Quality of life in refractory ascites: Transjugular intrahepatic portal-systemic shunting versus medical therapy


  • Potential conflict of interest: Nothing to report.


Uncontrolled studies suggest that transjugular intrahepatic portal-systemic shunting (TIPS) may improve quality of life in patients with refractory ascites. We hypothesized that any improvement of quality of life in patients with TIPS would be matched in controls due to the competing effects of improved ascites and worsened hepatic encephalopathy. Thus, an analysis of quality of life was performed using original data from the North American Study for the Treatment of Refractory Ascites, a multicenter trial of 109 patients randomized to TIPS or repeated large volume paracentesis (LVP) for refractory ascites. Short form 36 (SF-36) surveys were completed at baseline and at 6- and 12-month follow-up. Variables analyzed were: randomization group, number of LVP performed, cumulative volume from LVP, shortness of breath, abdominal distention, abdominal pain, diuretic usage, confusion, hospitalizations, and emergency room visits. Outcomes were changes in physical component scale (PCS) and mental component scale (MCS) of SF-36 results. We constructed multivariable, mixed effects models, including randomization group and baseline MCS and PCS. Changes in PCS and MCS from baseline were similar between the two randomization groups. In multivariate analysis, PCS improvement was associated with lack of confusion, improved ascites, and lack of hospitalizations, but not directly with randomization group. Improvement in MCS was associated with randomization to TIPS and lack of confusion. In conclusion, patients with refractory ascites randomized to TIPS or repeated LVP had similar changes in quality of life. Competing effects of hepatic encephalopathy, requirement for repeated LVP, and need for hospitalizations explain similar changes in quality of life between the two groups. (HEPATOLOGY 2005.)

Transjugular intrahepatic portal-systemic shunting (TIPS) is increasingly used for the treatment of refractory ascites. Most randomized trials have not demonstrated a survival advantage for TIPS over medical therapy.1–5 Therefore, any advantage that TIPS might have in this setting would lie in differences of cost and quality of life. Features which have been associated with poorer quality of life in patients with advanced liver disease include: ascites, requirement for hospitalizations, number and types of medications, Child-Turcotte-Pugh class, hepatic encephalopathy, muscle cramps, and pruritus.6–9 Quality of life is likely not affected by gender, race, and etiology of liver disease.6–9 Conflicting data on the association between age and quality of life have been reported.6–8

TIPS performed for the treatment of refractory ascites might be expected to improve quality of life. A successful TIPS may eliminate or minimize the requirement for diuretics and the need to undergo repeated large volume paracenteses (LVP). Patients might experience less shortness of breath, early satiety, and gastrointestinal bleeding, as well as fewer emergency department visits. In fact, limited, uncontrolled experience has suggested improved quality of life after performing TIPS for ascites.10–12 On the other hand, some patients may suffer deterioration in their quality of life due to hepatic encephalopathy, diarrhea and discomfort from lactulose, and repeated testing to ensure TIPS patency, including invasive venography procedures. Furthermore, TIPS may fail to resolve ascites in up to one half of cases.1–5

The present study was performed to determine differences in quality of life between patients undergoing TIPS versus repeated LVP for the treatment of refractory ascites. Clinical variables were correlated with changes in quality of life.


TIPS, transjugular intrahepatic portal-systemic shunting; LVP, large volume paracentesis; SF-36, short form 36; PCS, physical component scale; MCS, mental component scale.

