglobin, platelet count and INR, as well as full baseline characteristics, are shown in Table 1. Among patients in the prophylactic group, 106 received enoxaparin and 31 received heparin. Among patients in the therapeutic group, 25 received warfarin, 19 received enoxaparin, 13 received apixaban, 11 received heparin, five received rivaroxaban and one received argatroban.

Table 1.

Baseline characteristics

	Prophylactic (n = 137)	Therapeutic (n = 74)	Coagulopathy (n = 15)	P value
Age (median)	57	61	58	0.110
Sex	63% male	57% male	60% male	0.694
Pre‐procedure haemoglobin (mean ± SD)	10.4 ± 2.1	10.2 ± 2.3	9.2 ± 2.0	0.143
Pre‐procedure platelet count (mean ± SD)	188 ± 120	235 ± 137	51 ± 21	<0.001
Pre‐procedure INR (mean ± SD)	1.6 ± 0.5	1.8 ± 0.8	1.8 ± 0.5	0.052

P values were derived from ANOVA tests for differences in means for continuous variables or from chi‐square tests for categorical variables.

ANOVA, analysis of variance; INR, international normalised ratio; SD, standard deviation.

Among the 226 patients included in the final analysis, there were two instances of procedure‐related bleeding complications post‐paracentesis, one minor and one major, both of which occurred in the prophylactic group. One patient developed a rectus sheath hematoma several days after paracentesis, which delayed discharge from hospital but otherwise did not impact care, and there was no documentation of recurrence thereafter. This patient had two doses of prophylactic enoxaparin held prior to their paracentesis performed by radiology. One patient developed worsening multifactorial shock following paracentesis with cross‐sectional imaging demonstrating possible hemoperitoneum. This patient died 3 days post‐paracentesis, with an official cause of death documented as sepsis caused by spontaneous bacterial peritonitis. However, it is possible that post‐procedural intra‐abdominal haemorrhage contributed to this patient's poor outcome. This patient was continued on uninterrupted prophylactic heparin prior to their paracentesis performed by medicine. There were no thrombotic or intraprocedural complications among any patients in our sample.

The rates of PRBC transfusion within 48 h post‐paracentesis were similar between the three groups, occurring nine times (6.6%) in the prophylactic group, three times (4.1%) in the therapeutic group and one time (6.6%) in the coagulopathy group (P = 0.55). The change in pre‐ and post‐paracentesis haemoglobin concentrations was similar between the prophylactic, therapeutic and coagulopathy groups, with a decrease in median haemoglobin concentration of 0.40 g/dL (3.9%), 0.25 g/dL (2.5%) and 0.40 g/dL (4.7%) respectively (P = 0.93).

There was considerable heterogeneity among patients with regard to management of periprocedural anticoagulation. Among the 211 patients receiving any form of periprocedural anticoagulation, 58 (27.5%) had at least one dose held pre‐paracentesis. The rate of withholding anticoagulation was similar between the therapeutic and prophylactic groups at 32.4% and 24.8% respectively (P = 0.26). A full comparison of results between prophylactic, therapeutic and coagulopathy groups is displayed in Table 2.

Table 2.

Results by anticoagulation status

	Prophylactic (n = 137)	Therapeutic (n = 74)	Coagulopathy (n = 15)	P value
Change in haemoglobin (mean ± SD)	−0.40 ± 0.91 (−3.9%)	−0.25 ± 0.94 (−2.5%)	−0.40 ± 0.96 (−4.7%)	0.93
PRBC transfusion	9 (6.6%)	3 (4.1%)	1 (6.6%)	0.55
Thrombotic complication	0 (0%)	0 (0%)	0 (0%)	‐
Procedure‐related bleeding complication	2 (1.4%)	0 (0%)	0 (0%)	0.54
Intraprocedural complication	0 (0%)	0 (0%)	0 (0%)	‐
Zero doses held	103 (75.2%)	50 (67.6%)	‐	0.26
1+ doses held	34 (24.8%)	24 (32.4%)	‐	0.26

P values were derived from ANOVA tests for differences in means for continuous variables, or from Fisher's exact tests or chi‐square tests for categorical variables.

ANOVA, analysis of variance; PRBC, packed red blood cell; SD, standard deviation.

Among all patients with no doses of anticoagulation held pre‐paracentesis, there was a mean decrease in haemoglobin concentration of 0.42 g/dL (4.1% decrease). Comparatively, there was a mean decrease in haemoglobin concentration of 0.17 g/dL (1.6% decrease) among all patients with at least one dose of anticoagulation held pre‐paracentesis. The difference between groups was not statistically significant (P = 0.11). Results between patients with zero versus at least one dose of anticoagulation held pre‐paracentesis are shown in Table 3.

Table 3.

