Expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents for transjugular intrahepatic portosystemic shunt in people with liver cirrhosis

Abstract

Background

Transjugular intrahepatic portosystemic shunt (TIPS) is a widely used procedure for management of uncontrolled upper gastrointestinal bleeding and refractory ascites in people with liver cirrhosis. However, nearly half of the people experience shunt dysfunction and recurrent symptoms within one year of the procedure. Expanded polytetrafluoroethylene (ePTFE)‐covered stents are assumed to decrease shunt dysfunction by approximately 20% to 30%.

Objectives

To evaluate the benefits and harms associated with the use of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents in transjugular intrahepatic portosystemic shunts (TIPSs) for managing people with liver cirrhosis.

Search methods

We used standard, extensive Cochrane search methods. The latest search date was 28 February 2023.

Selection criteria

Randomised clinical trials comparing ePTFE‐covered stents versus bare stents in TIPS for treatment of people with liver cirrhosis.

Data collection and analysis

We used standard Cochrane methods. Our primary outcomes were 1. all‐cause mortality, 2. procedure‐related complications, and 3. health‐related quality of life. Our secondary outcomes were 4. upper gastrointestinal bleeding, 5. recurrence of ascites, 6. hepatic encephalopathy, 7. kidney failure, 8. early thrombosis, 9. non‐serious adverse events, and 10. shunt dysfunction. We used GRADE to assess certainty of evidence.

We analysed outcome data at the maximum follow‐up, except for the 'early thrombosis' outcome for which it was within 12 weeks after the TIPS procedure.

Main results

We included four trials with 565 randomised participants (age range: 18 to 75 years; male range: 63.6% to 75.0%). A total of 527 participants provided data for analyses because of losses to follow‐up. Two trials were conducted in China; one in France; and one in France, Spain, and Canada. Participants were classified with cirrhosis Child‐Pugh class A, B, or C, and for some, the class was not reported. We used intention‐to‐treat principle (four trials) and per‐protocol analysis (one trial) to meta‐analyse the data.

One trial compared ePTFE‐covered stents versus bare stents of the same diameter and three trials compared ePTFE‐covered stents versus stents of different diameters.

ePTFE‐covered stents versus bare stents of the same diameter

One trial with 258 participants compared 8 mm covered stent versus 8 mm bare stent.

Mortality in the covered stent group is possibly lower than in the bare stent group (risk ratio (RR) 0.63, 95% confidence interval (CI) 0.43 to 0.92; low‐certainty evidence). Upper gastrointestinal bleeding (RR 0.54, 95% CI 0.35 to 0.84), recurrence of ascites (RR 0.42, 95% CI 0.20 to 0.87), and shunt dysfunction (RR 0.42, 95% CI 0.28 to 0.61) occurred more often in the bare stent group than in the covered stent group (all low‐certainty evidence). There was no difference in hepatic encephalopathy between groups (RR 1.10, 95% CI 0.76 to 1.61; very low‐certainty evidence). The trial did not report data on procedure‐related complications, health‐related quality of life, early thrombosis, and segmental liver ischaemia (a non‐serious adverse event).

ePTFE‐covered stents versus bare stents of different stent diameters

Three trials compared ePTFE‐covered stents versus bare stents of different diameters (10.5 (standard deviation (SD) 0.9) mm versus 11.7 (SD 0.8) mm; 8 mm versus 10 mm; and one trial used 10‐mm stents that could be dilated from 8 mm to 10 mm).

There was no evidence of a difference between the ePTFE‐covered stents versus bare stents groups in mortality (RR 0.75, 95% CI 0.48 to 1.16; 3 trials, 269 participants), procedure‐related complications (RR 0.53, 95% CI 0.05 to 5.57; 1 trial, 80 participants), upper gastrointestinal bleeding (RR 0.46, 95% CI 0.15 to 1.38; 3 trials, 269 participants), hepatic encephalopathy (RR 0.93, 95% CI 0.66 to 1.30; 3 trials, 269 participants), and kidney failure (RR 7.59, 95% CI 0.40 to 143.92; 1 trial, 121 participants) (all very low‐certainty evidence). Recurrence of ascites (RR 0.30, 95% CI 0.11 to 0.85; 3 trials, 269 participants; low‐certainty evidence), shunt dysfunction (RR 0.50, 95% CI 0.28 to 0.92; 3 trials, 269 participants; low‐certainty evidence), and early thrombosis (RR 0.28, 95% CI 0.09 to 0.82; I² = 0%; 3 trials, 261 participants; very low‐certainty evidence) occurred more often in the bare stents group. There was no evidence of a difference in segmental liver ischaemia (RR 5.25, 95% CI 0.26 to 106.01; 1 trial, 80 participants; very low‐certainty evidence). No trial presented data on health‐related quality of life.

Funding

One trial did not clearly report funding sources. The remaining three trials declared that they had no funding with vested interests.

Authors' conclusions

Based on the small number of trials with insufficient sample size and events, and study limitations, we assessed the overall certainty of evidence in the predefined outcomes as low or very low. Therefore, we are uncertain which of the two interventions (ePTFE‐covered stents or bare stents of the same diameter and ePTFE‐covered stents versus bare stents of different stent diameters) is effective for the evaluated outcomes. None of the four trials reported data on health‐related quality of life, and data on complications were either missing or rarely reported.

We lack high‐quality trials to evaluate the role of ePTFE‐covered stents for TIPS for managing people with liver cirrhosis.

Plain language summary

What are the benefits and risks of covered stents compared with conventional bare stents for transjugular intrahepatic portosystemic shunt in people with liver cirrhosis?

Key messages

– Expanded polytetrafluoroethylene (ePTFE)‐covered stents may be more likely to reduce mortality (deaths), upper gastrointestinal bleeding, recurrence of ascites (a condition where fluid collects in spaces within your abdomen), and shunt dysfunction compared with bare stents of the same diameter.

– Beneficial effects of ePTFE‐covered stents remain unclear when comparing with bare stents of different diameters.

Technical terms: ePTFE

ePTFE is a variation of polytetrafluoroethylene (PTFE) that has been modified to have a porous structure. The material is created by stretching PTFE, which forms a network of interconnected nodes and fibrils (fibres). This porous structure allows the passage of fluids and gases through the material while retaining the desirable properties of PTFE, such as its chemical resistance.

What is liver cirrhosis?

Liver cirrhosis is a condition where healthy liver tissue is gradually replaced by scar tissue. This scarring can affect the liver's ability to function properly and can be caused by various factors such as chronic alcohol abuse, viral infections, or other diseases.

What is portal hypertension?

People with liver cirrhosis often develop a complication called portal hypertension, which is high blood pressure in the blood vessels that supply the liver. This can lead to serious complications such as internal bleeding, fluid accumulation, and enlargement of the veins in the oesophagus (food pipe) and stomach.

How is portal hypertension treated?

To treat portal hypertension, a procedure called transjugular intrahepatic portosystemic shunt (TIPS) is often performed. During TIPS, a stent (a short tube) is inserted to create a pathway between the portal vein and hepatic vein, redirecting the blood flow and reducing pressure. Unfortunately, the conventional bare stents used in the TIPS procedure are sometimes blocked within one year. Researchers have found that a new type of stent, covered with ePTFE, may help prevent this blockage.

What did we want to find out?

We wanted to find out if the ePTFE‐covered stents were better than bare stents in TIPS in people with liver cirrhosis to improve mortality, procedure‐related complications, and quality of life.

What did we do?

We searched for studies that looked at ePTFE‐covered stents compared with bare stents for TIPS in people with liver cirrhosis. We compared and summarised the results of the studies and rated our confidence in the evidence, based on factors such as study methods and sizes.

What did we find?

We found four studies that involved 565 people with liver cirrhosis. We found that using ePTFE‐covered stents yielded better results in terms of death from all causes (all‐cause mortality), upper gastrointestinal bleeding, recurrence of ascites, and shunt dysfunction when using the same diameter as bare stents. There were no difference in the occurrence of hepatic encephalopathy (a condition that affects mental and neurological function) in the two groups. However, there were no differences between ePTFE‐covered stents and bare stents of different diameters on number of deaths, procedure‐related complications, upper gastrointestinal bleeding, hepatic encephalopathy, kidney failure, or shunt dysfunction. There was a reduced number of people with recurrence of ascites in the ePTFE‐covered group compared to the bare stents of different diameters.

What are the limitations of the evidence?

We have very little or no confidence in the evidence mainly because studies were very small and people in the studies might have been aware of which treatment they were getting.

Funding

One study did not clearly report funding sources. The remaining three studies declared that they had no funding with vested interests.

How up to date is this evidence?

The evidence is up to date to February 2023.

Summary of findings

Summary of findings 1. ePTFE‐covered stents versus bare stents for transjugular intrahepatic portosystemic shunt (TIPS) in people with liver cirrhosis – same stent diameter.

ePTFE‐covered stents versus bare stents for transjugular intrahepatic portosystemic shunt in people with liver cirrhosis – same stent diameter
Patient or population: transjugular intrahepatic portosystemic shunt in people with liver cirrhosis Setting: interventional ward Intervention: ePTFE‐covered stents Comparison: bare stents
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (trials)	Certainty of the evidence (GRADE)	Comments
	Risk with bare stents	Risk with ePTFE‐covered stents
All‐cause mortality Follow‐up: 5 years	Study population		RR 0.63 (0.43 to 0.92)	258 (1 RCT)	⊕⊕⊝⊝ Lowa,b	—
	378 per 1000	238 per 1000 (163 to 348)
Upper gastrointestinal bleeding Follow‐up: 5 years	Study population		RR 0.54 (0.35 to 0.84)	258 (1 RCT)	⊕⊕⊝⊝ Lowa,b	—
	339 per 1000	183 per 1000 (119 to 284)
Recurrence of ascites Follow‐up: 5 years	Study population		RR 0.42 (0.20 to 0.87)	258 (1 RCT)	⊕⊕⊝⊝ Lowa,b	—
	165 per 1000	69 per 1000 (33 to 144)
Hepatic encephalopathy Follow‐up: 5 years	Study population		RR 1.10 (0.76 to 1.61)	258 (1 RCT)	⊕⊝⊝⊝ Very lowa,c	—
	283 per 1000	312 per 1000 (215 to 456)
Shunt dysfunction Follow‐up: 5 years	Study population		RR 0.42 (0.28 to 0.61)	258 (1 RCT)	⊕⊕⊝⊝ Lowa,b	—
	496 per 1000	208 per 1000 (139 to 303)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; ePTFE: expanded polytetrafluoroethylene; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate. The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited. The true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate. The true effect is likely to be substantially different from the estimate of effect.

^aDowngraded one level for risk of bias (the trial was rated at high risk of attrition bias and unclear risk of performance bias).
^bDowngraded one level for imprecision (although the 95% CIs were within the threshold of clinical appreciable benefit, the sample size was smaller than the optimal information size).
^cDowngraded two levels for imprecision (the sample size was smaller than the optimal information size; 95% CIs around estimate of effect were wide and included no effect and appreciable benefit/harm).

Summary of findings 2. ePTFE‐covered stents versus bare stents for transjugular intrahepatic portosystemic shunt in people with liver cirrhosis – different stent diameters.

ePTFE‐covered stents versus bare stents for transjugular intrahepatic portosystemic shunt (TIPS) in people with liver cirrhosis – different stent diameters
Patient or population: transjugular intrahepatic portosystemic shunt in people with liver cirrhosis Setting: interventional ward Intervention: ePTFE‐covered stents Comparison: bare stents
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (trials)	Certainty of the evidence (GRADE)	Comments
	Risk with bare stents	Risk with ePTFE‐covered stents
All‐cause mortalitya Follow‐up: 8–24 months	Study population		RR 0.75 (0.48 to 1.16)	269 (3 RCTs)	⊕⊝⊝⊝ Very lowb,c	—
	333 per 1000	250 per 1000 (160 to 387)
Procedure‐related complications Follow‐up: 2 years	Study population		RR 0.53 (0.05 to 5.57)	80 (1 RCT)	⊕⊝⊝⊝ Very lowb,c	—
	49 per 1000	26 per 1000 (2 to 272)
Upper gastrointestinal bleeding Follow‐up: 8–24 months	Study population		RR 0.46 (0.15 to 1.38)	269 (3 RCTs)	⊕⊝⊝⊝ Very lowb,c	—
	80 per 1000	37 per 1000 (12 to 110)
Recurrence of ascites Follow‐up: 8–24 months	Study population		RR 0.30 (0.11 to 0.85)	269 (3 RCTs)	⊕⊕⊝⊝ Lowb,d	—
	123 per 1000	37 per 1000 (14 to 105)
Hepatic encephalopathy Follow‐up: 8–24 months	Study population		RR 0.93 (0.66 to 1.30)	269 (3 RCTs)	⊕⊝⊝⊝ Very lowb,c	—
	333 per 1000	310 per 1000 (220 to 433)
Kidney failure Follow‐up: 2 years	Study population		RR 7.59 (0.40 to 143.92)	121 (1 RCT)	⊕⊝⊝⊝ Very lowb,c	—
	0 per 1000	0 per 1000 (0 to 0)
Shunt dysfunctiona Follow‐up: 8–24 months	Study population		RR 0.50 (0.28 to 0.92)	269 (3 RCTs)	⊕⊕⊝⊝ Lowb,d	—
	536 per 1000	268 per 1000 (150 to 493)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; ePTFE: expanded polytetrafluoroethylene; HR: hazard ratio; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate. The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited. The true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate. The true effect is likely to be substantially different from the estimate of effect.

^a We calculated both HR and RR. However, we reported RR since data were more informative for this outcome; out of the three trials, only one provided data with HR.
^b Downgraded one level for risk of bias (trials were at high risk of performance bias or detection bias, or both).
^c Downgraded two levels for imprecision: sample size was smaller than optimal information size; 95% CIs around estimate of effect were wide and included no effect and appreciable benefit/harm.
^d Downgraded one level for imprecision: although the 95% CIs were within the threshold of clinical appreciable benefit, the sample size was smaller than the optimal information size.

Background

Description of the condition

Liver cirrhosis is characterised by diffuse nodular regeneration and changes of the intrahepatic vascular bed. Alcohol fatty liver disease and hepatitis C virus infection are the main causes of cirrhosis in Western countries. The prevalence of non‐alcoholic steatohepatitis is rising and may become the most common cause of cirrhosis. In Asia, especially China, and in Saharan Africa, hepatitis B virus is the most common cause of cirrhosis (Tsochatzis 2014). Without treatment of the underlying cause of the liver disease, complications will ensue, including portal hypertension and its sequelae. In 2016, cirrhosis ranked as the 11th most common cause of death and the 15th leading cause of illness, contributing to approximately 2.2% of global deaths and 1.5% of disability‐adjusted life years (Global Health Estimates 2016).

