Severe haemorrhage following abdominal paracentesis for ascites in patients with liver disease


Dr M. Bilodeau, Centre hospitalier du l'Université de Montréal, Hôpital Saint-Luc, 264, boul. René-Lévesque est, Montréal, Québec, Canada, H2X 1P1.


Background : Bleeding is a recognized complication of abdominal paracentesis. Special concern has been raised when it is performed in patients with liver failure because of coagulation disorders and collaterals in the abdominal wall.

Aim : To assess the clinical characteristics of patients who developed haemorrhagic complications after paracentesis.

Methods : We reviewed all cases of severe haemorrhage occurring after paracentesis in patients admitted to the Liver Unit of our institution between 1994 and 2004.

Results : Nine cases were identified among 4729 procedures. The occurrence of severe haemorrhage represented 0.19% of all procedures with a death rate of 0.016%. Bleeding was not related to operator experience, elevated international normalized ratio or low platelets. It occurred in patients with high model for end-stage liver disease and Child-Pugh scores. Furthermore, some degree of renal failure was present in all but one patient.

Conclusion : Severe haemorrhage after abdominal paracentesis in patients with liver disease occurs in 0.2% of cases. It occurs in patients with severe liver failure and is often associated with significant pre-existing renal dysfunction.


Abdominal paracentesis is a routine diagnostic and therapeutic procedure in patients with liver disease. It is considered a safe and rarely contraindicated procedure.1, 2 Abdominal wall haematomas arise in <2% of punctures, and significant haemorrhage requiring transfusion occurs in <1% of patients.1, 3–5 It has been suggested that the complication rate may be higher when paracentesis is performed by inexperienced operators.6 The presence of abnormal coagulation parameters should preclude paracentesis only if there is clinical evidence of fibrinolysis or disseminated intravascular coagulation.1 No existent data support a cut-off value for coagulation parameters beyond which paracentesis should be avoided: platelet transfusion or fresh frozen plasma (FFP) administration before the procedure is not currently recommended in patients with coagulation disorders and/or thrombocytopenia.2, 5, 7–10 Recently, a large-scale study addressed the safety of paracentesis performed using a standardized approach in out-patient clinics. No haemorrhagic complication occurred in over 1000 procedures despite platelets counts being as low as 19 × 103/μL and international normalized ratio (INR) as high as 8.7.11

Although death secondary to haemorrhagic abdominal paracentesis is a known complication, few isolated cases have been reported in the literature.1, 3, 5, 12–16 The incidence of mortality caused by severe haemorrhage following this procedure has not been determined so far.

In this paper, we describe the largest number of severe haemorrhages resulting from abdominal paracentesis and, for the first time, determine the mortality rate associated with this complication in patients with liver disease-associated ascites. Only cases occurring outside the context of hepatocellular carcinoma were reviewed, as spontaneous haemoperitoneum can occur as a complication of this type of liver tumour.

Materials and Methods

The total number of paracentesis performed in the Liver Unit of Hôpital Saint-Luc of the Centre hospitalier de l'Université de Montréal was established by extracting data from the Government of Quebec's Health Insurance Plan Database as every Quebec citizen is covered under this health care programme. The period of study extended from 1 January 1994 to 31 December 2003. Cases of severe haemorrhage were identified through a computer-based search in the Medical Records Department that included the terms ‘haemorrhage’ or ‘hemoperitoneum’ in association with ‘ascites puncture’. Haemorrhage was considered severe if there was haemodynamic instability or a significant drop (>15 g/L) in the haemoglobin level.

The procedures were performed blindly either by residents or hepatologists, or under ultrasound guidance in the Radiology Department. A 14- to 18-gauge angiocath needle was used. Albumin infusion was administered when >2 L of ascites was removed. Haematological preparation with FFP, platelets and/or 1-deamino-8-d-arginine vasopressin (DDAVP) to correct coagulation disorders was occasionally delivered without any formal guidelines.

