NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
National Guideline Centre (UK). Cirrhosis in Over 16s: Assessment and Management. London: National Institute for Health and Care Excellence (NICE); 2016 Jul. (NICE Guideline, No. 50.)
10.1. Introduction
Oesophageal varices, which develop as a result of portal hypertension, are found in approximately 30% of people with cirrhosis at the time of diagnosis. People with cirrhosis without varices at the time of their diagnosis develop them at a rate of 5% (95%CI: 0.8–8.2%) at 1 year and 28% (21.0–35.0%) at 3 years.134 The factors precipitating variceal haemorrhage are still not clear but it is recognised that the risk of bleeding is related not only to the size of the varices (>5 mm) but also to the severity of liver disease and, in people with alcohol-related cirrhosis, whether or not they continue to drink. Once varices are present, they tend to enlarge; thus of people with small varices at the outset 12% (5.6–18.4%) will have large varices at 1 year and 31% (21.2–40.8%) at 3 years134, resulting in a higher risk of bleeding. The estimated 2-year incidence of bleeding is approximately 24%46 and most episodes of bleeding from varices (70%) occur within 2 years of diagnosis.
Although the in-hospital mortality associated with variceal bleeding has decreased in recent years due to improvements in endoscopic therapy and the use of antibiotic prophylaxis, the reported mortality rate, ranging from 12% to 44%, is still substantial. The risk of death within 6 weeks of the initial variceal haemorrhage is related closely to the severity of liver disease, as determined by the Child-Pugh grade: mortality is <10% in Child-Pugh class A compared to >32% in those in Child-Pugh class C.28
As approximately 30% of people with cirrhosis with oesophageal varices develop bleeding and 12–44% die as a result of the first bleed, prophylactic regimens to prevent bleeding have been developed. Non-selective beta-blocker therapy has been the main pharmacological approach for the primary prophylaxis of variceal haemorrhage because these drugs reduce azygos blood flow and variceal pressure.46 Endoscopic variceal ligation (EVL) has been advocated as an option for primary prophylaxis.100 Although EVL is a relatively simple endoscopic procedure, during which elastic bands are placed around the varices, repeated endoscopies are required both to achieve eradication of varices by EVL and for surveillance for variceal recurrence. Since there are 2 different treatment approaches, the GDG decided to examine the clinical and cost- effectiveness of non-selective beta-blockers and endoscopic band ligation both individually and head-to-head for the primary prevention of bleeding in patients with oesophageal varices due to cirrhosis.
10.2. Review question 1: What is the clinical and cost-effectiveness of non-selective beta-blockers for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis?
Review question 2: What is the clinical and cost-effectiveness of endoscopic band ligation for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis?
Review question 3: What is the clinical and cost-effectiveness of non-selective beta-blockers compared with endoscopic band ligation for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis?
For full details see review protocol in Appendix C.
Table 68
PICO characteristics of review question.
10.3. Clinical evidence
We searched for randomised trials comparing the effectiveness of endoscopic band ligation or oral non-selective beta-blockers in the primary prophylaxis of variceal bleeding in people with oesophageal varices due to cirrhosis. The non-selective beta-blockers considered were propranolol and carvedilol. Propranolol is licensed in the UK for the prophylaxis of variceal bleeding in portal hypertension (up to a maximum of 160 mg twice daily). Carvedilol is not licensed in the UK for this particular indication however the GDG wanted to include this evidence in the review, as carvedilol is currently widely used for this indication. Only 2 studies in the comparison of endoscopic band ligation versus non-selective beta-blockers used carvedilol. As evidence for non-selective beta-blockers was combined within the same class, any recommendation made would be for non-selective beta-blockers as a class and not for either propranolol or carvedilol individually.