Most people with liver cirrhosis remain asymptomatic until decompensation occurs. The symptoms of cirrhosis include variceal bleeding, ascites, non‐obstructive jaundice, and hepatic encephalopathy. The diagnosis of cirrhosis can be established with the finding of a nodular liver on conventional imaging modalities, such as ultrasound or computed tomography, along with impaired serological liver function tests. Liver biopsy, an invasive procedure, remains the 'gold standard' for diagnosing cirrhosis. Early cirrhosis may be missed by conventional imaging, necessitating liver biopsy. Newer imaging techniques, such as transient elastography, are alternative non‐invasive diagnostic tests (Tsochatzis 2014).

Primary haemorrhage from gastro‐oesophageal varices is one of the most common and lethal symptoms of cirrhosis. It occurs in approximately half of people with cirrhosis (Kovalak 2007). The development and growth of varices occurs at a rate of 7% per year (Groszmann 2005; Merli 2003). The one‐year risk of a first variceal haemorrhage is approximately 12% (D'Amico 1999). Risk factors for variceal bleeding have been reported to include large variceal size, endoscopic stigmata, and advanced liver disease (NIEC 1988). People with clinically decompensated cirrhosis (including the presence of ascites, jaundice, or symptomatic encephalopathy) are more likely to experience bleeding and to have worse outcomes (D'Amico 2014). In acute variceal haemorrhage, vasoconstrictors administered in conjunction with endoscopic therapies are recommended as the first‐line therapy, but the therapeutic benefits are debatable (Gøtzsche 2008; Roberts 2021). When this strategy fails, an urgent placement of a transjugular intrahepatic portosystemic shunt (TIPS) or potentially a Sengstaken‐Blakemore tube may stop the bleeding (de Franchis 2010; de Franchis 2015; de Franchis 2022; Roberts 2021). Early placement of a TIPS has been reported to be beneficial for people at high risk of treatment failure. Garcia‐Pagan and colleagues reported that in people with Child‐Pugh class C (Child score between 10 and 13) or Child‐Pugh class B classifications with active bleeding during endoscopy, the placement of a TIPS with covered stents within 72 hours reduced treatment failure and improved the one‐year survival rate compared with their endoscopic band ligation plus drug therapy (non‐selective beta‐blocker plus isosorbide‐5‐mononitrate) (one‐year survival: 86% for early TIPS versus 61% for endoscopic band ligation plus drug therapy) (Garcia‐Pagan 2010). Monescillo and colleagues also stated that for people with a hepatic venous pressure gradient greater than 20 mmHg, the placement of TIPS with bare stents within the first 24 hours could improve survival when compared with medical therapy (Monescillo 2004). UK guidelines have also recommended early TIPS within 72 hours for people at high risk of treatment failure after initial pharmacological and endoscopic therapy (de Franchis 2015; de Franchis 2022; Tripathi 2020). One recent meta‐analysis compared early TIPS versus current standard of care for acute variceal bleeding, with results showing that early TIPS was associated with a lower six‐week rebleeding risk, but without benefit on survival (Hussain 2022).

Refractory ascites is another common sign of decompensated liver cirrhosis and is defined as the inability to mobilise ascites despite the use of maximal doses of diuretics (Arroyo 1994). Mechanisms involved in the formation of ascites include increased sinusoidal pressure and the rising concentration of serum sodium‐retaining hormones (e.g. renin and aldosterone). Large‐volume paracentesis with albumin replacement is recommended as first‐line treatment for managing refractory ascites (Aithal 2021). In those people who are intolerant to repeated large‐volume paracentesis, TIPS placement may be adopted (Aithal 2021). One previous Cochrane Review with meta‐analysis that included five randomised clinical trials reported that TIPS was as effective at removing ascites as large‐volume paracentesis though TIPS was not associated with a significant reduction in mortality (Saab 2006).

Description of the intervention

TIPS involves inserting a stent between the hepatic vein and the portal vein, resulting in a direct reduction of portal venous pressure, thus reducing the risk of variceal bleeding (de Franchis 2015; de Franchis 2022; Tripathi 2020). TIPSs can also improve natriuresis and decrease the rate of filtration into the peritoneal space, thereby reducing ascites and hydrothorax (Aithal 2021). In addition, TIPSs may improve renal function by increasing glomerular filtration and urine output and are also associated with improved protein metabolism and nutrition, alongside improvements in quality of life (Punamiya 2011). However, a decrease in portal pressure can impair nutritive portal blood flow into the liver, which may increase the risk of liver failure. Furthermore, the increased cardiac preload can lead to heart failure in people with heart disease, especially in the context of pulmonary hypertension (Walser 2005). In addition, the risk of hepatic encephalopathy may be increased by the portosystemic shunt (Riggio 2008).

Procedure‐related complications such as bleeding, bile leakage, and infection can also occur when inserting the stents. Early or late stent occlusion is another important TIPS‐related complication. Early occlusion — within 12 weeks of stent insertion — is mainly caused by bile‐related thrombosis due to injury to the bile duct that occurred during TIPS placement (Vignali 2005). Late occlusion — occurring more than 12 weeks after placement — is mostly due to infiltration of the stent with mesenchymal cells, with resultant deposition of connective tissue along the luminal surface that produces a pseudo‐intimal layer (Vignali 2005). Stenosis may also occur at the junction of the hepatic vein and inferior vena cava due to intimal hyperplasia, which is stimulated by the increased shear stress at the end of the stent caused by high‐velocity or turbulent blood flow (Cura 2008). Approximately 25% to 50% of people with bare stents will develop stent stenosis and recurrent portal hypertension within six to 12 months of the primary TIPS procedure (Cejna 2001; Cura 2008), and the use of expanded polytetrafluoroethylene (ePTFE)‐covered stents may obviate this particular problem (Saad 2010).

The most commonly used ePTFE‐covered stent is the Viatorr stent (Gore, Flagstaff, Arizona, USA) (Hausegger 2004), which consists of a self‐expanding nitinol stent skeleton and an ePTFE‐film lining the inside of the stent lumen. To avoid being compressed by the liver parenchyma, the nitinol wire is bent into a zig‐zag configuration to increase radial strength. The stent is not fully covered by the ePTFE‐film and has a 2 cm bare region at its end. The remaining 4 cm to 6 cm of the stent is covered completely by the ePTFE‐film. Between the covered and uncovered regions, there is a circumferential radiopaque gold marker band. An additional radiopaque gold marker is embedded at the trailing edge of the device (Cejna 2001; Hausegger 2004).

The ePTFE‐film has three layers; the outer layer has large pores to allow superficial infiltration of fibrous connective tissue around the graft; the middle layer has much smaller pores and is impermeable to connective tissue and liquid bile; the innermost layer has small pores with a microstructure similar to that of the conventional vascular graft from the same manufacturer (Vignali 2005).

Another model of ePTFE‐covered stents is the Fluency stent (Bard, Tempe, Arizona, USA) (Saad 2010). Compared with the Viatorr stent, the Fluency stent is covered by the ePTFE‐film both inside and outside the lumen, and there are only 2 mm long bare regions at each end. The inner ePTFE‐film is carbon impregnated, which is designed to prevent platelet aggregation. On both ends of the stent, two radiopaque titanium markers are embedded. Fluency stents are mainly used in arterial diseases. For the TIPS procedure, a bare stent is used first and inserted into the portal vein. This is followed by the Fluency stent, which extends from the junction of the portal vein and liver parenchyma to the junction of the hepatic vein and inferior vena cava (Saad 2010).

Compared with covered stents, bare stents are those without an ePTFE film. The stent skeleton is usually composed of nitinol (an alloy of nickel and titanium), stainless steel, or another inert alloy. These stents are firm and flexible. However, it is possible for the connective tissue surrounding the stent to infiltrate the inner lumen. They are also permeable to bile.

The ePTFE‐covered stents are currently receiving increased interest in Europe and the USA. Covered stents have been recommended in guidelines (Boyer 2010; de Franchis 2015). Although covered stents are more expensive than their bare equivalent, studies suggest that they may be more cost‐effective because people with covered stents experience a reduction in clinical relapses as well as a decreased need for shunt revisions (Bercu 2015; Vignali 2005).

A multicentre clinical trial (NIHR130883) has been launched to clarify the effects of early TIPS (fundingawards.nihr.ac.uk/award/NIHR130883).

How the intervention might work

The ePTFE‐film on the covered stent is biocompatible, non‐thrombogenic, and microporous. As the film is impermeable to bile, early stenosis caused by bile‐related thrombosis can be prevented. In addition, the ePTFE covering can prevent infiltration of connective tissue and remain smooth, preventing the late occurrence of stenosis caused by connective tissue hyperplasia along the lumen. Placing the stent up to the junction of the hepatic vein and inferior vena cava reduces sheer stress around the junction, thereby preventing intimal hyperplasia at that location. With the use of ePTFE‐covered stents, one‐year primary patency rates could reach 80% to 84% (Cejna 2001; Cura 2008; Vignali 2005).

Why it is important to do this review

Many graft materials have been used for TIPSs, including silicone, polycarbonate urethane, and polyethylene terephthalate. None of these graft materials have been shown to result in better shunt patency proportions than those obtained with bare stents (Cura 2008; Haskal 1999a; Haskal 1999b; Tanihata 1997). Only ePTFE‐covered stents have been reported to have better shunt patency than that of bare stents. However, in recent years, ePTFE vascular grafts were reported to cause some serious adverse events, such as graft migration, disorder of the coagulation system, peri‐graft seroma, tissue inflammation, and fibrosis (Hsu 2017; Jung 2020; Kavalam 2019; Reyes Valdivia 2021; Shan 2021). Whether ePTFE‐covered stents are safe for people with cirrhosis during TIPS procedure needs to be elucidated. One meta‐analysis compared covered stents with bare stents in terms of maintaining the primary patency but included mainly non‐randomised and retrospective research (Yang 2010). Another meta‐analysis reported that PTFE‐covered stent grafts are associated with a better primary patency and survival and lesser rate of rebleeding than bare stents in people undergoing TIPS procedures. Yet, there is no difference in new‐onset hepatic encephalopathy between covered and bare stents (Triantafyllou 2018).

This review aimed to synthesise evidence from existing randomised clinical trials and to assess the beneficial and harmful effects of ePTFE‐covered stents compared with their bare equivalent. The review also aimed to identify further research gaps in comparing the effectiveness of ePTFE‐covered stents and their bare equivalent in the management of people with liver cirrhosis.

Objectives

To evaluate the benefits and harms associated with the use of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents in transjugular intrahepatic portosystemic shunts (TIPSs) for managing people with liver cirrhosis

Methods

Criteria for considering studies for this review

Types of studies

We included randomised clinical trials, with a parallel group design, comparing the use of ePTFE‐covered stents versus bare stents in TIPSs for managing people with liver cirrhosis. We included the trials irrespective of year and status of publication, language, blinding, or reported outcomes. We did not expect to find cross‐over or cluster randomised clinical trials.

We excluded trials with participants who were diagnosed with hepatocellular carcinoma or other malignant diseases before the TIPS procedure.

We excluded quasi‐randomised studies (i.e. pseudo‐randomised) as the allocation method is not truly random, and observational studies.

Types of participants

We included adults (i.e. at least 18 years old) diagnosed with liver cirrhosis either by liver biopsy or typical clinical signs regardless of the aetiology, manifestation, and severity of cirrhosis.

Types of interventions

We included trials comparing ePTFE‐covered stents, such as Viatorr stents or Fluency stents, a combination of both, or other types of ePTFE stents versus bare stents, in TIPS. We included trials irrespective of the diameter of the stent.

We allowed co‐interventions (e.g. endoscopic therapies, vasoactive drugs) if used equally in all trial groups.

Types of outcome measures

We analysed data on all outcomes at maximum follow‐up, except for the 'early thrombosis' outcome for which the primary time point for our main analysis was within 12 weeks of the TIPS procedure.

In addition to analysing outcomes with risk ratios (RR) (our main analysis approach), we also analysed outcomes with hazard ratios (HRs) if time‐to‐event data were available.

Primary outcomes

• All‐cause mortality.

• Procedure‐related complications, such as peritoneal bleeding, haemobilia, stent migration, sepsis, and others as defined by the trial authors.

• Health‐related quality of life, using any validated quality of life score. If individual trials used multiple scales, we planned to prioritise the 12‐item Short Form (SF‐12) (Ware 1996).

Secondary outcomes

• Upper gastrointestinal bleeding.

• Recurrence of ascites.

• Hepatic encephalopathy.

• Kidney failure.

• Thrombosis within 12 weeks after TIPS.

• Non‐serious adverse events, including segmental liver ischaemia.

• Shunt dysfunction (used as a surrogate outcome for stent patency, defined as a greater than 50% reduction of the lumen of the shunt at angiography or a portosystemic pressure gradient greater than 12 mmHg, or both).

Search methods for identification of studies

We followed the PRISMA 2020 guidance to plan and conduct the searches of databases and other resources (Page 2021a; Page 2021b).

Electronic searches

The Cochrane Hepato‐Biliary Group Information Specialist searched The Cochrane Hepato‐Biliary Group Controlled Trials Register via the Cochrane Register of Studies Web on 28 February 2023. In addition, we searched the Cochrane Central Register of Controlled Trials (2023, Issue 2) in the Cochrane Library; MEDLINE Ovid (1946 to 28 February 2023); Embase Ovid (Excerpta Medica Database; 1974 to 28 February 2023); LILACS (Latin American and Caribbean Health Science Information database, Bireme; 1982 to 28 February 2023); Science Citation Index Expanded (Web of Science; 1900 to 28 February 2023); and Conference Proceedings Citation Index – Science (Web of Science; 1990 to 28 February 2023). The latter two were searched simultaneously through the Web of Science. We also searched Wan Fang Data (1900 to 22 September 2022, www.wanfangdata.com.cn) and China National Knowledge Internet CNKI (1915 to 22 September 2022, www.cnki.net).

Appendix 1 provides the search strategies with the date range of the searches.

Searching other resources

We searched the online trial registries ClinicalTrial.gov (www.clinicaltrials.gov), the European Medicines Agency (EMA; www.ema.europa.eu/ema/), the World Health Organization International Clinical Trial Registry Platform (www.who.int/ictrp), the Food and Drug Administration (FDA; www.fda.gov/), and pharmaceutical company sources for planned and ongoing trials to 22 September 2022.

We searched the reference lists of included studies and any relevant systematic reviews to identify other published studies. We also reviewed any identified conference proceedings for preliminary data or interim‐analysis results.