Results by doses held

	0 Doses held (n = 153)			1+ Doses held (n = 58)
	Therapeutic (n = 50)	Prophylactic (n = 103)	Total (n = 153)	Therapeutic (n = 24)	Prophylactic (n = 34)	Total (n = 58)	P value
Change in haemoglobin (mean ± SD)	−0.40 ± 0.96 (−3.9%)	−0.42 ± 0.89 (−4.1%)	−0.42 ± 0.91 (−4.1%)	−0.11 ± 0.90 (−1.1%)	−0.20 ± 0.95 (−1.9%)	−0.17 ± 0.93 (−1.6%)	0.11
PRBC transfusion	3 (6.0%)	8 (7.8%)	11 (7.2%)	0 (0%)	1 (2.9%)	1 (1.7%)	0.16
Thrombotic complication	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	‐
Procedure‐related bleeding complication	0 (0%)	1 (1.0%)	1 (0.7%)	0 (0%)	1 (2.9%)	1 (1.7%)	0.53
Intraprocedural complication	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	‐
LOS (mean, median)	18.04, 7.5	12.17, 8.0	14.09, 8.0	16.92, 11.5	13.79, 6.0	15.09, 8.5	0.86

P values were derived to compare patients with 0 dose versus those with 1+ dose, based on t tests for differences in means for continuous variables, or from chi‐square tests for categorical variables.

LOS, length of stay; PRBC, packed red blood cell; SD, standard deviation.

A statistically significant difference was identified in mean change in haemoglobin (P = 0.007) and rates of withholding anticoagulant medications preprocedurally between performing services (P ≤ 0.001), and a near statistically significant difference was detected in rates of PRBC transfusion post‐paracentesis (P = 0.060). Patients receiving paracentesis by radiology providers experienced smaller decreases in haemoglobin concentration and were more likely to have at least one dose of anticoagulation held preprocedurally. A comparison of outcomes between different performing services is illustrated in Table 4.

Table 4.

Analysis of outcomes compared between service performing paracentesis

	Medicine (n = 117)	Radiology (n = 86)	Emergency Medicine (n = 18)	Anaesthesia (n = 5)	P value
Change in haemoglobin (mean ± SD)	−0.42 ± 0.99 (−4.1%)	−0.11 ± 0.76 (−1.1%)	−0.74 ± 0.85 (−6.8%)	−0.74 ± 0.74 (−7.7%)	0.007
PRBC transfusion	7 (6.0%)	3 (3.5%)	1 (5.5%)	2 (40%)	0.06
Thrombotic complication	0 (0%)	0 (0%)	0 (0%)	0 (0%)	‐
Procedure‐related bleeding complication	1 (0.8%)	1 (1.2%)	0 (0%)	0 (0%)	0.99
Intraprocedural complication	0 (0%)	0 (0%)	0 (0%)	0 (0%)	‐
Zero doses held (pre)	89 (76.1%)	43 (50%)	16 (88.9%)	5 (100%)	<0.001
1+ doses held (pre)	18 (15.4%)	40 (46.5%)	0 (0%)	0 (0%)
No AC because of coagulopathy	10 (8.5%)	3 (3.5%)	2 (11.1%)	0 (0%)	0.77
Patients on therapeutic AC	38 (32.5%)	28 (32.6%)	5 (27.8%)	2 (40%)
Patients on prophylactic AC	69 (59.0%)	55 (63.9%)	11 (61.1%)	3 (60%)

P values were derived from ANOVA tests for differences in means for continuous variables or from chi‐square tests for categorical variables.

AC, anticoagulation; ANOVA, analysis of variance; PRBC, packed red blood cell; SD, standard deviation.

Subgroup analyses were performed on patients receiving paracentesis by radiology providers, as well as internal or family medicine (medicine) providers. When controlling for performing service, no significant differences in change in haemoglobin, rates of transfusion, rates of intraprocedural complications, procedure‐related bleeding complications or thrombotic complications were identified between patients who had zero or one dose of anticoagulation withheld.

No significant differences were noted in thrombotic complications, intraprocedural complications, procedure‐related bleeding complications or length of stay in any analysis.

Discussion

To our knowledge, this study represents the largest published analysis of outcomes following abdominal paracentesis among patients receiving anticoagulant medications periprocedurally. Our study found no difference in outcomes between patients receiving therapeutic or prophylactic anticoagulation, regardless of whether these medications were continued or withheld preprocedurally. Although the Society of Interventional Radiology guidelines suggest that paracentesis may be performed under ultrasound guidance without interruption of anticoagulation, our study supports that clinicians continue to face uncertainty in this common clinical scenario. ¹ , ⁴ , ⁵ It is likely that this variance in provider practice patterns stems from a lack of strong evidence to guide this recommendation, as these recommendations are based on low‐quality evidence, principally from retrospective analysis. ⁴ , ⁶ At the authors' institution, there is no objective institutional guideline of policy for management of periprocedural anticoagulation pre‐ or post‐paracentesis, and, as such, all patients in this study were managed at the discretion of their providers.