For comparison 1 (non-selective beta-blockers versus placebo or no intervention), 9 papers reporting 5 studies were included in the review.8,42,93,153-155,158,159,193 All studies used propranolol as the intervention and the control group received placebo. Two of the studies153,159 were in populations of people with medium or large oesophageal varices and were analysed in this stratum. One study193 was in a population of people with small oesophageal varices and was analysed in this stratum. Two studies8,42 were in people with varices of all sizes, however they provided a subgroup analysis of small versus medium and large varices for the following outcomes and these data were analysed within these strata (Andreani 1990: variceal bleeding and upper gastrointestinal bleeding; Conn 1991: variceal bleeding). One study was excluded because it only provided data for this comparison in people with varices of all sizes with no subgroup analysis207 (see excluded studies list in Appendix L). The study characteristics are summarised in Table 69 below. Evidence from these studies is summarised in the clinical evidence summaries below (Table 72 and Table 73). See also the study selection flow chart in Appendix E, study evidence tables in Appendix H, forest plots in Appendix K, GRADE tables in Appendix J and excluded studies list in Appendix L.
Table 69
Summary of studies included in the review: non-selective beta-blockers versus placebo or no intervention.
Table 72
Clinical evidence summary: non-selective beta-blockers versus placebo or no intervention: medium or large varices.
Table 73
Clinical evidence summary: non-selective beta-blockers versus placebo or no intervention: small varices.
For comparison 2 (endoscopic band ligation versus no intervention), 5 RCTs were included in the review.118,125,191,221,232 There was some variation between the studies in the number of bands used in each ligation session, and the frequency of band ligation sessions which ranged from every 1–3 weeks (summarised in Table 70). In all studies, band ligation was performed until eradication of varices or until varices were too small to ligate. Sarin 1996191 included 6 people with another underlying cause of portal hypertension and this study was downgraded for population indirectness. Four of the studies118,125,191,221 were in populations of people with medium or large oesophageal varices and were analysed in this stratum. The final study232 was in people with varices of all sizes, however this study did provide a subgroup analysis of small versus medium and large varices for the outcome of upper gastrointestinal bleeding, and this outcome was analysed in these separate strata. Evidence from these studies is summarised in the clinical evidence summaries below (Table 74 and Table 75). See also the study selection flow chart in Appendix E, study evidence tables in Appendix H, forest plots in Appendix K, GRADE tables in Appendix J and excluded studies list in Appendix L.
Table 70
Summary of studies included in the review: endoscopic band ligation versus no intervention.
Table 74
Clinical evidence summary: endoscopic band ligation versus no intervention: medium or large varices.
Table 75
Clinical evidence summary: endoscopic band ligation versus no intervention: small varices.
For comparison 3 (endoscopic band ligation versus non-selective beta-blockers), 1 Cochrane systematic review88 and 2 RCTs198,203 were included in the review. The Cochrane review included 7 conference abstracts using the published data only. Evidence from these 7 studies has been included here however conference abstracts have not been routinely included elsewhere in this review. In total, 25 papers reporting 20 studies were included for this comparison in this review.1,4,6,38,51,53,54,82,88,106,107,117,127,149,161,166,192,194-196,198,203,206,227,234 All the included studies were analysed in the medium to large varices stratum, no studies were identified for the small varices stratum. The study characteristics are summarised in Table 71 below. Evidence from these studies is summarised in the clinical evidence summaries below (Table 76 and Table 77). See also the study selection flow chart in Appendix E, study evidence tables in Appendix H, forest plots in Appendix K, GRADE tables in Appendix J and excluded studies list in Appendix L.
Table 71
Summary of studies included in the review: endoscopic band ligation versus non-selective beta-blockers.
Table 76
Clinical evidence summary: endoscopic band ligation versus non-selective beta-blockers: medium or large varices.
Table 77
Clinical evidence summary: endoscopic band ligation versus non-selective beta-blockers: small varices.
The Cochrane review was partially included because the review did not include all the outcomes specified in our protocol. Consequently, the papers included in the Cochrane were examined individually to extract the additional outcomes: survival as a time-to-event outcome, freedom from variceal bleeding as a time-to-event outcome and hospital admissions. Where the individual studies reported survival and freedom from variceal bleeding as a time-to-event outcome, this was reported instead of mortality and variceal bleeding as dichotomous outcomes. For the outcome of upper gastrointestinal bleeding, if the individual study only reported variceal bleeding, this was also used for the upper gastrointestinal bleeding outcome in the Cochrane review. However, some studies report upper gastrointestinal bleeding (not including variceal bleeding) and some report upper gastrointestinal bleeding from varices and other sources and, where reported, these numbers were used. Analysis of the additional RCTs and of the evidence in comparisons 1 and 2 was also performed in this way.