We strived to obtain unpublished or published studies in the form of abstracts, letters, and any other format as well as missing data from published studies by contacting manufacturers of stents and the authors of relative studies. We applied no language limitations.

Data collection and analysis

Selection of studies

Two review authors (PZ and PS) examined titles and abstracts of the search results and removed obviously irrelevant reports. Then we retrieved the full text of all potentially relevant publications, linked multiple reports of the same trial, and identified the primary one. Two review authors (PZ and PS) who were not blinded to the trial authors and institutions of the full‐text publications assessed the articles for inclusion. The two review authors discussed any disagreements amongst themselves and if consensus was not reached, they requested a third review author (APB) to arbitrate. We included studies irrespective of whether measured outcome data were reported in a 'usable way'. We recorded and justified any reasons for exclusion of the ineligible studies in the Characteristics of excluded studies table.

We did not specifically search for adverse effects of the interventions in this review in observational studies, which is a known limitation of our systematic review. We are aware that by not looking for all observational studies on adverse effects, we allow the risks of putting more weight on potential benefits than on potential harms, and of overlooking uncommon and late adverse events (Storebø 2018). So, to minimise the risk of bias, we summarised the adverse events from the observational studies we found via our searches for randomised trials at the end of the Effects of interventions section.

Data extraction and management

Two review authors (PZ and PS) independently collected data using a data collection form piloted on four studies. We extracted data from trials with multiple reports of the same trial separately. Before we combined the data, we ensured that data were not duplicated or double counted by comparing the trial design, sample size, and year and place of trial conduct. The two review authors resolved disagreements by discussion and if required, a third review author (QZ) arbitrated. We sought any relevant missing data by contacting trial authors. We examined any retraction statements and errata for information on each trial. We extracted the following data.

• Trial information: trial ID; first author; country of author; location of the trial; year of publication; publication type; author contact details; funding sources; possible conflicts of interest; trial registration; ethics committee approval.

• Methods: trial design; total trial duration; sequence generation; allocation sequence concealment; blinding; duration of follow‐up; loss of follow‐up; intention‐to‐treat (ITT) analysis; confounding factors and methods to control confounding (for non‐randomised studies).

• Participants: total number; setting; age; sex; diagnostic criteria; inclusion criteria; exclusion criteria; baseline characteristics (Child‐Pugh class score, causes of cirrhosis, indications for TIPS, portocaval gradient before and after TIPS, etc.).

• Interventions: types of stents for TIPS.

• Outcomes: primary and secondary outcomes of the trial publication.

• Notes: correspondence required.

Assessment of risk of bias in included studies

Two review authors (XC and YS) independently assessed the risk of bias of each included trial according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a) and methodological studies (Kjaergard 2001; Moher 1998; Savović 2012a; Savović 2012b; Savović 2018; Schulz 1995; Wood 2008). We used the following definitions for the assessment of risk of bias.

Allocation sequence generation

• Low risk of bias: sequence generation was achieved using computer random number generation or a random number table. Drawing lots, tossing a coin, shuffling cards, and throwing dice were adequate if performed by an independent person not otherwise involved in the trial.

• Unclear risk of bias: the method of sequence generation was not specified.

• High risk of bias: the sequence generation method was not random. We only used these studies for the assessment of harm and not for determining benefit.

Allocation concealment

• Low risk of bias: the participant allocations could not have been foreseen in advance of or during enrolment. Allocation was controlled by a central and independent randomisation unit. The allocation sequence was unknown to the investigators (e.g. if the allocation sequence was hidden in sequentially numbered, opaque, and sealed envelopes).

• Unclear risk of bias: the method used to conceal the allocation was not described so that intervention allocations may not be foreseen in advance of or during enrolment.

• High risk of bias: the allocation sequence was likely to be known to the investigators who assigned the participants. We only used these studies for the assessment of harm and not for determining benefit.

Blinding of participants and personnel

• Low risk of bias: any of the following: no blinding or incomplete blinding, but the review authors judged that the outcome was not likely to be influenced by lack of blinding; or blinding of participants and key trial personnel ensured, and it was unlikely that the blinding could have been broken.

• Unclear risk of bias: any of the following: insufficient information to permit judgement of 'low risk' or 'high risk'; or the trial did not address this outcome.

• High risk of bias: any of the following: no blinding or incomplete blinding, and the outcome was likely to be influenced by lack of blinding; or blinding of key trial participants and personnel attempted, but likely that the blinding could have been broken, and the outcome was likely to be influenced by lack of blinding.

Blinded outcome assessment

• Low risk of bias: any of the following: no blinding of outcome assessment, but the review authors judged that the outcome measurement was not likely to be influenced by lack of blinding; or blinding of outcome assessment ensured, and it was unlikely that the blinding could have been broken.

• Unclear risk of bias: any of the following: insufficient information to permit judgement of 'low risk' or 'high risk'; or the trial did not address this outcome.

• High risk of bias: any of the following: no blinding of outcome assessment, and the outcome measurement was likely to be influenced by lack of blinding; or blinding of outcome assessment, but likely that the blinding could have been broken, and the outcome measurement was likely to be influenced by lack of blinding.

Incomplete outcome data

• Low risk of bias: missing data were unlikely to make treatment effects depart from plausible values. Sufficient methods, such as multiple imputations, were employed to handle missing data.

• Unclear risk of bias: there was insufficient information to assess whether missing data in combination with the method used to handle missing data was likely to bias the results.

• High risk of bias: the results were likely to be biased due to missing data.

Selective outcome reporting

• Low risk of bias: the trial reported the following predefined outcomes: all‐cause mortality, procedure‐related complications, hepatic encephalopathy, and shunt dysfunction. If the original trial protocol was available, the outcomes should have been those called for in that protocol. If the trial protocol was obtained from a trial registry (e.g. www.ClinicalTrials.gov), the outcomes sought should have been those enumerated in the original protocol if the trial protocol was registered before or at the time that the trial began. If the trial protocol was registered after the trial was begun, those outcomes were not considered to be reliable.

• Unclear risk of bias: not all predefined outcomes were reported fully, or it was unclear whether data on these outcomes was recorded or not.

• High risk of bias: one or more predefined outcomes were not reported.

Other bias

• Low risk of bias: the trial appeared free from other factors that could have put it at risk of bias.

• Unclear risk of bias: the trial may or may not have been free from other factors that could have put it at risk of bias.

• High risk of bias: there were other factors in the trial that could have put it at risk of bias.

Overall risk of bias

We judged a trial at an overall low risk of bias if it was assessed at low risk of bias in all the above domains. We judged a trial at an overall high risk of bias if it was assessed at unclear risk of bias or high risk of bias in one or more of the above domains.

We planned to assess risk of bias due to incomplete outcome data and lack of blinding separately for different key outcomes. We resolved differences in opinion by discussion. When unsettled disagreements arose, a third review author adjudicated (LT).

Measures of treatment effect

For dichotomous variables, we calculated RRs with 95% confidence intervals (CIs). For continuous data, we calculated mean differences (MDs) when studies used the same scale or standardised mean differences (SMDs) when studies used different scales with 95% CIs. We also used HRs with 95% CIs as relevant effect measures for the time‐to‐event outcome (mortality, etc.) when data were available.

Unit of analysis issues

We used the number of participants as randomised within the trial groups. For trials with more than two intervention groups, we planned to collect data for all intervention groups meeting the inclusion criteria of our review protocol. If the control group was a common comparator in a comparison, we planned to divide the number of participants by the number of relevant intervention groups in order to avoid counting the same data multiple times. If we identified cross‐over trials, though we considered it unlikely, we planned to use data from the first trial period only (Higgins 2011b). If we identified cluster‐randomised trials, though we considered it unlikely, we planned to use the intraclass correlation coefficient (ICC) to convert trials to their effective sample size before data synthesis (Higgins 2011b). Where the ICC was not available, we planned to use values for ICCs in other published studies.

Dealing with missing data

For data that were not missing at random, we contacted the trial authors to obtain the missing data. We performed ITT analyses according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). When analysis with ITT was not possible, we conducted available case analysis or per‐protocol analysis. We state this clearly in the Results section, and we discuss this further in the Potential biases in the review process section.

Assessment of heterogeneity

We assessed the clinical diversity across trials (liver function of trial participants, types of stents used for participants, etc.). We judged variability in trial design and risk of bias based on the domains listed previously. We assessed statistical heterogeneity by examining the I² statistic, which can be interpreted as the percentage of variation observed between the trials attributable to between‐trial difference rather than sampling error. We used the following thresholds to express heterogeneity: 0% to 30% (unimportant), 30% to 60% (moderate), 50% to 90% (substantial), and more than 75% (considerable). In addition, we used Chi² test to provide an indication of between‐trial heterogeneity (Deeks 2019).

Assessment of reporting biases

We conducted a comprehensive search for eligible studies, including grey literature and unpublished studies. We contacted trial authors to request missing information. We did not use funnel plots to assess the reporting biases because there were fewer than 10 trials for each outcome.

Data synthesis

We used Review Manager 5 to analyse data (Review Manager 2020).

Meta‐analysis

We presented our meta‐analyses with the random‐effects model which "involves an assumption that the effects being estimated in the different trials are not identical but follow some distribution" (DeMets 1987; DerSimonian 1986; Higgins 2011b). If there was a considerable variation in the results and considerable heterogeneity (more than 75%), and particularly if there was inconsistency in the direction of the effect, we planned to present the evidence in a narrative format instead of performing meta‐analyses (Deeks 2019).

Subgroup analysis and investigation of heterogeneity

We planned to perform the following subgroup analyses on all outcomes if data were available to explore important clinical differences amongst trials that could have affected the treatment effect.

• Brands of covered stents: Viatorr stents only and Viatorr plus Fluency stents compared to Fluency stents only because the scaffold structures are not identical, which may affect the outcome. Subgroup analysis was not performed for comparison of covered and bare stents of the same diameter because there were insufficient data. We did perform subgroup analysis for comparison of stents of different diameters.

• Trials including participants with acute upper gastrointestinal haemorrhage compared to trials excluding people with acute gastrointestinal haemorrhage. However, analysis for this subgroup was not performed because data were not available.

• Proportion of people with recurrence of ascites. However, analysis for this subgroup was not performed because data were not available.

Sensitivity analysis

We planned to examine the robustness of meta‐analyses by conducting the following sensitivity analysis on all‐cause mortality had there been sufficient data available:

Risk of bias

For the meta‐analysis of the primary outcome, we planned to analyse the effects of excluding trials that were judged to be at high risk of bias across one or more of the domains of randomisation (implied as randomised with no further details available) including allocation concealment, blinding, and outcome reporting. If the exclusion of trials at high risk of bias did not substantially alter the direction of the effect or the precision of the effect estimates, then we would have included data from these trials in the analysis. This sensitivity analysis was not performed because all the included trials were at high risk of bias in at least one domain (Kjaergard 2001; Moher 1998; Savović 2012a; Savović 2012b; Savović 2018; Schulz 1995; Wood 2008).

Fixed‐effect model

We repeated the analysis of primary outcomes with the fixed‐effect model to evaluate whether this altered the significance of the result.

Assumptions for lost binary data

For data missing at random, we performed sensitivity analyses to evaluate the influence of missing outcome data. We performed the following two ITT scenario analyses as sensitivity analyses for the outcome all‐cause mortality. We reported and discussed the outcome of sensitivity analyses when there was a substantial difference.

• Extreme case analysis favouring the experimental intervention ('best‐worst' case scenario): none of the dropouts/participants lost from the experimental group, but all the dropouts/participants lost from the control group experienced the outcome, including all randomised participants in the denominator.

• Extreme case analysis favouring the control ('worst‐best' case scenario): all the dropouts/participants lost from the experimental group, but none of the dropouts/participants lost from the control group experienced the outcome, including all randomised participants in the denominator.

Assessment of imprecision with Trial Sequential Analysis

Trial Sequential Analysis was performed using Trial Sequential Analysis software (TSA 2017). We applied Trial Sequential Analysis (Thorlund 2017; TSA 2017) as cumulative meta‐analyses are at risk of producing random errors due to sparse data and repetitive testing of the accumulating data (Wetterslev 2008). A required information size to detect or reject a certain intervention effect accounts for the diversity, present in a meta‐analysis (Wetterslev 2008; Wetterslev 2009; Wetterslev 2017). In our meta‐analysis, we calculated the diversity‐adjusted required information size (DARIS) based on the event proportion in the control group; the assumption of a plausible relative risk reduction (RRR) of 20% (Wetterslev 2009); type I error (α) of 2.5% because of our three primary outcomes and type I error (α) of 1.25% because of our seven secondary outcomes; a risk of 20% for type II error (β), and a diversity of 10%.

The underlying assumption of Trial Sequential Analysis is that testing for significance is performed each time a new trial is added to the meta‐analysis. We added the trials, irrespective of their risk of bias, according to the year of publication, and, if more than one trial was published in a year, we added trials alphabetically according to the last name of the first author. On the basis of the required information size, we constructed trial sequential monitoring boundaries (Thorlund 2017; Wetterslev 2008; Wetterslev 2017). These boundaries determine the statistical inference one may draw regarding the cumulative meta‐analysis that does not reach the required information size; if the trial sequential monitoring boundary for benefit or harm is crossed before the required information size is reached, firm evidence might be established, and further trials might be superfluous. In contrast, if the boundaries are not surpassed, it is most probably necessary to continue doing trials in order to detect or reject a certain intervention effect. That is determined by assessing if the cumulative Z‐curve crosses the trial sequential monitoring boundary for futility.

In Trial Sequential Analysis, we downgraded our assessment of imprecision by two levels when the accrued number of participants was below 50% of the DARIS, and one level when between 50% and 100% of DARIS. We did not downgrade when futility or DARIS was reached (Jakobsen 2014).

Summary of findings and assessment of the certainty of the evidence

Two review authors (ST and ZP) independently assessed the overall certainty of the evidence for each outcome using GRADEpro GDT (GRADEpro GDT). We solved differences in opinions by discussion. When unsettled disagreements arose, a third review author adjudicated (LT).

We created two summary of findings tables using GRADEpro GDT (GRADEpro GDT). We presented the comparisons 'ePTFE‐covered stents versus bare stents for TIPS in people with liver cirrhosis – same stent diameter' and 'ePTFE‐covered stents versus bare stents for TIPS in people with liver cirrhosis – different stent diameters'. In both tables, we presented the following outcome results for stents of the same diameter and stents of different diameters.