In 2005, Pache et al. demonstrated that performance of abdominal paracentesis may be safely performed in patients with liver disease. ⁷ However, while this and several additional published studies demonstrate the safety of abdominal paracentesis without correction of underlying coagulation abnormalities – including INR elevation and thrombocytopenia – there are minimal investigations into the safety of peri‐paracentesis anticoagulation. ³ , ⁵ , ⁸ , ⁹

In 2015, Devarbhavi et al. found no increased risk of haemorrhage in patients receiving abdominal paracentesis and were taking warfarin for Budd–Chiari. ¹⁰ A total of 32 patients with a mean INR of 3.1 had paracentesis performed without reversal of anticoagulation or administration of blood products. ¹⁰ It is worth noting that the paracenteses in this study were performed without ultrasound guidance. Regardless, there were no bleeding or hemoperitoneum complications noted, suggesting the safety of paracentesis despite therapeutic anticoagulation. ¹⁰ In the years since, no additional studies have been published to replicate the safety demonstrated in this report, despite the rising popularity of direct oral anticoagulants (DOACs) such as apixaban and rivaroxaban.

More recently, in 2023, Kuperman et al. reported two case reports of severe bleeding complications in patients taking apixaban with compensated, Child–Pugh Class B cirrhosis following abdominal paracentesis. ¹¹ Notably, these patients also had evidence of significant renal dysfunction. ¹¹ The investigators concluded that despite the general consensus of the safety of this procedure in patients with cirrhosis taking anticoagulation medications, further study is warranted to investigate the safety of DOACs in this population. ¹¹

Perhaps most notably, Barsuk et al. performed a prospective study investigating the safety of a ‘two‐probe’ technique in order to reduce haemorrhagic complications of abdominal paracentesis at Cedars‐Sinai Medical Center (CSMC) in 2017. ¹² While this study did not specifically investigate the effects or practice patterns regarding periprocedural anticoagulation, 5777 abdominal paracenteses were performed in 1639 patients and the authors indicated that ‘per the routine of the CSMC procedure service, antiplatelet or anticoagulant medications were not held for paracenteses’. ¹² However, it is unclear how many of these procedures were actually performed in patients receiving uninterrupted therapeutic or prophylactic anticoagulants. ¹² Furthermore, all procedures in this study were performed by a single, extensively experienced provider. The results demonstrated tremendous safety of abdominal paracentesis in general, but especially in those who received paracentesis using their multi‐probe technique by an experienced provider.

Our findings demonstrate the safety of performing paracentesis in patients without interruption of either therapeutic or prophylactic anticoagulation. Although there was a non‐statistically significant trend towards smaller decreases in haemoglobin concentration and reduced rates of PRBC transfusion in patients with at least one dose of anticoagulation held pre‐paracentesis, this was confounded by the finding that patients receiving paracentesis by radiology providers tended to have preprocedural anticoagulation held significantly more frequently. Procedures performed by radiology had anticoagulation interrupted 46.5% of the time, as compared to 15.4% of the time by medicine providers (P ≤ 0.001). When analysing outcomes by these departments – stratified by interruption of anticoagulation periprocedurally – there was no observed difference in change in haemoglobin or rates of PRBC transfusion between patients with zero or at least one dose held. This may suggest that the slight differences seen are more attributable to procedural experience and expertise by radiology providers rather than because of withholding or continuation of anticoagulation periprocedurally.

It is worth noting that the rate of PRBC transfusion following paracentesis was considerably higher than the rate of procedure‐related bleeding complications, occurring in 13 patients (5.8%) and two patients (0.88%) respectively. We suspect that this result is because of non‐procedure‐related factors that may have contributed to worsening anaemia in the hospitalised patients included in our analysis, such as acute illness causing a relative myelosuppression and iatrogenic anaemia from serum laboratory analysis.

Our study has several limitations. First, this was a retrospective review of non‐randomised patients at a single centre. Second, the rates of major adverse outcomes were very rare, thus making it challenging to draw definitive conclusions based on significance testing. As with all retrospective studies, it is possible that other clinical characteristics were not accounted for that may have contributed to differences between groups, though the authors feel that this is unlikely. Strengths of our study include the fact that paracenteses were performed by a variety of providers, both experienced and relatively inexperienced. Additionally, patients included in this analysis received a variety of anticoagulant medications.

Conclusion

Based on our study, there appears to be no significant difference in outcomes following abdominal paracentesis between patients receiving prophylactic or therapeutic anticoagulation regardless of whether anticoagulants are continued or withheld preprocedurally. Significant heterogeneity was noted between clinical departments with respect to periprocedural management of anticoagulation. Our results suggest significant differences in primary and secondary outcomes between performing services. Future high‐quality studies are needed to generate more well‐informed recommendations for the management of anticoagulant medications before and after paracentesis in hospitalised patients.