The Cochrane review population comprised of people with oesophageal varices due to portal hypertension (not specifically portal hypertension due to cirrhosis as an inclusion criterion). All the included studies were check individually and all but 1 study only included people with cirrhosis as the underlying cause of portal hypertension. Sarin 1999192 included 7 people with another underlying cause of portal hypertension and this study was downgraded for indirectness. The Cochrane review specified ‘high risk’ varices. In order to confirm all the studies fell into our predefined stratum of medium or large varices, studies were checked individually. All but 2 studies specifically mentioned the criteria which would fall into the category of medium or large varices. These 2 studies (Chen 1998 and Abdelfattah 2006) did not specify the size of the varices. Abdelfattah 2006 did specify high risk varices and was therefore included in the medium or large varices stratum. Chen 1998 did not specify the size or risk of varices and therefore and was included in the medium or large varices stratum but was downgraded for indirectness. Two studies specified that they included people with cirrhosis on the transplant waiting list (Gheorghe 2002)82;(Norberto 2007)149.
Two studies used carvedilol198,234 and the remaining studies used propranolol. One study (Lo 2004)124 was removed from the Cochrane review analysis as it used nadolol as the non-selective beta-blocker. Nadolol was excluded from the review protocol as it is not licensed or widely used in the UK for this indication.
No evidence was identified for the outcome of quality of life for any of the 3 comparisons.
10.4. Economic evidence
10.4.1. Published literature
One economic evaluation was identified that compared non-selective beta-blockers with band ligation for primary prevention of bleeding in patients with varices.101,106,149 This is summarised in the economic evidence profiles below (Table 78) and the economic evidence table in Appendix I.
Table 78
Economic evidence profile: band ligation versus non-selective beta-blockers.
No relevant economic evaluations were identified that compared non-selective beta-blockers with no prophylaxis.
No relevant economic evaluations were identified that compared band ligation with no prophylaxis.
See also the economic article selection flow chart in Appendix F.
10.4.2. Unit costs
See Tables 92 and 93 in Appendix O.
10.5. Evidence statements
10.5.1. Clinical
People with cirrhosis and small oesophageal varices
- For the comparison of non-selective beta-blockers with placebo or no intervention, no evidence was identified for the critical outcomes of quality of life, mortality as a time-to-event outcome or freedom from variceal bleeding as a time-to-event outcome. Evidence of Very Low quality demonstrated no clinically important difference between beta-blockers and placebo or no intervention for the outcomes of mortality, variceal bleeding and upper gastrointestinal bleeding (1 study with 150 patients, 3 studies with 237 patients and 2 studies with 182 patients for each outcome, respectively).
- For the comparison of endoscopic variceal band ligation with no intervention, no evidence was identified for the critical outcomes of quality of life, survival and freedom from variceal bleeding. Evidence of Very Low quality suggested a clinical harm of band ligation on the outcome of upper gastrointestinal bleeding, but evidence was only available from 1 study in a subgroup analysis of 31 patients with small oesophageal varices.
- For the comparison of endoscopic variceal band ligation with non-selective beta-blockers, no evidence was identified for this population stratum.
People with cirrhosis and medium or large oesophageal varices
- For the comparison of non-selective beta-blockers with placebo or no intervention, no evidence was identified for the critical outcomes of quality of life and freedom from variceal bleeding as a time-to-event outcome. Evidence from 2 studies with 398 patients suggested a clinical harm of beta-blockers on survival, but there was some uncertainty and evidence was of Low quality. Evidence of Very Low quality suggested a clinical benefit of beta-blockers on variceal bleeding (3 studies with 268 patients). Evidence of Moderate quality demonstrated a clinically important benefit of beta-blockers on upper gastrointestinal bleeding and bleeding-related mortality (3 studies with 448 patients and 2 studies with 398 patients, respectively).