• All‐cause mortality

• Procedure‐related complications

• Upper gastrointestinal bleeding

• Recurrence of ascites

• Hepatic encephalopathy

• Kidney failure

• Shunt dysfunction

The GRADE approach appraises the certainty of a body of evidence based on the extent to which one was confident that an estimate of effect or association reflects the item being assessed (GRADEpro GDT). The certainty of a body of evidence considers within‐trial risk of bias, indirectness of the evidence, heterogeneity of the data, imprecision of results (wide CIs) (Jakobsen 2014), and risk of publication bias (Balshem 2011; Guyatt 2011a; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2011g; Guyatt 2011h; Guyatt 2013a; Guyatt 2013b; Guyatt 2013c; Guyatt 2013d; Guyatt 2017; Mustafa 2013).

The certainty of evidence is defined as high, moderate, low, or very low, and the grades are defined as follows.

• High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

• Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.

• Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

We followed our protocol during the preparation of this review, and we reported the differences between the protocol and the review in the Differences between protocol and review section.

Results

Description of studies

See: Characteristics of included studies, Characteristics of excluded studies, and Characteristics of studies awaiting classification tables.

Results of the search

We identified 676 records, 631 from electronic databases and 45 from other resources through the searches up to 28 February 2023. In the compiling of the search results from the individual databases, we identified and removed 334 duplicates. From the remaining 342 records, we excluded 322 records based on titles and abstracts. We assessed the full‐text of the remaining 20 records and subsequently excluded eight studies (11 records). The excluded studies with reasons are listed in the Characteristics of excluded studies table. One study (with two records) is awaiting classification (Characteristics of studies awaiting classification table). We included four trials in the review (one trial had four publications) (Characteristics of included studies table). The study selection process is shown in Figure 1.

1.

Study flow diagram (Page 2021a; Page 2021b). Date of search 22 September 2022 and 28 February 2023

Included studies

We included four trials (Bureau 2007; Huang 2010; Perarnau 2014; Wang 2016a; see Characteristics of included studies table). Two trials were single‐centre conducted in China (Huang 2010; Wang 2016a). The remaining trials were multicentre: one conducted in France (Perarnau 2014), and one conducted in France, Spain, and Canada (Bureau 2007). Each trial had a full‐text publication. Three trials provided data with ITT analysis (Bureau 2007; Huang 2010; Perarnau 2014), and one trial provided per‐protocol data (Wang 2016a).

Participants

The four trials included 565 randomised participants; 368 men and 166 women; the sex of the remaining 31 participants was not reported. Not all included trials applied the ITT principle to meta‐analyse outcome data; therefore, our analyses included data from 527 participants (about 9% less than those randomised). The proportion of men ranged from 63.6% to 75.0%. The ages of all included participants ranged from 18 to 75 years, and the mean ages ranged from 45 to 55 years. Two trials with 140 participants made the diagnosis of cirrhosis using liver biopsy or typical clinical signs (Bureau 2007; Huang 2010). One trial with 137 participants made the diagnosis on histological findings or typical clinical signs (Perarnau 2014). The diagnostic criteria of the remaining 288 participants in the fourth trial remained unknown due to lack of reporting (Wang 2016a). One trial did not report the Child‐Pugh classification (Bureau 2007). Of the other three trials (505 participants), 100 participants were classified as Child‐Pugh class A, 241 participants as Child‐Pugh class B, 112 participants as Child‐Pugh class C, and there was no information for the remaining 52 participants (Huang 2010; Perarnau 2014; Wang 2016a). The causes of cirrhosis included posthepatitis (206 participants), alcohol (153 participants), viral (75 participants), and non‐alcoholic steatohepatitis (seven participants). The indications for TIPS were refractory ascites or refractory hydrothorax (192 participants), upper gastrointestinal bleeding (acute uncontrolled bleeding, 23 participants), and rebleeding (143 participants). All trials excluded participants with portal vein thrombosis, a history of malignant diseases, and chronic hepatic encephalopathy.

Interventions

All were two‐group trials comparing ePTFE‐covered stents versus bare stents used in TIPS procedures. For the covered stents group, two trials used Fluency stents (Bard, USA) with a diameter of 8 mm (Huang 2010; Wang 2016a), one trial used Viatorr stents (Arizona, USA) alone with a diameter of 10.5 (standard deviation (SD) 0.9) mm (Bureau 2007), and one trial used stents with a diameter of 10 mm (Perarnau 2014).

In the bare stents group, the doctors chose the stents, and the stent diameter varied from 8 mm to 11.7 (SD 0.8) mm (Bureau 2007; Huang 2010; Perarnau 2014; Wang 2016a).

Follow‐up

Two trials reported a total follow‐up of two years (Bureau 2007; Perarnau 2014), one trial reported five years (Wang 2016a). The follow‐up period was 8.34 (SD 4.42) months in the bare‐stent group and 6.16 (SD 3.89) months in the covered‐stent group in one trial (Huang 2010).

Funding of trials

One trial did not clearly report funding sources (Huang 2010). The authors of the remaining three trials declared that they had no funding with vested interests (Bureau 2007; Perarnau 2014; Wang 2016a).

Excluded studies

Based on the full‐text publications, we excluded eight studies with 11 records (Characteristics of excluded studies table). Amongst these eight studies, three were retrospective studies (Jung 2009; Mittal 2010; Wu 2010). Interventions in one study were not relevant to our review question (Bureau 2015). Two studies only compared covered stents with different diameters (Boatta 2009; Riggio 2010), and the remaining study compared the technique of using a combination of stents and stent‐grafts with using a single stent‐graft to construct a TIPS (Wang 2016b).

Studies awaiting classification

One study was published twice as conference abstracts. However, the randomisation method was not explained and there was insufficient information for data analysis. We were unsuccessful in our attempts to contact the corresponding authors (Bandi 2010).

Ongoing trials

We identified no ongoing studies.

Risk of bias in included studies

Figure 2 shows the risk of bias graph and Figure 3 shows the risk of bias summary.

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies

3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Allocation

Random sequence generation

Three trials reported adequate allocation sequence generation (low risk of bias; Bureau 2007; Perarnau 2014; Wang 2016a), whereas in one trial generation of the allocation sequence was unclear (Huang 2010).

Allocation concealment

Allocation concealment was at low risk of bias in all four trials.

Blinding

Performance bias

In one trial, only the participants were blinded (Perarnau 2014). Two trials lacked blinding (Bureau 2007; Huang 2010). Therefore, we assessed all three trials at high risk of performance bias. The remaining trial ensured double blinding and was at low risk of performance bias (Wang 2016a).

Detection bias

Two trials reported that the outcomes assessors were blinded to the allocation, and were at low risk of detection bias (Perarnau 2014; Wang 2016a). The other two trials did not blind the outcome assessors to the allocation, and were at high risk of detection bias (Bureau 2007; Huang 2010).

Incomplete outcome data

No participants were lost to follow‐up in two trials (Bureau 2007; Huang 2010), and the ITT analysis was explicitly stated in one trial (Perarnau 2014). Therefore, we rated these three trials at low risk of attrition bias. Though the remaining trial reported reasons for the loss to follow‐up, it did not perform an ITT analysis and was at high risk of bias (Wang 2016a).

Selective reporting

Two trials were free of reporting bias (low risk of bias; Bureau 2007; Perarnau 2014). The other two trials did not report adverse events clearly and were subsequently rated at unclear risk of reporting bias (Huang 2010; Wang 2016a).

Other potential sources of bias

We did not identify any other obvious biases.

Effects of interventions

See: Table 1; Table 2

We included four trials with 565 randomised participants. The trials compared ePTFE‐covered stents versus bare stents (Bureau 2007; Huang 2010; Perarnau 2014; Wang 2016a). A total of 527 participants provided data for analyses because of losses to follow‐up.

Expanded polytetrafluoroethylene‐covered stents versus bare stents of the same stent diameter

One trial, with 258 participants, compared stents of the same diameter (8 mm) (Wang 2016a).

Primary outcomes

All‐cause mortality

Mortality in the ePTFE‐covered stent group was considerably lower than in the bare stent group (RR 0.63, 95% CI 0.43 to 0.92; P = 0.02; 258 participants; Analysis 1.1; low‐certainty evidence).

1.1. Analysis.

Comparison 1: ePTFE‐covered stents versus bare stents – same stent diameter, Outcome 1: All‐cause mortality – RR

Procedure‐related complications

The trial reported no data on procedure‐related complications.

Health‐related quality of life

The trial reported no data on health‐related quality of life.

Secondary outcomes

Upper gastrointestinal bleeding

Upper gastrointestinal bleeding occurred more often in the bare stent group than in the ePTFE‐covered stent group (RR 0.54, 95% CI 0.35 to 0.84; P = 0.006; 258 participants; Analysis 1.2; low‐certainty evidence).

1.2. Analysis.

Comparison 1: ePTFE‐covered stents versus bare stents – same stent diameter, Outcome 2: Upper gastrointestinal bleeding

Recurrence of ascites

Recurrence of ascites occurred more often in the bare stent group than in the ePTFE‐covered stent group (RR 0.42, 95% CI 0.20 to 0.87; P = 0.02; 258 participants; Analysis 1.3; low‐certainty evidence).

1.3. Analysis.

Comparison 1: ePTFE‐covered stents versus bare stents – same stent diameter, Outcome 3: Recurrence of ascites

Hepatic encephalopathy

There was no evidence of a difference in the incidence of hepatic encephalopathy between the ePTFE‐covered stent group and the bare stent group (RR 1.10, 95% CI 0.76 to 1.61; P = 0.60; 258 participants; Analysis 1.4; very low‐certainty evidence).

1.4. Analysis.

Comparison 1: ePTFE‐covered stents versus bare stents – same stent diameter, Outcome 4: Hepatic encephalopathy – RR

Kidney failure

The trial reported no data on kidney failure.

Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt

The trial reported no data on thrombosis within 12 weeks after TIPS.

Segmental liver ischaemia (non‐serious adverse event)

The trial reported no data on segmental liver ischaemia.

Shunt dysfunction

Stunt dysfunction occurred more often in the bare stent group than in the covered stent group (RR 0.42, 95% CI 0.28 to 0.61; P < 0.00001; 258 participants; Analysis 1.5; low‐certainty evidence).

1.5. Analysis.

Comparison 1: ePTFE‐covered stents versus bare stents – same stent diameter, Outcome 5: Shunt dysfunction – RR

Expanded polytetrafluoroethylene‐covered stents versus bare stents of different stent diameters

Three trials with 267 participants used different stent diameters between groups (Bureau 2007; Huang 2010; Perarnau 2014). The stents had the following diameters: 10.5 (SD 0.9) mm ePTFE‐covered stent versus 11.7 (SD 0.8) mm bare stent in Bureau 2007; 8 mm ePTFE‐covered stent versus 10 mm bare stent in Huang 2010; 8 mm to 10 mm ePTFE‐covered stent versus 8 mm to 10 mm bare stent in Perarnau 2014.

Primary outcomes

All‐cause mortality

Three trials provided data on mortality (Bureau 2007; Huang 2010; Perarnau 2014). There was no evidence of a difference in mortality between ePTFE‐covered stents and bare stents of different diameters (RR 0.75, 95% CI 0.48 to 1.16; P = 0.19; I² = 17%; 269 participants; Analysis 2.1; very low‐certainty evidence).

2.1. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 1: All‐cause mortality – RR

Two trials providing data with HRs indicated no clear differences between ePTFE‐covered stents and bare stents of different diameters (HR 0.82, 95% CI 0.50 to 1.36; P = 0.45; I² = 0%; 209 participants; Analysis 2.2; very low‐certainty evidence) (Bureau 2007; Perarnau 2014).

2.2. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 2: All‐cause mortality – HR (logrank)

Procedure‐related complications

One trial reported procedure‐related complications. There was no evidence of a difference in the incidence of procedure‐related complications between ePTFE‐covered stents and bare stents of different diameters (RR 0.53, 95% CI 0.05 to 5.57; P = 0.59; 80 participants; Analysis 2.3; very low‐certainty evidence) (Bureau 2007).

2.3. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 3: Procedure‐related complications

Health‐related quality of life

None of the trials reported data on health‐related quality of life.

Secondary outcomes

Upper gastrointestinal bleeding

There was no evidence of a difference in the incidence of upper gastrointestinal bleeding between ePTFE‐covered stents and bare stents of different diameters (RR 0.46, 95% CI 0.15 to 1.38; P = 0.17; I² = 0%; 3 trials, 269 participants; Analysis 2.4; very low‐certainty evidence) (Bureau 2007; Huang 2010; Perarnau 2014).

2.4. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 4: Upper gastrointestinal bleeding

Recurrence of ascites

Three trials provided data on recurrence of ascites (Bureau 2007; Huang 2010; Perarnau 2014). Recurrence of ascites occurred more often with bare stents than with ePTFE‐covered stents (RR 0.30, 95% CI 0.11 to 0.85; P = 0.02; I² = 0%; 269 participants; Analysis 2.5; low‐certainty evidence).

2.5. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 5: Recurrence of ascites

Hepatic encephalopathy

Three trials provided data on hepatic encephalopathy (Bureau 2007; Huang 2010; Perarnau 2014). There was no evidence of a difference between ePTFE‐covered stents and bare stents of different diameters (RR 0.93, 95% CI 0.66 to 1.30; P = 0.66; I² = 0%; 269 participants; Analysis 2.6; very‐low certainty evidence).

2.6. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 6: Hepatic encephalopathy – RR

Two trials also used HRs to illustrate hepatic encephalopathy (Bureau 2007; Perarnau 2014). There was no evidence of a difference between ePTFE‐covered stents and bare stents of different diameters (HR 0.63, 95% CI 0.29 to 1.37; P = 0.25; I² = 64%; 2 trials; Analysis 2.7; very low‐certainty evidence).

2.7. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 7: Hepatic encephalopathy – HR (logrank)

Kidney failure

One trial reported data on kidney failure (Perarnau 2014). There was no evidence of a difference in the occurrence of kidney failure between ePTFE‐covered stents and bare stents of different diameters (RR 7.59, 95% CI 0.40 to 143.92; P = 0.18; 121 participants; Analysis 2.8; very low‐certainty evidence).

2.8. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 8: Kidney failure

Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt

Three trials reported early thrombosis (Bureau 2007; Huang 2010; Perarnau 2014). The meta‐analysis demonstrated that there was a lower probability of early thrombosis with the use of ePTFE‐covered stents than with the use of bare stents (RR 0.28, 95% CI 0.09 to 0.82; P = 0.02; I² = 0%; 261 participants; Analysis 2.9; very low‐certainty evidence).

2.9. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 9: Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt

Segmental liver ischaemia (non‐serious adverse event)

One small trial reported segmental liver ischaemia (Bureau 2007). There was no evidence of a difference between ePTFE‐covered stents and bare stents of different diameters (RR 5.25, 95% CI 0.26 to 106.01; P = 0.28; 80 participants; Analysis 2.10; very low‐certainty evidence).