The Child-Pugh score and the model for end-stage liver disease (MELD)–United Network for Organ Sharing (UNOS) scores were calculated using the usual formulas.17 Creatinine clearance was estimated by the Cockroft-Gault formula. Mean ± s.d. values are presented.


Over the 10-year period of this study, 4729 abdominal paracenteses were performed in our Liver Unit. Nine patients presented severe haemorrhage, always following evacuation of >2 L of ascites. The frequency of severe haemorrhagic complication after paracentesis was therefore 0.19% with a lethal outcome in 0.016%. The clinical characteristics are presented in Table 1.

Table 1.  Clinical characteristics of the nine cases of haemorrhage after abdominal paracentesis
 Case no. 1Case no. 2Case no. 3Case no. 4Case no. 5Case no. 6Case no. 7Case no. 8Case no. 9
  1. CP, Child-Pugh; CT, CAT scan; DDAVP, 1-deamino-8-d-arginine vasopressin; HD, haemodialysis; HE, hepatic encephalopathy; INR, international normalized ratio; MELD, model for end-stage liver disease; NASH, non-alcoholic steato-hepatitis; ND, not determined; OLT, orthotopic liver transplantation; Portal pressure, hepatic venous pressure gradient; TIPS, transjugular intrahepatic portal-systemic shunt; US, ultrasound.

Age (years)566053684567422560
Liver diseaseRecurrent cirrhosis after OLTPrimary biliary cirrhosisRecurrent cirrhosis after OLTAutoimmune cirrhosisAcetaminophen intoxicationAcute liver failureAlcoholic cirrhosisIdiopathic ductopeniaNASH
Portal pressure9ND17ND1915NDND29
Platelets (109/L)14611411385761201514071
Creatine clearance (mL/min)274226 (HD)7218 (HD)35424918 (HD)
OperatorRadiologist/ ultrasoundSenior Hepatology ResidentSenior Hepatology ResidentJunior Internal Medicine ResidentHepatologistSenior Hepatology ResidentHepatologistHepatologistSenior Hepatology Resident
Haematologic preparationNoneNoneNoneNoneNoneNoneNoneNoneDDAVP
SymptomsAbdominal distensionPain, nauseaConvulsionsPainPain, haemodynamic instabilityPain, haemodynamic instabilityPainPain, haemodynamic instabilityHaemodynamicinstability
Time interval24 h6 h3 h6 h5 h10 h12 h24 h6 h
ComplicationHaemoperitoneumHaemoperitoneumAbominal wallhaematomaAbdominalwallhaematomaHaemoperitoneumHaemoperitoneumAbdominalwallhaematomaHaemoperitoneumHaemoperitoneum
Death(interval)Yes, sepsis(18 days)Yes, HE(31 days)Yes, HE afterTIPS (19 days)NoNoYes, haemorrhagicshock (10 h)Yes, sepsis(8 days)Yes, sepsis(6 days)Yes, sepsis(9 days)

Every case with a haemorrhagic complication occurred in hospitalized patients with severe liver disease. Indeed, the mean MELD score of these patients was 30 ± 13 and the mean Child-Pugh score 11 ± 2. Coagulation parameters were moderately abnormal with a mean platelet count of 102 ± 37 and a mean INR of 2.0 ± 0.9. Significant renal dysfunction was observed in almost every patient. Mean creatinine clearance was 37 ± 17 mL/min and mean serum urea was 17 ± 7 mmol/L. Three patients were undergoing haemodialysis. In every case where portal pressure was available, it was found to be elevated. The patient who had the lowest MELD score and the slightest renal dysfunction (case no. 4) had received heparin and aspirin after paracentesis because of cardiac chest pain. All but one paracentesis were performed by experienced operators.