- For the comparison of endoscopic variceal band ligation with no intervention, no evidence was identified for the critical outcome of quality of life. Two studies with 253 patients reported survival and freedom from variceal bleeding as time-to-event outcomes. These studies provided Moderate quality evidence demonstrating a clinically important benefit of band ligation on survival and variceal bleeding. Evidence of Very low and Low quality demonstrated a clinically important benefit of band ligation on upper gastrointestinal bleeding and bleeding-related mortality (5 studies with 444 patients and 3 studies with 297 patients, respectively).
- For the comparison of endoscopic variceal band ligation with non-selective beta-blockers, no evidence was identified for the critical outcome of quality of life. Evidence of Moderate quality suggested no clinical difference between band ligation and beta-blockers on survival (7 studies with 790 patients). However, a clinically important benefit of band ligation was observed from Very Low quality evidence reporting mortality as a dichotomous outcome (12 studies with 790 patients). Very Low quality evidence demonstrated a clinically important benefit of band ligation on freedom from variceal bleeding (7 studies with 805 patients reported time-to-event data). A similar clinically important benefit of band ligation on variceal bleeding reported as a dichotomous outcome was demonstrated from Moderate quality evidence (10 studies with 554 patients reported as a dichotomous outcome). There was a clinically important benefit of band ligation on upper gastrointestinal bleeding (Low quality, 20 studies with 1610 patients), bleeding- related mortality (Moderate quality, 15 studies with 1258 patients), hospitalisation (Low quality, 1 study with 89 patients) and lethargy due to beta-blockers (Moderate quality, 2 studies with 163 patients).
10.5.2. Economic
People with cirrhosis and small oesophageal varices
- No relevant economic evaluations were identified.
People with cirrhosis and medium or large oesophageal varices
- For the comparison of non-selective beta-blockers with placebo or no intervention, no relevant economic evaluations were identified.
- For the comparison of endoscopic variceal band ligation with placebo or no intervention, no relevant economic evaluations were identified.
- For the comparison of endoscopic variceal band ligation with non-selective beta-blockers, 1 cost-consequences analysis found that band ligation was more costly and more effective compared to beta-blockers for primary prevention of bleeding in patients with varices (£1,850 more per patient, 0.03 fewer deaths per patient, and 0.03 fewer patients with bleeding episodes). This analysis was assessed as partially applicable with potentially serious limitations.
10.6. Recommendations and link to evidence
| Recommendation |
|
|---|---|
| Research recommendation |
|
| Relative values of different outcomes | The GDG discussed the merits of various outcome measures that compared band ligation to non-selective beta-blockers in the prevention of primary bleeding in people with cirrhosis and oesophageal varices. The GDG agreed that health-related quality of life, survival, and freedom from variceal haemorrhage were the critical outcomes for decision-making. Whilst the GDG agreed that survival and freedom from variceal bleeding reported as time-to-event data (hazard ratio) were the preferred measures for decision-making (as the effect of the interventions on the time elapsed before the event occurred was considered an important factor), they wanted to retain as much evidence as possible for these particular outcomes and also considered dichotomous data even though this was quality downgraded and the GDG placed less weighting on dichotomous outcomes for decision-making. The GDG agreed that important outcomes were hospital admission, hospital length of stay, episodes of primary upper gastrointestinal bleeding (regardless of source), bleeding-related mortality and the occurrence of the adverse event of fatigue with beta-blockers. The GDG consensus of opinion was that band ligation, unlike other interventions such as sclerotherapy, is associated with few long-term complications that should be considered as adverse events. The GDG noted that the outcome of upper gastrointestinal bleeding was heterogeneous in the way it was reported by individual studies. Some studies only reported variceal bleeding, others reported upper gastrointestinal bleeding (excluding variceal bleeding) and others reported upper gastrointestinal bleeding from varices and other sources. |
