Clinical review protocols

National Guideline Centre (UK)

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.)

Cirrhosis in Over 16s: Assessment and Management.

Show details

Contents

< Prev Next >

Appendix CClinical review protocols

C.1. Risk factors and risk assessment tools

C.1.1. Risk factors

Table 1Review protocol: Risk factors

Component	Description
Review question	What are the risk factors that indicate the populations at specific risk for cirrhosis?
Objectives	To estimate the prognostic value of different risk factors to predict the future development of cirrhosis and to facilitate the decision to test for cirrhosis in primary care (that is, those at higher risk of developing cirrhosis in the future should be considered for testing for cirrhosis)
Population	Adults and young people who are 16 years or older
Presence or absence of prognostic variable	Obesity (BMI ≥30, or a lower BMI for people of Asian family origin) Alcohol misuse Viral hepatitis B Viral hepatitis C Type 2 diabetes
Outcomes	Critical outcomes: Cirrhosis: time-to-event. If time-to-event data is not available, categorical data will be used (that is, the relative risk of developing cirrhosis at different time points).
Study design	Prospective and retrospective cohort Systematic reviews of the above
Exclusions	Studies not taking into account all the essential confounding factors at analysis (in multivariate analysis) or design stage. Studies not taking into account all the confounding factors will be considered if no other evidence is available. Studies with univariate analyses if studies with multivariable analysis are available. Studies that do not have at least 10 events per covariate in the multivariate analysis will be downgraded for risk of bias. If sufficient evidence is available, these studies will be excluded.
How the information will be searched	The databases to be searched are Medline, Embase and The Cochrane Library. Studies will be restricted to English language only. No date restriction will be applied.
Key confounders	The following are key confounders for each risk factor. Studies must have taken these confounders into consideration, either by adjusting for in the multivariate analysis or accounting for at design stage (for example excluding people with one of the other risk factors) or describing baseline characteristics between these groups. Obesity (BMI ≥30, BMI >25 for people of an Asian family origin): age, ethnicity, treatments for obesity (weight loss or surgery), all of the other risk factors. Alcohol misuse: gender, age, ethnicity, level and pattern of alcohol misuse, all of the other risk factors. Viral hepatitis B: gender, age, ethnicity, treatment for hepatitis B, all of the other risk factors. Viral hepatitis C: gender, age, ethnicity, treatment for hepatitis C, all of the other risk factors. Type 2 diabetes: gender, age, ethnicity, treatment for type 2 diabetes, all of the other risk factors.
The review strategy	Meta-analysis may be considered, if appropriate. If no other study designs are available, case-control studies will be considered. We will consider whether the severity/level of the prognostic variable (that is, BMI level, level of alcohol consumed, severity of type 2 diabetes) influences the development of cirrhosis, if available in the literature.

C.1.2. Risk tools

Table 2Review protocol: Risk tools

Component	Description
Review question	Are there any validated risk tools that indicate the populations at specific risk for cirrhosis?
Objectives	To assess the discriminative ability and calibration of the risk factor tools in predicting the future risk of cirrhosis
Population	Adults and young people who are 16 years or older Strata: male/female
Risks stratification tools	Any validated risk factor tools
Reference standard/target condition	Development of cirrhosis (confirmed on liver biopsy)
Outcomes (in terms of discrimination/calibration)	Critical outcomes: ROC area under the curve (of each risk tool for each outcome)/concordance c- statistic. Sensitivity, specificity, predictive values. Predicted risk, observed risk/calibration plot (reproduced with author permissions) (that is, predicted x-year mean risk % verses Kaplan-Meier x-year event rate). Narrative of agreement between observed and predicted risk and whether underestimation/overestimation of predicted risk). Other outcomes: D statistics, R2 statistic and Brier score.
Study design	Cohort (preferably prospective)
How the information will be searched	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only. No date restriction will be applied.
The review strategy	Meta-analysis may be considered, if appropriate.