2.10. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 10: Segmental liver ischaemia

Shunt dysfunction

Three trials reported shunt dysfunction (Bureau 2007; Huang 2010; Perarnau 2014). Results showed that shunt dysfunction occurred more often in the bare stents group than in the ePTFE‐covered stent group (RR 0.50, 95% CI 0.28 to 0.92; P = 0.03; I² = 60%; 269 participants; Analysis 2.11; low‐certainty evidence).

2.11. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 11: Shunt dysfunction – RR

Two trials also used HRs to illustrate shunt dysfunction (Bureau 2007; Perarnau 2014). The result indicated a clear difference between the ePTFE‐covered stents versus bare stents favouring the ePTFE‐covered stents (HR 0.41, 95% CI 0.17 to 0.99; P = 0.05; I² = 69%; 209 participants; Analysis 2.12; low‐certainty evidence).

2.12. Analysis.

Comparison 2: ePTFE‐covered stents versus bare stents – different stent diameter, Outcome 12: Shunt dysfunction – HR (logrank)

Subgroup analysis: Viatorr stents versus Fluency stents (test for differences between brands of stents)

We tested for subgroup differences between trials using Viatorr stents and Fluency stents on all outcomes with sufficient data. One trial used Viatorr stents (Bureau 2007) and one trial used Fluency stents (Huang 2010).

Primary outcomes

All‐cause mortality

Results indicated there were no clear reductions of mortality using Viatorr stents (RR 0.66, 95% CI 0.37 to 1.18; P = 0.16; 80 participants) or Fluency stents (RR 0.11, 95% CI 0.01 to 1.98; P = 0.13; 60 participants) compared with its bare equivalent. Test of subgroup differences indicated that there were no clear differences in all‐cause mortality between Viatorr stents groups and Fluency stents groups (Chi² = 1.42, df = 1 (P = 0.23); I² = 29.8%; Analysis 3.1).

3.1. Analysis.

Comparison 3: Subgroup analysis: types of covered stents, Outcome 1: All‐cause mortality

Secondary outcomes

Upper gastrointestinal bleeding

When compared with bare stents, there was no evidence of a difference in the incidence of upper gastrointestinal bleeding in the Viatorr stents group (RR 0.79, 95% CI 0.19 to 3.30; P = 0.74; 1 trial, 80 participants) and Fluency stents group (RR 0.17, 95% CI 0.02 to 1.30; P = 0.09; 1 trial, 60 participants). Test of subgroup differences indicated that there were no clear differences in the incidence of upper gastrointestinal bleeding between Viatorr stents groups and Fluency stents groups (Chi² = 1.48, df = 1 (P = 0.22); I² = 32.4%; Analysis 3.2).

3.2. Analysis.

Comparison 3: Subgroup analysis: types of covered stents, Outcome 2: Upper gastrointestinal bleeding

Recurrence of ascites

There were no clear differences in the incidence of recurrence of ascites between bare stents and Viatorr stents (RR 0.13, 95% CI 0.02 to 1.00; P = 0.05; 1 trial, 80 participants) and Fluency stents (RR 0.20, 95% CI 0.01 to 4.00; P = 0.29; 1 trial, 60 participants). Test of subgroup differences indicated that there were no clear differences in the incidence of recurrence of ascites between Viatorr stents groups and Fluency stents groups (Chi² = 0.05, df = 1 (P = 0.82); I² = 0%; Analysis 3.3).

3.3. Analysis.

Comparison 3: Subgroup analysis: types of covered stents, Outcome 3: Recurrence of ascites

Hepatic encephalopathy

There were no clear differences in the incidence of hepatic encephalopathy between bare stents and Viatorr stents (RR 0.95, 95% CI 0.43 to 2.08; P = 0.89; 1 trial, 80 participants) and Fluency stents (RR 0.83, 95% CI 0.28 to 2.44; P = 0.74; 1 trial, 60 participants). Test of subgroup differences indicated that there were no clear differences in the incidence of hepatic encephalopathy between Viatorr stents groups and Fluency stents groups (Chi² = 0.03, df = 1 (P = 0.85); I² = 0%; Analysis 3.4).

3.4. Analysis.

Comparison 3: Subgroup analysis: types of covered stents, Outcome 4: Hepatic encephalopathy

Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt

There were no clear differences in the incidence of early thrombosis between bare stents and Viatorr stents (RR 0.15, 95% CI 0.01 to 2.81; P = 0.20; 1 trial, 80 participants) and Fluency stents (RR 0.08, 95% CI 0.00 to 1.31; P = 0.08; 1 trial, 60 participants). Test of subgroup differences indicated that there were no clear differences between Viatorr stents groups and Fluency stents groups (Chi² = 0.10, df = 1 (P = 0.75); I² = 0%; Analysis 3.5).

3.5. Analysis.

Comparison 3: Subgroup analysis: types of covered stents, Outcome 5: Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt

Shunt dysfunction

Compared with bare stents, results showed that shunt dysfunction occurred less frequently in Viatorr stents groups (RR 0.35, 95% CI 0.16 to 0.79; P = 0.01; 1 trial, 80 participants) and Fluency stents groups (RR 0.36, 95% CI 0.15 to 0.87; P = 0.02; 1 trial, 60 participants). Test of subgroup differences indicated that there were no clear differences in the incidence of shunt dysfunction between Viatorr stents groups and Fluency stents groups (Chi² = 0.00, df = 1 (P = 0.98); I² = 0%; Analysis 3.6).

3.6. Analysis.

Comparison 3: Subgroup analysis: types of covered stents, Outcome 6: Shunt dysfunction

The subgroup analyses on other predefined factors, comparisons, and outcomes were not applicable due to insufficient data.

Sensitivity analysis: all‐cause mortality at maximum follow‐up (five years)

Risk of bias

The sensitivity analysis was not applicable because all the included trials were at high risk of bias in at least one domain.

Fixed‐effect model

The sensitivity analysis on the comparison of the same diameter was not applicable because this comparison included only one trial.

The sensitivity analysis was only applicable for the all‐cause mortality outcome under the comparison between covered stents and bare stents of different diameters. The results were consistent between fixed‐effect and random‐effects models (Analysis 4.1).

4.1. Analysis.

Comparison 4: Sensitivity analysis (fixed‐effect model) – ePTFE‐covered stents versus bare stents (different stent diameters), Outcome 1: All‐cause mortality – RR

Assumptions for lost binary data

Best‐worst case scenario (same stent diameter)

The sensitivity analysis indicated that results were consistent between the incorporation of lost data with or without best‐worst case scenario when comparing covered and bare stents of the same diameter. The ePTFE‐covered stents group was associated with lower mortality than the bare stents group of the same diameter (Analysis 5.1).

5.1. Analysis.

Comparison 5: Sensitivity analysis (missing data – best‐worst) – ePTFE‐covered stents versus bare stents (same stent diameter), Outcome 1: All‐cause mortality – RR

Best‐worst case scenario (different stent diameters)

Results showed no clear differences in mortality between covered stents and bare stents of different diameters in both assumption groups. This indicated that results were consistent between the incorporation of lost data with or without best‐worst case scenario (Analysis 6.1).

6.1. Analysis.

Comparison 6: Sensitivity analysis (missing data – best‐worst) – ePTFE‐covered stents versus bare stents (different stent diameter), Outcome 1: All‐cause mortality – RR

Worst‐best case scenario (same stent diameter)

The sensitivity analysis indicated results were inconsistent between the incorporation of lost data with or without worst‐best cases scenario (Analysis 7.1). Under the worst‐best case scenario, there were no clear differences in mortality between the covered and bare stents groups of the same diameter (RR 0.92, 95% CI 0.65 to 1.28; P = 0.61; 1 trial, 288 participants). However, without the worst‐best case scenario, results showed that ePTFE‐covered stents group had lower mortality than the bare stents group of the same diameter (RR 0.63, 95% CI 0.43 to 0.92; P = 0.02; 1 trial, 258 participants).

7.1. Analysis.

Comparison 7: Sensitivity analysis (missing data – worst‐best) – ePTFE‐covered stents versus bare stents (same stent diameter), Outcome 1: All‐cause mortality – RR

Worst‐best case scenario (different stent diameters)

The sensitivity analysis indicated that results were consistent between the incorporation of lost data with or without worst‐best case scenario when comparing covered and bare stents of different diameters. There was no clear difference in mortality between the ePTFE‐covered stents group and the bare stents group of different diameters (Analysis 8.1).

8.1. Analysis.

Comparison 8: Sensitivity analysis (missing data ‐ worst‐best) – ePTFE‐covered stents versus bare stents (different stent diameter), Outcome 1: All‐cause mortality – RR

Trial Sequential Analysis (same stent diameter)

All‐cause mortality at maximum follow‐up

Trial Sequential Analysis showed that the DARIS was 1646 participants, calculated based on the proportion of death of 38% amongst participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 2.5%; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve crossed the conventional boundary based on the one included trial (258 participants). However, it did not cross the trial sequential monitoring boundary, suggesting there was no conclusive evidence that ePTFE‐covered stent could reduce mortality (Figure 4). Compared with the GRADE assessment of imprecision, which was downgraded one level because the sample size did not meet the optimal information size (Table 1), the Trial Sequential Analysis assessment on all‐cause mortality downgraded the evidence two levels for imprecision.

4.

All‐cause mortality (same stent diameter)

Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (same stent diameter) for TIPS in people with liver cirrhosis on the outcome of all‐cause mortality.

The diversity‐adjusted required information size (DARIS) of 1646 was calculated based upon the proportion of death in the controlled group (38%); a relative risk reduction (RRR) of 20%; an alpha of 2.5% (type I error); a beta (type II error) of 20%; and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted‐CI was 0.63 (0.18 to 2.20). The blue curve presents the cumulative meta‐analysis Z‐score and the inward‐sloping red curves present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The inner triangle represents the area of futility.

Upper gastrointestinal bleeding

Trial Sequential Analysis showed a DARIS of 2338 participants, calculated based on the proportion of upper gastrointestinal bleeding of 34% amongst participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 1.25%; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve crossed the conventional boundary based on the one included trial (258 participants). However, it did not cross the trial sequential monitoring boundary, suggesting there was no firm evidence of ePTFE‐covered stent in reducing upper gastrointestinal bleeding (Figure 5). Compared with the GRADE assessment of imprecision which was downgraded one level since the sample size did not meet the optimal information size (Table 1), the Trial Sequential Analysis assessment on upper gastrointestinal bleeding downgraded the evidence two levels for imprecision.

5.

Upper gastrointestinal bleeding (same stent diameter)

Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (same stent diameter) for TIPS in people with liver cirrhosis on the outcome of upper gastrointestinal bleeding.

The diversity‐adjusted required information size (DARIS) 2276 was calculated based upon a proportion of upper gastrointestinal bleeding of 34% in the controlled group; a relative risk reduction (RRR) of 20%; an alpha of 1.25% (type I error); a beta of 20% (type II error); and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted CI was 0.54 (0.10 to 3.00). The blue curve presents the cumulative meta‐analysis Z‐score and the inward‐sloping red curves present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The triangular area represents the area of futility.

Recurrence of ascites

Because of insufficient information, Trial Sequential Analysis could not be constructed. Compared with the GRADE assessment of imprecision which was downgraded one level since the sample size did not meet the optimal information size (Table 1), the Trial Sequential Analysis assessment on recurrence of ascites downgraded the evidence two levels for imprecision.

Hepatic encephalopathy

Trial Sequential Analysis showed a DARIS of 1897 participants. DARIS was calculated based on a proportion of hepatic encephalopathy of 28.3% of participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 1.25; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve did not cross either the conventional or the trial sequential monitoring boundaries based on the one included trial (258 participants), demonstrating none of the interventions reached superiority and that the trial sequential monitoring boundaries of futility were not reached (Figure 6). Therefore, we downgraded the Trial Sequential Analysis assessment of hepatic encephalopathy for imprecision two levels, which was consistent with the GRADE assessment (Table 1).

6.

Hepatic encephalopathy (same stent diameter)

Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (same stent diameter) for TIPS in people with liver cirrhosis on the outcome of hepatic encephalopathy.

The diversity‐adjusted required information size (DARIS) of 2937 was calculated based upon a proportion of hepatic encephalopathy of 28.3% in the controlled group; a relative risk reduction (RRR) of 20%; an alpha of 1.25% (type I error); a beta of 20% (type II error); and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted CI was 1.1 (0.17 to 7.17). The blue curve presents the cumulative meta‐analysis Z‐score and the red lines at the left bottom corners present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The inner triangle represents the area of futility.

Shunt dysfunction

Trial Sequential Analysis showed a DARIS of 1260 participants, calculated based on a proportion of shunt dysfunction of 50% amongst participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 1.25%; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve crossed the conventional boundary based on the one included trial (258 participants). However, it did not cross the trial sequential monitoring boundaries, which suggested that no firm evidence of ePTFE‐covered stent in reducing shunt dysfunction could be established. Hence, more trials should be recommended (Figure 7). Compared with the GRADE assessment of imprecision which was downgraded one level because the sample size did not meet the optimal information size (Table 1), the Trial Sequential Analysis assessment on shunt dysfunction downgraded the evidence two levels for imprecision.

7.

Shunt dysfunction (same stent diameter)
Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (same stent diameter) for TIPS in people with liver cirrhosis on the outcome of shunt dysfunction.

The diversity‐adjusted required information size (DARIS) was 1226 calculated based upon a proportion of shunt dysfunction of 50% in the control group; a relative risk reduction (RRR) of 20%; an alpha of 1.25% (type I error); a beta of 20% (type II error); and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted CI was 0.42 (0.15 to 1.71). The blue curve presents the cumulative meta‐analysis Z‐score and the inward‐sloping red curves present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The triangular area present the area of futility.

Trial Sequential Analysis (different stent diameter)

All‐cause mortality at maximum follow‐up

Trial Sequential Analysis showed that the DARIS was 2000 participants. DARIS was calculated based on a proportion of death of 33.3% amongst participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 2.5%; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve did not cross the conventional boundary, based on the three included trials (269 participants), neither did it cross the trial sequential monitoring boundaries for benefit or harm, suggesting there was no conclusive evidence that ePTFE‐covered stent could reduce mortality (Figure 8). Therefore, we downgraded the Trial Sequential Analysis assessment of all‐cause mortality for imprecision by two levels, which was consistent with the GRADE assessment (Table 2).

8.

All‐cause mortality (different stent diameter)

Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (different stent diameter) for TIPS in people with liver cirrhosis on the outcome of all‐cause mortality.