| Trade-off between clinical benefits and harms | The GDG considered 2 pre-specified population strata separately (people with small oesophageal varices and people with medium or large oesophageal varices). It is widely accepted that the intervention would act differently in people with small varices and in people with medium or large varices. For example, band ligation is an invasive procedure and it is unlikely that it would be considered in people with small varices because of the technical difficulties involved. This was reflected in the evidence, with only 1 study reporting band ligation in people with small varices (subgroup data from a study that included people with both small and large varices). This clinical difference between the two population strata is also reflected in the evidence by the rarity of studies including both people with small varices and people with larger varices. Three studies were identified that recruited both populations overall.8,42,232 Outcomes were only extracted and analysed where the data were provided separately for the 2 strata in a subgroup analysis. Studies that recruited both populations overall and reported outcomes generalised for the whole study population without a subgroup analysis were excluded and listed in the excluded studies list. The GDG did not consider this evidence, as it wished to make separate recommendations for the 2 population strata. People with small oesophageal varices Non-selective beta-blockers versus placebo or no intervention Only 3 studies8,42,193 reported on this comparison. The studies did not report outcomes considered as critical by the GDG (quality of life, survival as time-to-event data, or freedom from variceal bleeding as time-to-event data). Dichotomous outcomes were available for mortality, variceal bleeding and upper gastrointestinal bleeding. The GDG considered that there was no clinically important difference between beta-blockers and placebo or no intervention for any of the outcomes reported. Band ligation versus placebo or no intervention Only 1 study232 compared band ligation with placebo in people with small varices as a subgroup analysis (the study recruited people with all sizes of varices). As with the above comparison, this study did not report on the critical outcomes. Only upper gastrointestinal bleeding was reported and this showed clinical harm for band ligation. However, the GDG noted that this was a very small study with only 1 person in the band ligation group having this outcome. Band ligation versus non-selective beta-blockers There was no evidence for this comparison in people with small varices. People with medium or large oesophageal varices Non-selective beta-blockers versus placebo or no intervention Only 4 studies8,42,153,159 reported this comparison in people with medium or large varices and only 1 critical outcome (survival) was reported which showed a clinical harm for beta-blockers. The GDG noted imprecision in this result and the evidence was of Low quality. Beta-blockers were of clinical benefit over placebo for the outcomes of variceal bleeding, when reported as a dichotomous outcome, and for upper gastrointestinal bleeding and bleeding-related mortality. Band ligation versus placebo or no intervention Five studies118,125,191,221,232 reported this comparison in people with medium or large varices. Band ligation was of clinical benefit over placebo or no intervention for the critical time-to-event outcomes of survival and freedom from variceal bleeding, and for the important outcomes of upper gastrointestinal bleeding and bleeding-related mortality. Band ligation versus non-selective beta-blockers Twenty studies (25 papers)1,4,6,38,51,53,54,82,88,106,107,117,127,149,161,166,192,194-196,198,203,206,227,234 reported this comparison for people with medium or large varices. No clinically important difference between beta-blockers and band ligation was observed for survival. However, a clinically important benefit of band ligation was reported for mortality (dichotomous outcome), variceal bleeding (when reported as time-to- event or dichotomous data), upper gastrointestinal bleeding, bleeding-related mortality, hospitalisation and lethargy due to beta-blockers. No evidence was available for any of the comparisons for the quality of life outcome for people with medium or large varices. |
| Trade-off between net clinical effects and costs | One cost-consequence analysis (Norberto 2007)149 was identified that directly compared band ligation with beta-blockers in people with medium or large varices. No relevant studies were identified comparing either band ligation or beta-blockers with no intervention, or in people with small varices. Norberto 2007 reported that overall costs were £1,850 greater per person for band ligation compared to beta-blockers with a small (not statistically significant) clinical benefit of 0.03 fewer patients with variceal bleeding and 0.03 fewer deaths per patient. The difference in total costs was mainly due to the higher intervention cost of band ligation, as the follow-up and hospital costs were similar in both arms. However, the GDG noted that this study (also included in the clinical review) had clinical results less favourable for band ligation than the meta-analysed results of the clinical review as a whole. The additional cost of band ligation should therefore be compared against the increased effectiveness shown in the pooled clinical effectiveness figures from the meta-analyses in this chapter, rather than the clinical effectiveness demonstrated in the Norberto 2007 study alone. The GDG also noted that the excess cost of band ligation might be expected to be lower in