Haemorrhagic complications presented with a wide variety of symptoms but most patients had abdominal pain. Two-thirds of the complications consisted of haemoperitoneum and the other third, abdominal wall haematomas. Symptoms appeared within an average of 11 ± 8 h following the procedure. Diagnosis was made either by performance of a repeat paracentesis in the cases of haemoperitoneum or by radiological procedures [ultrasound and/or computed tomography (CT) scan] for abdominal wall haematomas.

Every case but one was managed conservatively by administration of blood products (FFP, platelets, packed red blood cells), appropriate medication (DDAVP, amines) and volume. In the only patient who underwent surgery for control of bleeding, small venules in the abdominal wall were actively bleeding.

Despite control of the haemorrhage in all but one case (case no. 6), death occurred during the course of the hospital stay in an additional six patients (almost 80%). Mortality was related to the underlying liver disease and was not a direct result of the procedure.


Considered a safe and cost-effective procedure, abdominal paracentesis is frequently performed in hospitalized patients with liver failure, even in the presence of coagulopathy and thrombocytopenia. Haemorrhage is usually a rare complication. We report herein the largest case series of such complication, its incidence and death rate.

Several mechanisms have been proposed to explain the haemorrhagic complications following abdominal paracentesis. First, bleeding from direct puncture of a superficial abdominal wall vein or from mesenteric varices has been hypothesized and sometimes demonstrated, as in our first case.18 It has also been suggested that haemorrhage might occur from mesenteric variceal rupture precipitated by sudden release of abdominal wall pressure following paracentesis in the face of an elevated intravariceal pressure.15, 19 In our five patients where these data were available, portal pressure was elevated but there was no definitive evidence that bleeding was variceal.

Although most reports detected symptoms during the first 6 to 48 h after paracentesis, delayed symptoms up to 1 week after the procedure have also been described.18, 20 In our experience, most patients present with abdominal pain, distension and haemodynamic instability. The diagnosis of such complications requires the performance of a new diagnostic paracentesis, abdominal ultrasound and/or CT scan. As demonstrated in our cases, most bleeding can be handled by medical treatment, such as fluid resuscitation and correction of the coagulation disorders. Laparoscopy with vessel ligature should be considered when haemodynamic instability persists despite medical treatment.20 Alternatively, transjugular intrahepatic portal-systemic shunt has also been used.15 Clinical evidence of fibrinolysis or disseminated intravascular coagulation should preclude paracentesis, but none of our patient met these criteria.

Two studies showed a potentially higher risk of bleeding in liver patients with renal failure and suggested that this context might require prophylactic administration of FFP.5, 14 In our study, the majority of the patients who presented a hemorrhagic complication had and estimated serum creatinine clearance under 50 mL/min prior to paracentesis. However, our data are insufficient to support the systematic prophylactic administration of either DDAVP or FFP in these patients.

Our reports also confirm that hemorrhagic complications do not necessarily occur in the setting of severe thrombocytopenia and/or prolonged coagulation time and that hematologic preparation is not mandatory in patients with these types of abnormalities alone.10, 11 Indeed, only two of nine patients who bled presented platelets levels under 50 000 or INR >2 just prior to the event. The study recently published by Grabau et al.11 clearly underlines the safety of performing paracentesis using a standardized procedure in stable patients followed in the out-patient clinic. Our report shows that complications do occur in the more severely ill patients who require in-hospital care and who have renal dysfunction.

In conclusion, severe haemorrhage after abdominal paracentesis is a rare (0.19%) but potentially lethal complication. We report death following this complication in 0.02% of all paracenteses performed. Most haemorrhages occurred in severely ill patients and were rarely seen in patients with severe thrombocytopenia and/ore elevated INR. All but one occurred in the context of some degree of renal failure. The management of this complication can be conservative. Short-term prognosis is nonetheless poor and seems related to the underlying liver disease. Utility of the systematic prophylactic administration of agents that improve platelet function in patients with renal failure requiring abdominal paracentesis could be assessed prospectively. Unfortunately, this would be difficult to realize because of the very large patient sample required.