Patients and Methods

A secondary analysis was performed of original patient data from the North American Study for Treatment of Refractory Ascites, an international, multicenter, prospective, randomized controlled trial.1 The study included 109 subjects with refractory ascites, stringently defined according to International Ascites Club criteria.13 Removal of at least 10 L of ascites in the preceding 2 months for symptom relief was required for entry into the study. Furthermore, patients had either failed to respond to a maximum of 160 mg/d of furosemide and 400 mg/d of spironolactone (or equivalent doses of loop-acting and distal-acting diuretics), or they could not tolerate high-dose diuretics because of hyponatremia, hyperkalemia, azotemia, or encephalopathy. All patients were on dietary sodium restriction between 50 and 66 meq/d. Patients included in the study had serum creatinine levels of less than 1.5 mg/dL for at least 7 days before study entry, as well as total bilirubin levels of less than 5 mg/dL and an international normalized ratio of less than 2. Subjects were randomized to either repeated LVP and intravenous albumin infusion (n = 57) or to TIPS (n = 52). Patients randomized to the TIPS arm underwent LVP at the time of TIPS. All subjects remained on a sodium-restricted diet and diuretics. LVP was performed as necessary during study follow-up for recurrence of tense, symptomatic ascites in patients with a weight gain of more than 10 lbs. (4.5 kg) from immediately previous nadir weight despite maximal diuretic therapy, or if patients had recurrence and were unable to use an effective dose of diuretics as a result of diuretic-related side effects. TIPS was placed successfully in 49 patients (94%). Assisted stent patency at 1 year was more than 90% in surviving subjects.

Extensive and carefully documented clinical, laboratory, and radiological data were collected during the course of the study. Over the course of 1 year patients had at least 10 follow-up visits per protocol. At each visit, investigators assessed whether subjects had experienced the following interval symptoms: shortness of breath, abdominal distention, abdominal pain, and confusion. Other variables assessed during follow-up were visits to the emergency department, hospitalizations, requirement for LVP, number of LVP performed, quantity of ascites fluid removed, TIPS thrombosis, gastrointestinal bleeding, requirement for lactulose, and diuretic dosage. Clinical outcomes were reported for the first and second 6 months of follow-up. If a patient was symptomatic during at least one visit, that patient was counted as having experienced the symptom during the corresponding 6-month period. Diuretic dosage at the end of each 6-month period was reported. Dosages were considered low if patients were taking no more than 40 mg/d furosemide or 100 mg/d of spironolactone (or up to 1 mg/d bumetanide and 5 mg/d amiloride). Dosages were considered high if patients were taking 120 mg/d or more of furosemide or 300 mg/d or more of spironolactone (or 3 mg/d or more of bumetanide or 15 mg/d or more of amiloride). Patients were censored from the study at the time of death or liver transplantation.

Short form 36 (SF-36) questionnaires were administered to all patients at baseline randomization and at 6- and 12-month follow-up. The SF-36 is a general quality of life assessment with 36 questions divided into eight scales: (1) physical function, (2) physical role limitation, (3) bodily pain, (4) general health, (5) vitality, (6) social functioning, (7) emotional role limitation, and (8) mental health. Two standardized indices, the physical component scale (PCS) and the mental component scale (MCS), can be calculated from SF-36 raw data.14 Ranges for the MCS and PCS are 8 to 73 and 10 to 74, respectively. The SF-36 is widely used and has been validated in diverse disease states.14 It has also proven useful for the evaluation of quality of life in patients with advanced liver disease.6, 7, 15, 16 In fact, the SF-36 is a standard to which new liver disease quality of life assessments are compared.16

Statistical Analysis.

Parametric variables were reported as means and standard deviations, and nonnormally distributed variables were reported as quartiles and ranges. Tests for equality of proportions were used for comparing dichotomous variables between randomization groups. Mixed effects models, including baseline PCS or MCS score, assessed the association between clinical variables and change in mean PCS or MCS, looking for effect modification by time of follow-up (6- or 12-month). Multivariable mixed effects models were generated via stepwise backward elimination, using a P value greater than .05 to remove variables not significantly associated with outcome. All univariate variables with P values less than .20 were considered for multivariate analyses. A randomization group was also included in each multivariate model. Baseline characteristics (age, sex, race, and cause of liver disease) were similar between the two randomization groups.1 Statistical programs used were SAS 8.2 (SAS Institute, Cary, NC) and Stata 8.1 (Stata Corporation, College Station, TX). A P value of less than .05 was considered significant.


One hundred six (97%) patients completed the baseline SF-36 survey. At 6-month follow-up, 56 of 64 patients (88%) completed a follow-up SF-36, and 51 of 54 patients (93%) were surveyed at 12 months (Fig. 1). Reasons for failure to complete the survey were not obtained. Fifty-four subjects exited the study at various time ponts (26 deaths, 27 liver transplantations, and 1 voluntary withdrawal). Baseline and follow-up survey data were available for a similar proportion of patients in the TIPS and LVP groups (P = .27).