a study representative of the clinical review. Whilst the higher cost of the initial band ligation procedure would remain, follow-up and hospital costs would be lower for band ligation than for beta-blockers due to the lower rates of variceal bleeding and rehospitalisation in people treated with band ligation. Using the decreased rates of all-cause mortality and variceal bleeding found in the clinical review, with costs of £1,326 for band ligation and £2,653 for treating variceal bleeding (as used in the original economic model for this guideline, see Appendix N) and £56.71 for 1 year of propranolol at 40 mg 3 times per day (NHS Drug Tariff), shows that band ligation would be expected to be cost-effective at a threshold of £20,000 per QALY gained. For band ligation versus no treatment, band ligation would have an incremental cost of £710 (£1,793−£1,083) and incremental effectiveness of 0.597 QALYs, giving an ICER of £1,190 per QALY gained. For band ligation versus beta-blockers, band ligation would have an incremental cost of £1,054 (£1,496−£441) and an incremental effectiveness of 0.072 QALYs, giving an ICER of £14,641 per QALY gained. This assumes that death leads to a loss of 3 QALYs and bleeding leads to a loss of 0.03 QALYs, but does not include the additional financial benefit of decreased subsequent hospitalisations (161 fewer per 1,000 people with band ligation compared to beta-blockers) as the length of these hospitalisations is not known. The GDG noted that these ICERs are estimates, but are consistent with what would be expected if Noberto 2007 was updated with clinical data representative of the clinical review. The GDG concluded that band ligation was likely to be cost-effective compared to beta-blockers for people with medium or large varices at a cost-effectiveness threshold of £20,000 per QALY. As the GDG found insufficient clinical evidence to make a recommendation regarding people with small varices, the economics of treating this patient group were not considered. |
| Quality of evidence | The GDG discussed the included studies and noted the following:
Non-selective beta-blockers versus placebo or no intervention There were only 3 studies of small sample size and the reported outcomes were of Very Low quality. Band ligation versus placebo or no intervention Subgroup evidence from only 1 study of Very Low quality was available. People with medium or large oesophageal varices Non-selective beta-blockers versus placebo or no intervention Evidence for the critical outcomes was of Low and Very Low quality. Evidence for the important outcomes of upper gastrointestinal bleeding and bleeding-related mortality was of Moderate quality. Band ligation versus placebo or no intervention Evidence for the critical outcomes was of Moderate quality and for the important outcomes of upper gastrointestinal bleeding and bleeding-related mortality was of Very low and Low quality. Band ligation versus non-selective beta-blockers Evidence was of Moderate quality for the outcomes of survival, variceal bleeding (dichotomous), bleeding-related mortality and adverse events. For the outcomes of variceal bleeding (time-to-event), upper gastrointestinal bleeding and hospitalisation, evidence was of Low or Very Low quality. |
| Other considerations | Small varices Whilst the GDG noted no clinical benefit of beta-blockers in people with small varices there was a paucity of evidence which was of Very Low quality leading to uncertainty over the true effect of beta-blockers in the stratum. Overall the GDG agreed there was insufficient evidence to make a recommendation in people with small varices and instead chose to develop a research recommendation in this area. Medium or large varices The GDG:
Bleeding from oesophageal varices is a major complication of cirrhosis. Approximately half of patients with cirrhosis have oesophageal varices and one-third of all patients with varices will experience bleeding at some point. Despite improvements in the management of acute haemorrhage in recent decades, the 6-week mortality associated with variceal bleeding remains of the order of 10–20%. Risk of variceal bleeding increases with variceal size. Whether non-selective beta- blockers are of benefit as primary prophylaxis in people with cirrhosis and small oesophageal varices has not been adequately studied. |
- Introduction
- Review question 1: What is the clinical and cost-effectiveness of non-selective beta-blockers for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis? Review question 2: What is the clinical and cost-effectiveness of endoscopic band ligation for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis? Review question 3: What is the clinical and cost-effectiveness of non-selective beta-blockers compared with endoscopic band ligation for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis?
- Clinical evidence
- Economic evidence
- Evidence statements
- Recommendations and link to evidence
- Prophylaxis of variceal haemorrhage - Cirrhosis in Over 16sProphylaxis of variceal haemorrhage - Cirrhosis in Over 16s
Your browsing activity is empty.
Activity recording is turned off.
See more...