The diversity‐adjusted required information size (DARIS) of 2000 was calculated based upon a proportion of all‐cause mortality of 33.3% in the controlled group; a relative risk reduction (RRR) of 20%; an alpha of 2.5% (type I error); a beta of 20% (type II error); and and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted CI was 0.75 (0.12 to 4.50). The blue curve presents the cumulative meta‐analysis Z‐score and the red lines at the left bottom comers present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The inner triangle represents the area of futility.

Procedure‐related complications

Because of insufficient information, Trial Sequential Analysis could not be constructed. We downgraded the Trial Sequential Analysis assessment of procedure‐related complications for imprecision by two levels, which was consistent with the GRADE assessment (Table 2).

Upper gastrointestinal bleeding

Because of insufficient information, Trial Sequential Analysis could not be constructed. We downgraded the Trial Sequential Analysis assessment of upper gastrointestinal bleeding for imprecision by two levels, which was consistent with the GRADE assessment (Table 2).

Recurrence of ascites

Because of insufficient information, Trial Sequential Analysis could not be constructed. Compared with the GRADE assessment of imprecision, which was downgraded one level because the sample size did not meet the optimal information size (Table 2), the Trial Sequential Analysis assessment of imprecision on recurrence of ascites was downgraded by two levels.

Hepatic encephalopathy

Trial Sequential Analysis showed a DARIS of 2345 participants. DARIS was calculated based on a proportion of hepatic encephalopathy of 33.3% of participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 1.25%; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve did not cross either the conventional or the trial sequential monitoring boundaries based on the three included trials (269 participants), demonstrating none of the interventions reached superiority and that the trial sequential monitoring boundaries of futility were not reached (Figure 9). Therefore, we downgraded the Trial Sequential Analysis assessment of hepatic encephalopathy for imprecision by two levels, which was consistent with the GRADE assessment (Table 2).

9.

Hepatic encephalopathy (different stent diameter)

Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (different stent diameter) for TIPS in people with liver cirrhosis on the outcome of hepatic encephalopathy.

The diversity‐adjusted required information size (DARIS) 2345 was calculated based upon a proportion of hepatic encephalopathy of 33.3% in the controlled group; a relative risk reduction (RRR) of 20%; an alpha of 1.25% (type I error); a beta of 20% (type II error); and and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted CI was 0.93 (0.24 to 3.65). The blue curve presents the cumulative meta‐analysis Z‐score and the two red lines on the left bottom present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The inner triangle represents the area of futility.

Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt

Because of insufficient information, Trial Sequential Analysis could not be constructed. Compared with the GRADE assessment of imprecision which was downgraded one level because the sample size did not meet the optimal information size, the Trial Sequential Analysis assessment of imprecision on early thrombosis was downgraded two levels.

Kidney failure

Because of insufficient information, Trial Sequential Analysis could not be constructed.

Segmental liver ischaemia

Because of insufficient information, Trial Sequential Analysis could not be constructed.

Shunt dysfunction

Trial Sequential Analysis showed a DARIS of 1077 participants. DARIS was calculated based on a proportion of shunt dysfunction of 53.6% amongst participants in the bare stents group; an RRR of 20%; an alpha (type I error) of 1.25%; a beta (type II error) of 20%; and a diversity of 10%. The cumulative Z‐curve crossed the conventional boundary, based on the three included trials (269 participants). It did not cross the trial sequential monitoring boundary for benefit or harm, which suggested no firm evidence of ePTFE‐covered stent in reducing shunt dysfunction could be established. Hence, more trials should be recommended (Figure 10). Compared with the GRADE assessment of imprecision, which was downgraded one level because the sample size did not meet the optimal information size (Table 2), the Trial Sequential Analysis assessment of imprecision on shunt dysfunction was downgraded two levels.

10.

Shunt dysfunction (different stent diameter)

Trial Sequential Analysis of expanded polytetrafluoroethylene (ePTFE)‐covered stents versus bare stents (different stent diameter) for TIPS in people with liver cirrhosis on the outcome of shunt dysfunction.

The diversity‐adjusted required information size (DARIS) 1077 was calculated based upon a proportion of shunt dysfunction of 53.6% in the controlled group; a relative risk reduction (RRR) of 20%; an alpha of 1.25% (type I error); a beta of 20% (type II error); and an assumed squared diversity (D) of 10%. The 95% Trial Sequential Analysis adjusted CI was 0.50 (0.04 to 5.88). The blue curve presents the cumulative meta‐analysis Z‐score and red lines at the left bottom present the adjusted threshold for statistical significance according to the two‐sided trial sequential monitoring boundaries. The inner triangle represents the area of futility.

Harms identified in non‐randomised studies identified with the searches for randomised clinical trials

We also searched the adverse events for participants who received ePTFE‐covered stents during TIPS procedure in non‐randomised studies. However, we did not identify any other adverse events that were not reported in the randomised trials. For ePTFE‐vascular grafts that were used in liver transplantation or vascular replacement, there were some serious adverse events, such as graft migration, lethal bleeding, graft infection, and perigraft seroma (Hsu 2017; Jung 2020; Kavalam 2019; Reyes Valdivia 2021; Shan 2021).

Discussion

Comparison 1: expanded polytetrafluoroethylene‐covered stents versus bare stents of the same diameter

For the comparison of ePTFE‐covered stents versus bare stents of the same diameter, we identified one trial with a high risk of bias and a small sample size (258 participants; Wang 2016a; Table 1). Low‐certainty evidence showed that ePTFE‐covered stents were associated with lower mortality at five‐year follow‐up compared with the bare stent equivalent. In addition, there was a lower incidence of upper gastrointestinal bleeding, recurrence of ascites, and shunt dysfunction in the ePTFE‐covered stent group compared with the bare stent group of the same diameter. There was no evidence on health‐related quality of life, procedure‐related complications, or other predefined secondary outcomes.

Comparison 2: expanded polytetrafluoroethylene‐covered stents versus bare stents of different diameters

For the comparison of ePTFE‐covered stents versus bare stents of different diameters, we identified three trials at high risk of bias (Bureau 2007; Huang 2010; Perarnau 2014). Very low‐certainty evidence showed a difference in mortality between the ePTFE‐covered stents versus bare stents of different diameters at a range of eight to 24 months' follow‐up despite using different brands of covered stents. In addition, low‐certainty evidence showed that the ePTFE‐covered stents, compared with bare stents of different diameters despite the brands of covered stents, may be associated with a lower incidence of recurrence of ascites. We do not know if ePTFE‐covered stents can be associated with a lower incidence of early thrombosis as the certainty of evidence was very low. For other reported outcomes, there was no evidence of a difference between ePTFE‐covered stents and bare stents. There was no evidence on health‐related quality of life. However, the results of predefined outcomes should be interpreted carefully due to very low‐certainty evidence.

Overall completeness and applicability of evidence

In this review, we included all randomised clinical trials that enroled people with liver cirrhosis, diagnosed by liver biopsy or typical clinical signs, regardless of the aetiology, manifestation, and severity of the cirrhosis. Most trials addressed efficacy outcomes of ePTFE‐covered stents versus bare stents. However, two trials at high risk of bias provided no details regarding adverse events, and none of the trials reported health‐related quality of life or non‐serious adverse events, except for segmental liver ischaemia. None of the included trials reported outcomes for participants who deteriorated or had a non‐functioning stent, and who may have been eligible for a liver transplant. Furthermore, although previous meta‐analyses have indicated that TIPS placement can significantly improve transplant‐free survival time (Qi 2014; Salerno 2007), we did not identify any comparative data relating to time to liver transplantation in the four included trials. This was also the case for the positioning of the TIPS; where previous research has suggested that placement of the stent in the left portal vein branch may stem the development of hepatic encephalopathy (Luo 2019; Qi 2014). We identified no evidence in the present trials to add to this growing body of evidence.

Quality of the evidence

Overall, the certainty of the evidence was low or very low, downgraded because of study limitations such as high risk of performance and detection bias, and imprecision. Two trials had a high risk of performance and detection bias due to the reporting of non‐blinding of participants, personnel, and assessors. One trial had a high risk of performance bias due to the reporting of a single‐blind setting. One trial had a high risk of attrition bias due to the reporting of more than 10% dropout rate and non‐reporting of ITT analysis. One trial did not clearly describe the random sequence generation and thus had an unclear risk of selection bias. Two trials did not clearly report adverse events, and thus had an unclear risk of reporting bias. One trial did not specify the funding source and had an unclear risk of other bias. Imprecision was mainly due to a small sample size, low event rate, or wide CIs.

Potential biases in the review process

We classified one study (two reports) published as conference abstracts as awaiting classification because the process of randomisation was unclear, and they did not contain sufficient information for data analysis. Furthermore, we received no reply from the study authors. If this trial had been eligible for inclusion, we might have been able to present additional relevant evidence regarding the efficacy of using covered versus bare stents in TIPSs for the management of people with liver cirrhosis.

In addition, only one trial reported no missing data. For the analysis, we used ITT data set from two trials to minimise the potential bias. However, one trial did not provide ITT data. The use of per‐protocol data set from that trial may lead to potential attrition bias.

Agreements and disagreements with other studies or reviews

We found three previous meta‐analyses that compared the use of covered versus bare stents in TIPSs for managing people with liver cirrhosis (Qi 2017; Triantafyllou 2018; Yang 2010).

The meta‐analysis by Qi 2017 included the same four trials included in our review, and assessed overall survival, shunt patency, and hepatic encephalopathy (527 participants). They did not assess procedure‐related complications, upper gastrointestinal bleeding, recurrence of ascites, and early thrombosis. In addition, Qi and colleagues did not take the diameters of the stents into account when comparing covered and bare stents. Compared with Qi 2017, results from our review remained consistent in terms of mortality and shunt dysfunction when compared the stents of the same diameter, but were inconsistent when comparing stents of different diameters. Moreover, contrary to Qi 2017, we did not observe any evidence of a difference in hepatic encephalopathy between ePTFE‐covered versus bare stents despite the stents' diameters.

Yang 2010 included six studies, but only one of them was a randomised trial.

Triantafyllou 2018 included 14 studies with various study designs. Amongst them, there were four randomised trials, which were the same as the ones we included. They assessed shunt patency, rebleeding, hepatic encephalopathy, and overall survival. They did not take the diameter of the stents into account when comparing ePTFE‐covered stents and bare stents.

Authors' conclusions

Implications for practice.

This review provides low‐ or very low‐certainty evidence on the use of expanded polytetrafluoroethylene (ePTFE)‐covered stents in transjugular intrahepatic portosystemic shunt (TIPS) procedures for managing people with liver cirrhosis. Due to the low‐ or very low‐certainty evidence, we cannot conclude with certainty about the effects of ePTFE‐covered stents on the evaluated outcomes compared to bare stents. Based on the results, ePTFE‐covered stents of the same diameter may decrease mortality and reduce the incidence of shunt dysfunction compared with the bare stent equivalent. In addition, the incidence of upper gastrointestinal bleeding and refractory ascites may also decrease. Data are equivocal regarding the incidence of hepatic encephalopathy as it occurred similarly in the covered and bare stent group. Of different diameters, the ePTFE‐covered stents may be associated with a lower incidence of recurrence of ascites and early thrombosis. But the benefits of reducing mortality, procedure‐related complication, upper gastrointestinal bleeding, hepatic encephalopathy, kidney failure, shunt dysfunction, and segmental liver ischaemia, compared with the bare stents, remain unknown. Overall, the high risk of systematic error (bias) and imprecision bias of the included trials suggest that caution must be applied to these results until further, higher‐quality evidence is conducted that will help us conclude with greater certainty.

None of the four trials reported data on health‐related quality of life, and data on complications were either missing or rarely reported. We lack high‐quality trials to evaluate the role of ePTFE‐covered stents for TIPSs for managing people with liver cirrhosis.

Implications for research.

We need larger, long‐term, randomised clinical trials at a low risk of bias that compare the use of the same diameter covered TIPS versus bare TIPS for the management of people with liver cirrhosis and portal hypertension. These trials should include people with any liver cirrhosis and portal hypertension and assess both the efficacy and the harms of using covered stents in TIPSs. In addition, trials should assess quality of life, and be designed following the SPIRIT (Standard Protocol Items: Recommendation for the Interventional Trials) statement (SPIRIT 2013a; SPIRIT 2013b) and reported following the CONSORT guidelines (Moher 2001; www.consort-statement.org/).