Figure 1.

Patient flow diagram. QOL, quality of life; TIPS, transjugular intrahepatic portal-systemic shunt group; LVP, large volume paracentesis group

Clinical characteristics differed between TIPS and LVP groups during follow-up (Table 1). Patients randomized to TIPS required fewer LVP with less cumulative ascites fluid removal. TIPS patients also required lower diuretic dosages. Furthermore, TIPS patients had lower rates of shortness of breath and abdominal distention, but higher rates of confusion and hospitalizations.

Table 1. Clinical Characteristics of Each Randomization Group During First and Second 6 Months of Follow-up
VariableTIPS Group: Number (%) or Median (25%, 75%, range)*LVP Group: Number (%) or Median (25%, 75%, range)*P Value
  • *

    Only patients with complete follow-up for each time period are considered.

  • Significant at P = .05 level.

LVP required   
 0–6 months10 (27%)29 (88%)<.001
 6–12 months2 (7%)17 (63%)<.001
Number LVP   
 0–6 months0 (0, 1, 0-14)4 (2, 7, 0-12)<.001
 6–12 months0 (0, 0, 0-8)1 (0, 7, 0-16)<.001
Cumulative ascites (L)   
 0–6 months0 (0, 2, 0-116)17 (9, 52, 0-130)<.001
 6–12 months0 (0, 0, 0-117)7 (0, 54, 0-132)<.001
Shortness of breath   
 0–6 months16 (43%)26 (79%).002
 6–12 months7 (23%)15 (56%).01
Abdominal distention   
 0–6 months31 (84%)32 (97%).07
 6–12 months9 (30%)22 (81%)<.001
Diuretic usage   
 0–6 months  .01
  No diuretics9 (24%)4 (13%) 
  Low dose19 (51%)7 (23%) 
  Moderate dose5 (14%)8 (26%) 
  High dose4 (11%)12 (39%) 
 6–12 months  .006
  No diuretics5 (19%)6 (24%) 
  Low dose13 (48%)6 (24%) 
  Moderate dose7 (26%)7 (28%) 
  High dose2 (7%)6 (24%) 
Abdominal pain   
 0–6 months27 (73%)23 (70%).76
 6–12 months15 (50%)14 (52%).88
 0–6 months18 (49%)8 (24%).03
 6–12 months9 (30%)6 (22%).51
Taking lactulose   
 0–6 months31 (84%)24 (73%).26
 6–12 months24 (80%)17 (65%).22
 0–6 months21 (57%)8 (24%).006
 6–12 months11 (37%)10 (38%).89
Emergency department visits   
 0–6 months15 (41%)14 (42%).87
 6–12 months9 (30%)6 (23%).56
Gastrointestinal bleeding   
 0–6 months0 (0%)3 (9%).06
 6–12 months1 (3%)3 (12%).25

Age, sex, race, and cause of liver disease were not associated with statistically significant changes in PCS and MCS. Changes in PCS and MCS during follow-up were not significantly different between the two groups (Table 2), although there may have been a trend toward greater quality of life improvement in MCS after TIPS (P = .06). On univariate analysis, significant predictors of change in PCS were recurrent ascites, severe recurrent ascites (three or more LVP or >15 L cumulative ascites removed), confusion, and shortness of breath. Presence of abdominal distention and requirement for hospitalization were significant predictors only at 12-month follow-up (Table 3). On multivariate analysis, randomization to TIPS group was not a significant predictor of PCS. Severe ascites (three or more LVP), confusion, and need for hospitalizations by the 12-month visit were associated with declines in PCS (Table 4).

Table 2. Changes in PCS and MCS by Treatment Group
 GroupBaseline6 Months*12 Months*P Value
  • *

    Change compared with baseline.

  • TIPS versus LVP change from baseline.

PCSTIPS28.4 ± 8.47.0 ± 8.45.0 ± 10.5.29
 LVP26.8 ± 9.36.3 ± 9.32.0 ± 11.7 
MCSTIPS44.7 ± 12.63.7 ± 8.33.3 ± 12.6.06
 LVP41.0 ± 10.33.4 ± 12.70.5 ± 11.9 
Table 3. Univariate Predictors of PCS
VariableCoefficient (CI)*P Value
  • *

    P = .05 is considered significant.

Randomization to TIPS2.14 (−1.82, 6.10).29
Requirement for LVP−4.22 (−8.02, −0.42).03*
Three or more LVP−5.41 (−9.00, −1.81).005*
More than 15 L cumulative ascites removed−5.68 (−9.63, −1.73).007*
Shortness of breath−3.68 (−7.21, −0.15).05*
Abdominal distention  
 At 6-month visit4.08 (−4.16, 12.33).34
 At 12-month visit−5.82 (−10.41, −1.22).02*
Abdominal pain−0.57 (−4.18, 3.05).76
Confusion−4.16 (−7.91, −0.41).03*
Taking lactulose−0.40 (−4.70, 3.90).86
 At 6-month visit2.96 (−1.36, 7.29).19
 At 12-month visit−7.00 (−11.99, −2.01).008*
Emergency department visit0.54 (−2.99, 4.08).76
Diuretics (any vs. none)1.56 (−2.86, 5.98).49
Diuretics (dosage effect)0.35 (−1.39, 2.08).70
Table 4. Predictors of PCS, Multivariate Analysis
VariableCoefficient (CI)*P Value
  • *

    P = .05 is considered significant.

Baseline PCS−0.53 (−0.72, −0.33)<.001*
Randomization to TIPS0.31 (−3.74, 4.36).88
Three or more LVP−5.12 (−9.14, −1.09).02*
Confusion−4.10 (−7.73, −0.47).03*
 At 6 months1.63 (−2.97, 6.23).49
 At 12 months−5.55 (−10.10, −0.99).02*

On univariate analysis, no clinical variable was associated with change in MCS scores (Table 5). On multivariate analysis, TIPS patients had improved quality of life after adjusting for the negative effect of confusion on MCS changes (Table 6).

Table 5. Univariate Predictors of MCS
VariableCoefficient (CI)*P Value
Randomization to TIPS4.59 (−0.05, 9.23).06
Requirement for LVP−2.96 (−7.82, 1.90).24
Three or more LVP−2.79 (−7.45, 1.87).25
More than 15 L cumulative ascites removed−3.00 (−8.05, 2.06).25
Shortness of breath−1.81 (−6.53, 2.90).45
Abdominal distention2.67 (−2.32, 7.65).30
Abdominal pain0.54 (−3.76, 4.83).81
Confusion−1.99 (−6.61, 2.63).40
Taking lactulose0.39 (−4.74, 5.52).88
Hospitalizations−0.18 (−4.29, 3.94).93
Emergency department visit−0.55 (−4.87, 3.77).80
Diuretics (any vs. none)−2.50 (−7.79, 2.79).36
Diuretics (dosage effect)−0.60 (−2.67, 1.47).57
Table 6. Predictors of MCS, Multivariate Analysis
VariableCoefficient (CI)*P Value
  • *

    P = .05 is considered significant.

Baseline MCS−0.57 (−0.75, −0.40)<.001*
Randomization to TIPS5.29 (0.94, 9.64).02*
 At 6 months1.39 (−4.42, 7.20).64
 At 12 months−8.06 (−14.70, −1.43).02*


Several uncontrolled studies have suggested that patients who undergo TIPS for refractory ascites experience improved quality of life.10–12 Demonstrating a postprocedure improvement, however, is no substitute for a controlled comparison. Enrolling in a study, participating in quality of life surveys, undergoing closer medical scrutiny, and simply allowing for the passage of time may lead to improved perceptions of quality of life.17 In contrast, we analyzed quality of life data from a randomized, controlled trial between continued medical therapy and performance of TIPS for refractory ascites. Within each randomization group, quality of life improved during follow-up. Although the difference in improvement between the groups was not statistically significant, there was a trend toward enhanced quality of life in MCS after TIPS.