History

Protocol first published: Issue 9, 2016

Appendices

Appendix 1. Search strategies

Database	Date range	Search strategy
The Cochrane Hepato‐Biliary Group Controlled Trials Register (via the Cochrane Register of Studies Web)	28 February 2023	((polytetrafluoroethylen* or ePTFE) and stent*) AND (((transjugular intrahepatic or peritone*) and (shunt* or anastomos*)) or TIPS)
Cochrane Central Register of Controlled Trials in the Cochrane Library	2023, Issue 2	#1 MeSH descriptor: [Polytetrafluoroethylene] explode all trees #2 MeSH descriptor: [Stents] explode all trees #3 (polytetrafluoroethylen* or ePTFE) and stent* #4 (#1 and #2) or #3 #5 MeSH descriptor: [Portasystemic Shunt, Transjugular Intrahepatic] explode all trees #6 MeSH descriptor: [Peritoneovenous Shunt] explode all trees #7 ((transjugular intrahepatic or peritone*) and (shunt* or anastomos*)) or TIPS #8 #5 or #6 or #7 #9 #4 and #8
MEDLINE Ovid	1946 to 28 February 2023	1. exp Polytetrafluoroethylene/ 2. exp Stents/ 3. 1 and 2 4. ((polytetrafluoroethylen* or ePTFE) and stent*).mp. [mp=title, book title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 5. 3 or 4 6. exp Portasystemic Shunt, Transjugular Intrahepatic/ 7. exp Peritoneovenous Shunt/ 8. (((transjugular intrahepatic or peritone*) and (shunt* or anastomos*)) or TIPS).mp. [mp=title, book title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 9. 6 or 7 or 8 10. 5 and 9 11. (randomized controlled trial or controlled clinical trial or retracted publication or retraction of publication).pt. 12. clinical trials as topic.sh. 13. (random* or placebo*).ab. or trial.ti. 14. 11 or 12 or 13 15. exp animals/ not humans.sh. 16. 14 not 15 17. 10 and 16
Embase Ovid	1974 to 28 February 2023	1. exp politef/ 2. exp stent/ 3. 1 and 2 4. ((polytetrafluoroethylen* or ePTFE) and stent*).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 5. 3 or 4 6. exp transjugular intrahepatic portosystemic shunt/ 7. exp peritoneum vein shunt/ 8. (((transjugular intrahepatic or peritone*) and (shunt* or anastomos*)) or TIPS).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 9. 6 or 7 or 8 10. 5 and 9 11. Randomized controlled trial/ or Controlled clinical study/ or randomization/ or intermethod comparison/ or double blind procedure/ or human experiment/ or retracted article/ 12. (random$ or placebo or parallel group$1 or crossover or cross over or assigned or allocated or volunteer or volunteers).ti,ab. 13. (compare or compared or comparison or trial).ti. 14. ((evaluated or evaluate or evaluating or assessed or assess) and (compare or compared or comparing or comparison)).ab. 15. (open adj label).ti,ab. 16. ((double or single or doubly or singly) adj (blind or blinded or blindly)).ti,ab. 17. ((assign$ or match or matched or allocation) adj5 (alternate or group$1 or intervention$1 or patient$1 or subject$1 or participant$1)).ti,ab. 18. (controlled adj7 (study or design or trial)).ti,ab. 19. (erratum or tombstone).pt. or yes.ne. 20. or/11‐19 21. (random$ adj sampl$ adj7 ('cross section$' or questionnaire$ or survey$ or database$1)).ti,ab. not (comparative study/ or controlled study/ or randomi?ed controlled.ti,ab. or randomly assigned.ti,ab.) 22. Cross‐sectional study/ not (randomized controlled trial/ or controlled clinical study/ or controlled study/ or randomi?ed controlled.ti,ab. or control group$1.ti,ab.) 23. (((case adj control$) and random$) not randomi?ed controlled).ti,ab. 24. (Systematic review not (trial or study)).ti. 25. (nonrandom$ not random$).ti,ab. 26. 'Random field$'.ti,ab. 27. (random cluster adj3 sampl$).ti,ab. 28. (review.ab. and review.pt.) not trial.ti. 29. 'we searched'.ab. and (review.ti. or review.pt.) 30. 'update review'.ab. 31. (databases adj4 searched).ab. 32. (rat or rats or mouse or mice or swine or porcine or murine or sheep or lambs or pigs or piglets or rabbit or rabbits or cat or cats or dog or dogs or cattle or bovine or monkey or monkeys or trout or marmoset$1).ti. and animal experiment/ 33. Animal experiment/ not (human experiment/ or human/) 34. or/21‐33 35. 20 not 34 36. 10 and 35
LILACS (Bireme)	1982 to 28 February 2023	((polytetrafluoroethylen$ or ePTFE) and stent$) [Words] and (((transjugular intrahepatic or peritone$) and (shunt$ or anastomos$)) or TIPS) [Words]
Science Citation Index Expanded (Web of Science)	1900 to 28 February 2023	#5 #4 AND #3 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=(((transjugular intrahepatic or peritone*) and (shunt* or anastomos*)) or TIPS) #1 TS=((polytetrafluoroethylen* or ePTFE) and stent*)
Conference Proceedings Citation Index – Science (Web of Science)	1990 to 28 February 2023	#5 #4 AND #3 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=(((transjugular intrahepatic or peritone*) and (shunt* or anastomos*)) or TIPS) #1 TS=((polytetrafluoroethylen* or ePTFE) and stent*)
Wan Fang Data	1900 to 22 September 2022	(覆膜 or polytetrafluoroethylene* or ePTFE) and 支架) AND (((经颈静脉 or 肝内) and (分流 or 转流)) or TIPS or 门体静脉分流 or 门体分流)
China National Knowledge Internet (CNKI)	1915 to 22 September 2022	(覆膜 or polytetrafluoroethylene* or ePTFE) and 支架) AND (((经颈静脉 or 肝内) and (分流 or 转流)) or TIPS or 门体静脉分流 or 门体分流)

Data and analyses

Comparison 1. ePTFE‐covered stents versus bare stents – same stent diameter.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 All‐cause mortality – RR	1	258	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.43, 0.92]
1.2 Upper gastrointestinal bleeding	1	258	Risk Ratio (M‐H, Random, 95% CI)	0.54 [0.35, 0.84]
1.3 Recurrence of ascites	1	258	Risk Ratio (M‐H, Random, 95% CI)	0.42 [0.20, 0.87]
1.4 Hepatic encephalopathy – RR	1	258	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.76, 1.61]
1.5 Shunt dysfunction – RR	1	258	Risk Ratio (M‐H, Random, 95% CI)	0.42 [0.28, 0.61]

Comparison 2. ePTFE‐covered stents versus bare stents – different stent diameter.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 All‐cause mortality – RR	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.48, 1.16]
2.2 All‐cause mortality – HR (logrank)	2		Hazard Ratio (IV, Random, 95% CI)	0.82 [0.50, 1.36]
2.3 Procedure‐related complications	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.53 [0.05, 5.57]
2.4 Upper gastrointestinal bleeding	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.46 [0.15, 1.38]
2.5 Recurrence of ascites	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.30 [0.11, 0.85]
2.6 Hepatic encephalopathy – RR	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.66, 1.30]
2.7 Hepatic encephalopathy – HR (logrank)	2		Hazard Ratio (IV, Random, 95% CI)	0.63 [0.29, 1.37]
2.8 Kidney failure	1	121	Risk Ratio (M‐H, Random, 95% CI)	7.59 [0.40, 143.92]
2.9 Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt	3	261	Risk Ratio (M‐H, Random, 95% CI)	0.28 [0.09, 0.82]
2.10 Segmental liver ischaemia	1	80	Risk Ratio (M‐H, Random, 95% CI)	5.25 [0.26, 106.01]
2.11 Shunt dysfunction – RR	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.50 [0.28, 0.92]
2.12 Shunt dysfunction – HR (logrank)	2		Hazard Ratio (IV, Random, 95% CI)	0.41 [0.17, 0.99]

Comparison 3. Subgroup analysis: types of covered stents.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 All‐cause mortality	2	140	Risk Ratio (M‐H, Random, 95% CI)	0.46 [0.10, 2.03]
3.1.1 Trials with only Viatorr stents	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.66 [0.37, 1.18]
3.1.2 Trials with only Fluency stents	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.11 [0.01, 1.98]
3.2 Upper gastrointestinal bleeding	2	140	Risk Ratio (M‐H, Random, 95% CI)	0.43 [0.10, 1.96]
3.2.1 Trials with only Viatorr stents	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.19, 3.30]
3.2.2 Trials with only Fluency stents	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.17 [0.02, 1.30]
3.3 Recurrence of ascites	2	140	Risk Ratio (M‐H, Random, 95% CI)	0.15 [0.03, 0.81]
3.3.1 Trials with only Viatorr stents	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.13 [0.02, 1.00]
3.3.2 Trials with only Fluency stents	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.20 [0.01, 4.00]
3.4 Hepatic encephalopathy	2	140	Risk Ratio (M‐H, Random, 95% CI)	0.91 [0.48, 1.71]
3.4.1 Trials with only Viatorr stents	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.95 [0.43, 2.08]
3.4.2 Trials with only Fluency stents	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.83 [0.28, 2.44]
3.5 Thrombosis within 12 weeks after transjugular intrahepatic portosystemic shunt	2	140	Risk Ratio (M‐H, Random, 95% CI)	0.11 [0.01, 0.81]
3.5.1 Trials with only Viatorr stents	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.15 [0.01, 2.81]
3.5.2 Trials with only Fluency stents	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.08 [0.00, 1.31]
3.6 Shunt dysfunction	2	140	Risk Ratio (IV, Random, 95% CI)	0.35 [0.19, 0.64]
3.6.1 Trials with only Viatorr stents	1	80	Risk Ratio (IV, Random, 95% CI)	0.35 [0.16, 0.79]
3.6.2 Trials with only Fluency stents	1	60	Risk Ratio (IV, Random, 95% CI)	0.36 [0.15, 0.87]

Comparison 4. Sensitivity analysis (fixed‐effect model) – ePTFE‐covered stents versus bare stents (different stent diameters).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 All‐cause mortality – RR	3	269	Risk Ratio (M‐H, Fixed, 95% CI)	0.72 [0.50, 1.05]

Comparison 5. Sensitivity analysis (missing data – best‐worst) – ePTFE‐covered stents versus bare stents (same stent diameter).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 All‐cause mortality – RR	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
5.1.1 Without best‐worst case scenario analysis	1	258	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.43, 0.92]
5.1.2 With best‐worst case scenario analysis	1	288	Risk Ratio (M‐H, Random, 95% CI)	0.48 [0.33, 0.68]

Comparison 6. Sensitivity analysis (missing data – best‐worst) – ePTFE‐covered stents versus bare stents (different stent diameter).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 All‐cause mortality – RR	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
6.1.1 Without best‐worst case scenario analysis	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.48, 1.16]
6.1.2 With best‐worst case scenario analysis	3	277	Risk Ratio (M‐H, Random, 95% CI)	0.70 [0.48, 1.00]

Comparison 7. Sensitivity analysis (missing data – worst‐best) – ePTFE‐covered stents versus bare stents (same stent diameter).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 All‐cause mortality – RR	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
7.1.1 Without worst‐best case scenario analysis	1	258	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.43, 0.92]
7.1.2 With worst‐best case scenario analysis	1	288	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.65, 1.28]

Comparison 8. Sensitivity analysis (missing data ‐ worst‐best) – ePTFE‐covered stents versus bare stents (different stent diameter).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
8.1 All‐cause mortality – RR	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
8.1.1 Without worst‐best case scenario analysis	3	269	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.48, 1.16]
8.1.2 With worst‐best case scenario analysis	3	277	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.44, 1.43]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Bureau 2007.

Study characteristics
Methods	Trial design: parallel‐group, multicentre, randomised clinical trial Blinding: none Trial duration: 2 years Setting: hospital Location: France, Spain, Canada
Participants	Total sample size (at randomisation): 80 participants Age (mean): 55 (SD 10) years in intervention group; 55 (SD 12) years in control group Sex (male/female): 27/12 in intervention group; 30/11 in control group Diagnostic criteria of cirrhosis: documented by previous liver biopsy or typical clinical signs Child‐Pugh classification: not reported Child‐Pugh score (mean): 9 (SD 2) in intervention group; 9 (SD 2) in control group Cause of cirrhosis: alcohol (44 participants) Indications for TIPS: uncontrolled variceal bleeding (23 participants); recurrent variceal bleeding after failure of the usual pharmacological and endoscopic methods (25 participants); refractory ascites (32 participants) Portocaval gradient before TIPS (mean): 20 (SD 7) mmHg in intervention group; 20 (SD 5) mmHg in control group Portocaval gradient after TIPS (mean): 7 (SD 4) mmHg in intervention group; 7 (SD 4) mmHg in control group Inclusion criteria: uncontrolled variceal bleeding; refractory ascites or recurrent variceal bleeding after failure of the usual pharmacological and endoscopic methods; presence of cirrhosis, as documented by previous liver biopsy or typical clinical signs; aged 18–75 years Exclusion criteria: history of chronic or recurrent encephalopathy; portal vein thrombosis; hepatocellular carcinoma; cardiac failure; hepatic polycystosis; intrahepatic bile duct dilation; unable to give informed consent; pregnant or breastfeeding women
Interventions	Intervention group: e‐PTFE covered stents (39 participants) Manufacturer: Viatorr (Gore, Flagstaff, Arizona, USA) (39 participants) Stent diameter: 10.5 (SD 0.9) mm Duration of follow‐up: mean 585 days; median 678 days Control group: bare stents (41 participants) Manufacturer: Memotherm Flexx (BARD, Voisins le Bretonneux, France) (22 participants); Wallstent (Boston Scientific, Barcelona, Spain) (15 participants); Luminex (BARD, Ontario, Canada) (2 participants); Sinus Stent (MEDCARE, Franconville, France) (2 participants) Stent diameter: 11.7 (SD 0.8) mm Duration of follow‐up: mean 430 days; median 322 days
Outcomes	Primary outcomes All‐cause mortality at 2 years Procedure‐related complications Secondary outcomes Shunt dysfunction at 2 years Upper gastrointestinal bleeding at 2 years Recurrence of ascites at 2 years Hepatic encephalopathy at 2 years Early thrombosis (thrombosis within 12 weeks after TIPS) Outcomes not used (not predefined in our protocol): number of reinterventions (5 angioplasties and 4 coaxial new stents) at 2 years; results of the analyses for factors associated with outcomes
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "80 were finally randomised …" "Assignment to treatment group was obtained by phone from Toulouse. Adjustment was performed in blocks of 10 patients (5 in each group)." (p.470, Bureau 2004) Comments: block randomisation performed by an independent centre.
Allocation concealment (selection bias)	Low risk	Quote: "Patients were enrolled by senior physicians participating in the trial to coated or uncoated stent groups by using opaque sealed envelopes." (p.470, Bureau 2004) Comments: allocation concealment ensured.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Patients and physicians were not blinded to the type of prosthesis." (p.470, Bureau 2004) Comments: not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "Investigators were not blinded to the type of prosthesis used." (p.473, Bureau 2004) Comments: not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comments: study authors did not report any missing data. ITT analysis was performed.
Selective reporting (reporting bias)	Low risk	Comments: trial protocol available and the review outcomes of interest all‐cause mortality, procedure‐related complications, shunt dysfunction rate, and hepatic encephalopathy were reported. No discrepancy in reporting of outcomes between the study protocol and the full published study report. All outcomes stated in protocol were reported.
Other bias	Low risk	Comments: none obvious.

Huang 2010.

Study characteristics
Methods	Trial design: parallel‐group, single‐centre, randomised clinical trial Blinding: none Trial duration: 12 months Setting: hospital Location: China
Participants	Total sample size (at randomisation): 60 participants Age (mean): 50 (SD 10) years in intervention group; 47 (SD 11) years in control group Sex (male/female): 23/7 in intervention group; 22/8 in control group Diagnostic criteria of cirrhosis: proved by typical clinical signs or liver biopsy Child‐Pugh classification (A/B/C): 5/17/8 in intervention group; 6/17/7 in control group Child‐Pugh score (mean): 8.6 (SD 2.1) in intervention group; 8.4 (SD 2.0) in control group Cause of cirrhosis: viral (57 participants); alcohol (3 participants) Indications for TIPS: uncontrolled active bleeding (51 participants); refractory ascites (9 participants) Portocaval gradient before TIPS (mean): before: 25.7 (SD 6.6) mmHg in intervention group; 26.5 (SD 5.1) mmHg in control group Portocaval gradient after TIPS (mean): 9.5 (SD 2.9) mmHg in intervention group; 13.2 (SD 1.5) mmHg in control group Inclusion criteria: presence of cirrhosis, as proved by typical clinical signs or liver biopsy; uncontrolled variceal bleeding; refractory ascites; recurrent variceal bleeding not amenable to usual pharmacological and endoscopic methods; aged 18–75 years Exclusion criteria: primary prevention of variceal bleeding; congestive heart failure; multiple hepatic cysts; uncontrolled systemic infection; unrelieved biliary obstruction; severe pulmonary hypertension; hepatoma; portal vein thrombosis; severe coagulopathy
Interventions	Intervention group: covered stents (30 participants) Manufacturer: Fluency (Angiomed GmbH Co, USA) (30 participants) Stent diameter: 8 mm Duration of follow‐up (mean): 6.16 (SD 3.89) months Control group: bare stents (30 participants) Manufacturer: Wallstents (Angiomed GmbH Co, USA) (30 participants) Stent diameter: 10 mm Duration of follow‐up (mean): 8.34 (SD 4.42) months
Outcomes	Primary outcomes All‐cause mortality at 12 months Secondary outcomes Patency rate (measured by portosystemic pressure gradient or Doppler ultrasonography) at 3, 6, 12 months Upper gastrointestinal bleeding at 6, 12 months Recurrence of ascites at 6, 12 months Hepatic encephalopathy at 6, 12 months Early thrombosis Outcomes not used (not predefined in our protocol): all‐cause mortality at 3, 6 months; portosystemic pressure gradient at 12 months; ultrasound measured portal venography. Unable to use (no data): procedure‐related complications (the authors stated "no major complication had occurred" in the Result section of their publication. As we found this information insufficiently clear, we did not include it in our data analysis. We wrote to the authors but received no reply).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Randomisation was confirmed by fax or mail." (p.354, Huang 2010) Comment: insufficient information about the sequence generation process to permit judgement of low or high risk.
Allocation concealment (selection bias)	Low risk	Quote: "Patients were allocated to covered or uncovered stent groups by using opaque sealed envelopes." (p.354, Huang 2010) Comments: allocation concealment ensured.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Patients and physicians were not blinded to the type of prosthesis." (p.354, Huang 2010) Comments: not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Quote: "Patients and physicians were not blinded to the type of prosthesis." (p.354, Huang 2010) Comments: not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comments: no missing data.
Selective reporting (reporting bias)	Unclear risk	Comments: adverse events were not reported clearly. We did not have access to the trial protocol to compare consistency of reporting through to the final published trial report.
Other bias	Low risk	Comment: none obvious.

Perarnau 2014.

Study characteristics
Methods	Trial design: parallel‐group, multicentre, randomised controlled trial Blinding: participants and outcome assessors were blinded Trial duration: 2 years Setting: hospital Location: France
Participants	Total sample size (at randomisation): 137 participants Age (mean): 57 (range 53–64) years in intervention group; 59 (range 52–65) years in control group Sex (male/female): 43/22 in intervention group; 59/12 in control group Diagnostic criteria of cirrhosis: documented by histological or typical clinical signs Child‐Pugh classification (A/B/C): 14/35/16 in intervention group; 8/53/9 in control group Child‐Pugh mean score: 8 (range 7–9) in intervention group; 8 (range 7–9) in control group Cause of cirrhosis: alcoholic (113 participants); viral (18 participants); non‐alcoholic steatohepatitis (7 participants); others (3 participants) Indications for TIPS: prevention of rebleeding (42 participants); refractory ascites (100 participants); hydrothorax (9 participants) Portocaval gradient before TIPS (mean): 16 (range 13–20) mmHg in intervention group; 15 (range 12–17) mmHg in control group Portocaval gradient after TIPS (mean): 6 (range 4–8) mmHg in intervention group; 5.5 (range 4–7) mmHg in control group Inclusion criteria: people who need TIPS for validated indication according to the Baveno IV criteria; aged 18–75 years; cirrhosis previously documented by histological or typical clinical signs; and Child‐Pugh score C < 12 Exclusion criteria: acute uncontrolled haemorrhage; total portal thrombosis; hepatocellular carcinoma; heart failure; pulmonary hypertension; hepatic polycystosis; dilation of intrahepatic bile ducts; history of recurrent spontaneous hepatic encephalopathy; hepatic arterial insufficiency; pregnancy, breastfeeding, or childbearing potential without contraception
Interventions	Intervention group: covered stents (66 participants) Manufacturer: Fluency (Bard, Tempe, Arizona, USA) (1 participant); Fluency + Bare (19 participants); Viatorr (Gore, Flagstaff, Arizona, USA) (40 participants) Stent diameter: 10 mm (could be dilated from 8 mm to 10 mm) Duration of follow‐up (median): 23.6 (interquartile range 14.0–24.1) months Control group: bare stents (71 participants) Luminexx (Bard, Tempe, Arizona, USA) (11 participants); Palmaz genesis (Cordis Bridgewater, New Jersey, USA) (18 participants); Smart (Cordis Bridgewater, New Jersey, USA) (12 participants); Wallstent (Boston Scientific, Natick, Massachusetts, USA) (13 participants); Zilver (Cook Bloomington, Indiana, USA) (7 participants). Stent diameter: 10 mm (could be dilated from 8 mm to 10 mm) Duration of follow‐up: mean 585 days; median 21.8 (interquartile range 5.7–24.1) months
Outcomes	Primary outcomes All‐cause mortality at 2 years Secondary outcomes Shunt dysfunction at 2 years Upper gastrointestinal bleeding Recurrence of ascites Hepatic encephalopathy Renal failure Early thrombosis Outcomes not used (not predefined in our protocol): all‐cause mortality at early (≤ 1 month); early procedure‐related complications (≤ 1 month). Unable to use (no data to calculate): quality of life (12‐item Short Form) at 2 years.
Notes	This trial reported related information of early complications (within 30 days). Participants in the covered stents group experienced fever (6 participants), peritoneal bleeding (1 participant), sepsis (1 participant), and miscellaneous complications including acute cytolysis, hepatic encephalopathy, haemolysis, tachycardia, etc. (9 participants), while participants in the bare stents group experienced fever (3 participants), peritoneal bleeding (1 participant), haemobilia (1 participant), stent migration (1 participant), cervix haematoma (1 participant), and miscellaneous complications including acute cytolysis, hepatic encephalopathy, haemolysis, and tachycardia (10 participants).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomisation was done the day before the TIPS procedure, using a secure, computer‐generated, interactive web‐response system." (p.963, Perarnau 2014) Comments: the investigators described a random component in the sequence generation process.
Allocation concealment (selection bias)	Low risk	Quote: "Sequence generation was handled by the Tours Clinical Investigation Centre. It was stratified by centres and performed using blocks of variable sizes." (p.963, Perarnau 2014) Comments: allocation concealment ensured.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "In the present trial, only the patients were blinded." (p.963, Perarnau 2014) Comments: blinding of participants ensured.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "These radiologists were not aware of the patients' randomisation groups and had no practice at all with vascular stents." (p.963, Perarnau 2014) Comments: blinding of outcome assessment ensured.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "All analyses were conducted on a sub‐sample of patients who had TIPS, i.e., after excluding randomized patients whose procedure was cancelled. By doing this, we did exclude some randomized patients from the analysis, however we reasoned that there was no logical link between the allocation result and the fact that these patients did not undergo TIPS. We are therefore confident that excluding such patients did not introduce any bias." Comment: ITT analysis performed on 129 participants (62 in the covered stents group and 67 in the bare stents group).
Selective reporting (reporting bias)	Low risk	Comment: the trial protocol was available and most prespecified outcomes of interest in the review were reported between both the trial protocol and the full published trial report. All outcomes stated in trial protocol were reported.
Other bias	Low risk	Comment: none obvious.

Wang 2016a.

Study characteristics
Methods	Trial design: parallel‐group, single‐centre, randomised controlled trial Blinding: double‐blind Trial duration: 5 years Setting: hospital Location: China
Participants	Total sample size (at randomisation): 288 participants Age (mean): 45.4 (SD 7.0) years in intervention group; 46.7 (SD 5.0) years in control group Sex (male/female): 88/43 in intervention group; 76/51 in control group Diagnostic criteria of cirrhosis: not reported Child‐Pugh classification (A/B/C): 38/58/35 in intervention group; 29/61/37 in control group Child‐Pugh score (mean): 6.98 (SD 1.4) years in intervention group; 7.10 (SD 1.8) years in control group Cause of cirrhosis: posthepatitic (206 participants); other (52 participants) Indications for TIPS: prevention of rebleeding (245 participants); refractory ascites or hydrothorax (42 participants) Portocaval gradient after TIPS (mean): not reported Portocaval gradient before TIPS (mean): not reported Inclusion criteria: portal hypertension, people with defined indications for TIPS treatment; scheduled for elective TIPS; aged 18–70 years Exclusion criteria: combined with hepatic encephalopathy before the treatment; combined with portal vein thrombosis; combined with malignant liver tumour or malignancies at the other sites; combined with haemorrhage of gastrointestinal ulcer; cases underwent emergent TIPS; people who met the inclusion criteria but did not accept the randomly assigned stents, or did not establish the shunt channel as expected due to specific conditions in the operation Note: people who were lost to follow‐up or not followed up as scheduled, underwent liver transplantation, developed liver cancer or other malignancies after the operation, or died of unrelated diseases failed to meet the criteria during the operation or follow‐up
Interventions	Intervention group: covered stents (144 participants) Manufacturer: Fluency (Bard, USA) (131 participants)a Stent diameter: 8 mm Duration of follow‐up: 5 years Control group: bare stents (144 participants) Manufacturer: Smart (Cordis, USA) (127 participants)a Stent diameter: 8 mm Duration of follow‐up: 5 years
Outcomes	Primary outcomes All‐cause mortality at 5 years Secondary outcomes Shunt dysfunction at 5 years (rate of secondary interventional after follow‐up for 5 years; cumulative restenosis rate at 5 years) Recurrence of upper gastrointestinal bleeding at 5 years Recurrence of refractory hydrothorax/ascites at 5 years Hepatic encephalopathy at 5 years Outcomes not used (not predefined in our protocol): all‐cause mortality at 1, 2, 3, 4 years; cumulative restenosis rate at 1, 2, 3, 4 years
Notes	aWe contacted authors to request data from participants who were lost to follow‐up, but received no response.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Each patient was assigned with an ID from 1 to 288 in order on admission, then a random number generator was used to randomly divide the 288 patients into group 0 or 1, with 144 patients in each." (p.6, Wang 2016a) Comment: the investigators described a random component in the sequence generation process.
Allocation concealment (selection bias)	Low risk	Quote: "The results of randomisation were sealed in envelops." (p.6, Wang 2016a) Comment: allocation concealment ensured.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "This was a double‐blinded trial. Patients and physicians allocated to the intervention group were unaware of the allocated arm, outcome assessors and data analysts were kept blinded to the allocation." (p.6, Wang 2016a) Comment: blinding of participants and personnel ensured.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "This was a double‐blinded trial … outcome assessors and data analysts were kept blinded to the allocation." (p.6, Wang 2016a) Comments: blinding of outcome assessment ensured.
Incomplete outcome data (attrition bias) All outcomes	High risk	Comment: the participants who were lost to follow‐up or not followed up as scheduled underwent liver transplantation, developed liver cancer or other malignancies after the operation, or died of unrelated diseases were excluded as censored cases to ensure the accuracy, randomness, and controllability of the present trial. There were 288 participants who met the inclusion criteria (with 144 in the intervention and control group each); however, 30 participants failed to meet the criteria during the operation or follow‐up. Therefore, 258 participants (131 in the intervention group and 127 in the control group) completed the study. ITT analysis was not performed.
Selective reporting (reporting bias)	Unclear risk	Comment: adverse events were not reported clearly. We did not have access to the trial protocol to compare consistency of reporting through to the final published trial report.
Other bias	Low risk	Comment: none obvious.

ITT: intention to treat; SD: standard deviation; TIPS: transjugular intrahepatic portosystemic shunt.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Angeloni 2004	Prospective, non‐randomised trial.
Boatta 2009	Randomised clinical trial. Compared covered stents with different diameters between groups, but did not compare covered stents with bare stents.
Bureau 2015	Randomised clinical trial. The main objective was to compare the liver transplantation‐free survival of people with cirrhosis and recurrent ascites between those treated by covered TIPS and those treated by large volume paracentesis + albumin infusion.
Jung 2009	Retrospective, non‐randomised trial.
Mittal 2010	Retrospective, non‐randomised trial.
Riggio 2010	Randomised clinical trial. Compared covered stents with different diameters between groups, but did not compare covered stents with bare stents.
Wang 2016b	Randomised clinical trial. Compared covered plus bare stents versus covered stents.
Wu 2010	Retrospective, non‐randomised trial.

TIPS: transjugular intrahepatic portosystemic shunt.

Characteristics of studies awaiting classification [ordered by study ID]

Bandi 2010.

Methods	Trial design: randomised clinical trial Blinding: awaiting full text Trial duration: 2 years Setting: awaiting full text Location: awaiting full text
Participants	Total sample size (at randomisation): 66 participants Setting: awaiting full text Age (mean): awaiting full text Sex (male/female): awaiting full text Diagnostic criteria of cirrhosis: awaiting full text Child‐Pugh classification (A/B/C): awaiting full text Child‐Pugh score (mean): awaiting full text Cause of cirrhosis: awaiting full text Indications for TIPS: awaiting full text Portocaval gradient after TIPS (mean): awaiting full text Portocaval gradient before TIPS (mean): awaiting full text Inclusion criteria: awaiting full text Exclusion criteria: awaiting full text
Interventions	Intervention group: covered stents (33 participants) Awaiting full text Control group: bare stents (33 participants) Awaiting full text
Outcomes	TIPS dysfunction; clinical relapse; survival rates
Notes	Conference abstract only; insufficient detail to ensure inclusion or exclusion.

TIPS: transjugular intrahepatic portosystemic shunt.

Differences between protocol and review

We removed the domain 'For‐profit bias' from the risk of bias domains text as it is part of the GRADE domain – publication bias.

We assessed segmental liver ischaemia as a non‐serious adverse event.

Contributions of authors

Review development: PZ, SD, PS, APB, YS, XC, QZ, TL.

Study screening: PZ, PS, APB.

Data extraction: PZ, PS, QZ.

Data analysis: XC, YS, SD, TL, PZ.

Report writing: PZ, TL, SD.

Future update: LT in the existing team will be responsible for future updates.

Sources of support

Internal sources

• Wuhan Union Hospital, China

  Support accesses to databases

External sources

• Cochrane Hepato‐Biliary, Denmark

  Provided help with the review preparation until publication, including the peer review process

Declarations of interest

PZ: none.

SD: none.

PS: none.

APB: none.

YS: none.

XC: none.

QZ: none.

TL: none.

New