Baseline measurements of quality of life in our study population were comparable with those from previous studies.6–8 For example, Arguedas et al.7 found that in individuals with Child-Turcotte-Pugh class C cirrhosis, mean PCS and MCS were 28 ± 11 and 43 ± 13—results that closely match our data. To put these numbers in perspective, mean PCS and MCS in individuals with Child-Turcotte-Pugh class A cirrhosis, individuals with chronic obstructive pulmonary disease, and the general United States population are 35 ± 13 and 45 ± 12, 36 ± 10 and 48 ± 11, and 50 ± 10 and 50 ± 10, respectively.7, 8 Therefore, the increases in PCS and MCS of 0.5 to 7 points demonstrated in our study should be considered a modest improvement.

Randomization to TIPS as compared with repeated LVP was associated with greater improvement in the mental component scale of the SF-36 after adjusting for the development of confusion in some subjects. Patients who achieve ascites control after TIPS may believe they are in better health. They no longer experience repeated reaccumulation of ascites fluid and thus are not dependent on repeated LVP. Furthermore, some patients may have a better self-image and fit better into their usual clothes. They may also have improved appetite and breathing mechanics. Nutritional status and pleasure derived from eating may also increase. However, if encephalopathy develops, these possible benefits may be negated. If the development of hepatic encephalopathy after TIPS could be predicted accurately, then TIPS could be reserved for those patients who would experience a quality of life benefit.

Randomization to TIPS was not directly associated with changes in the physical component scale. Instead, predictors of improvement in the physical component scale were: lack of severe ascites, confusion, and requirement for hospitalization. Logically, these three predictors of quality of life changes are likely affected by the performance of TIPS. Therefore, although TIPS has no direct impact on changes in PCS, it seems to cause changes in PCS indirectly through its effects on these three clinical variables.

We identified severe ascites, confusion, and hospitalizations as the variables most associated with quality of life changes among patients with refractory ascites. Our findings are consistent with those of other investigators.6–8 As part of our trial protocol, patients were selected to have relatively preserved model for end-stage liver disease scores. Therefore, we could not investigate how model for end-stage liver disease score influences quality of life in this population. Other variables that may influence quality of life changes include medications, muscle cramps, and pruritus.6–9 Our database did not allow us to accurately assess changes in muscle cramps and pruritus between the two randomization groups. We did examine requirement for lactulose and found that confusion was more closely associated with quality of life changes. Diuretic dosage was not associated with changes in quality of life in our patient population.

Our analysis could be criticized because only 51 patients completed quality of life assessment at 12-month follow-up, compared with a starting number of 106. Dropout was balanced between the TIPS and LVP groups and was due to the high rate of liver transplantation and death in this study population with decompensated cirrhosis. Ninety-seven percent of enrolled patients completed baseline surveys, and 90% of patients who did not undergo interval transplantation or death completed follow-up surveys. Furthermore, survey completion rates were similar between the two randomization groups. Because the study was a randomized, controlled trial, patients were similar with respect to baseline demographic and clinical variables.

For patients with advanced liver disease, physicians and patients may have markedly different perceptions of quality of life.15, 18, 19 Therefore, objective instruments are essential to evaluate any quality-of-life benefit from TIPS. The SF-36 is a general assessment of quality of life not specifically designed for the evaluation of patients with advanced liver disease. However, this tool has been widely used for the evaluation of decompensated cirrhosis.6, 7, 15, 16 Furthermore, its widespread use in the medical literature for the evaluation of other disease states makes our findings more easily accessible.

In conclusion, patients with refractory ascites randomized to TIPS or continued medical management had similar changes in quality of life during 1 year of follow-up. Similar changes in quality of life between treatment arms were explained by competing effects of severe ascites, hospitalizations, and the development of confusion. TIPS did lead to improvement in the mental component scale, only after adjusting for the negative effects of confusion in some subjects. In the absence of the ability to predict how TIPS will affect the course of encephalopathy, ascites, and hospitalizations in the individual patient, the decision to perform TIPS cannot be justified based on expectations for improved quality of life.