C.2. Diagnostic tests

Table 3Review protocol: Blood fibrosis test

Component	Description
Review question	In people with suspected (or under investigation for) cirrhosis, what is the most accurate blood fibrosis test to identify whether the condition is present (as indicated by the reference standard, liver biopsy)?
Objectives	The first-line approach will be to review RCT test and treat studies. Patients are randomised to one test (with appropriate treatment for a positive result) versus another test (with appropriate treatment for a positive result) and look at patient outcomes. Patient outcomes for test-and-treat studies: Survival (time-to-event) or mortality at 5 years (dichotomous) Health-related quality of life (continuous) Incidence of a decompensating event: ascites, variceal bleeding, HCC, HRS (dichotomous) Adverse effects of testing (dichotomous) Referral to secondary or tertiary care (dichotomous) Need for liver transplant (dichotomous) The second-line approach will be to review diagnostic accuracy studies of each test compared to the reference standard, liver biopsy. Diagnostic accuracy studies will be reviewed unless RCT test and treat studies are available for all index tests.
Study design	RCTs (for test and treat) Cross-sectional studies, cohort studies, case series (including both retrospective and prospective analyses) Exclusions: case control studies
Population	Adults and young people >16 years with suspected (or under investigation for) cirrhosis. Stratify studies based on the underlying cause. Alcohol misuse disorders Hepatitis C Non-alcoholic fatty liver disease People with multiple aetiologies PBC or PSC (reported separately) Studies including mixed aetiologies which do not provide results subgrouped by aetiology will be excluded from the review. Exclusions: Patients under 16 years old General population or patients not suspected to have cirrhosis (not thought to be at- risk population and without signs or symptoms) Current diagnosis of cirrhosis (hepatic decompensation compatible with cirrhosis, encephalopathy, variceal bleeding, ascites) Patients with hepatitis B
Setting	Primary and secondary care
Index test	Blood fibrosis tests: FibroTest for all aetiologies (haptoglobin, α2M, Apo A1, γGT, Bilirubin, age, sex) Enhanced liver fibrosis (ELF) (PIIINP, hyaluronic acid, TIMP-1) Note: ELF has changed since inception and the newer test excludes age as an additional variable). Validated in HCV and some metabolic liver diseases. APRI (aspartate aminotransferase (AST)/platelet ratio index) FIB-4 (platelets, ALT, AST) AST/ALT ratio Only tests that have been validated in an independent validation cohort for the aetiology will be included.
Reference standard/target condition	Cirrhosis diagnosed by liver biopsy using one of the following scoring systems: Knodell score (F4) Ishak fibrosis score (F5 or F6) METAVIR (F4) For NAFLD populations only, specific fibrosis scoring systems defined Kleiner 2005 and Brunt 2001 references Liver biopsy should be at least 6 portal tracts or a length of 15 mm or more. Studies which do not specify this requirement will be excluded, unless no studies with this reference standard are identified (for each aetiology strata). A biopsy less than 25 mm or 10 portal tracts will reduce the accuracy of the reference standard test and the quality of evidence will be downgraded. Exclusions: Studies that used scoring systems other than METAVIR, Ishak and Knodell scores for diagnosis of cirrhosis or using different cut-off values to those specified above to indicate a positive test result. Liver biopsy length or number of portal tracts not stated or less than 15 mm and 6 portal tracts
Statistical measures	Critical outcomes: Specificity Sensitivity Important outcomes: ROC curve or area under curve The GDG set the critical measure for decision making as sensitivity. The GDG set a level of 95% as an acceptable level for the sensitivity of the test (this level will be used to assess imprecision).
Search strategy	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only.
Review strategy	Subgroups where diagnostic tests may be more or less accurate – to investigate heterogeneity: People who are drinking alcohol or have ceased but previously drank alcohol at harmful levels (for the alcohol strata) (>80% with people still drinking; <80%) Appraisal of methodological quality: The methodological quality of each study will be assessed using the QUADAS-II checklist (per target condition). Extract data on the number of valid test readings for use in assessing the methodological quality. Synthesis of data: Diagnostic meta-analysis will be conducted where appropriate using hierarchical methods. If limited evidence is available for each aetiology we will, in order of preference: Consider evidence from conference abstracts and contact the authors Consider extrapolating evidence from another aetiology strata if evidence is available Consider evidence from studies reporting the accuracy in mixed aetiologies

Table 4Review protocol: Non-invasive imaging

Component	Description
Review question	In people with suspected (or under investigation for) cirrhosis, what is the most accurate non-invasive imaging test (transient elastography [fibroscan or ARFI], ultrasound or MR elastography) to identify whether cirrhosis is present (as indicated by the reference standard, liver biopsy)?
Objectives	The first-line approach will be to review RCT test and treat studies. Patients are randomised to one test (with appropriate treatment for a positive result) versus another test (with appropriate treatment for a positive result) and look at patient outcomes. Patient outcomes for test and treat studies: Survival (time-to-event) or mortality at 5 years (dichotomous) Health-related quality of life (continuous) Incidence of a decompensating event: ascites, variceal bleeding, HCC, HRS (dichotomous) Adverse effects of testing (dichotomous) Referral to secondary or tertiary care (dichotomous) Need for liver transplant (dichotomous) The second-line approach will be to review diagnostic accuracy studies of each test compared to the reference standard, liver biopsy. Diagnostic accuracy studies will be reviewed unless RCT test and treat studies are available for all index tests.
Study design	RCTs (for test and treat) Cross sectional studies, cohort studies, case series (including both retrospective and prospective analyses) Exclusions: case control studies
Population/Target condition	Adults and young people >16 years with suspected (or under investigation for) cirrhosis. Stratify studies based on the underlying cause. Alcohol misuse conditions Hepatitis C Non-alcoholic fatty liver disease People with multiple aetiologies PBC or PSC (reported separately) Studies including mixed aetiologies which do not provide results subgrouped by aetiology will be excluded from the review. Exclusions: Patients under 16 years old General population or patients not suspected to have cirrhosis (not thought to be at- risk population and without signs or symptoms) Current diagnosis of cirrhosis (hepatic decompensation compatible with cirrhosis, encephalopathy, variceal bleeding, ascites) Patients with hepatitis B
Setting	Primary and secondary care
Index test	Transient elastography Acoustic radiation force impulse (ARFI) imaging Point shear wave elastography (pSWE) Ultrasound MRI (all forms, including MR elastography) The index test should be carried out according to the manufacturer's guidelines on performance standards (for example in the percentage of the transient elastography scan that needs to be successful for a valid scan). Exclusions: Index tests using ultrasound and liver microbubble transit time.
Reference standard (could be more than one)	Cirrhosis diagnosed by liver biopsy using one of the following scoring systems: Knodell score (F4) Ishak fibrosis score (F5 or F6) METAVIR (F4) For NAFLD populations only, specific fibrosis scoring systems defined Kleiner 2005 and Brunt 2001 references. Liver biopsy should be at least 6 portal tracts or a length of 15 mm or more. Studies which do not specify this requirement will be excluded, unless no studies with this reference standard are identified (for each aetiology strata). A biopsy less than 25 mm or 10 portal tracts will reduce the accuracy of the reference standard test and the quality of evidence will be downgraded. Exclusions: Studies that used scoring systems other than METAVIR, Ishak and Knodell scores for diagnosis of cirrhosis or using different cut-off values to those specified above to indicate a positive test result. Liver biopsy length or number of portal tracts not stated or less than 15 mm and 6 portal tracts.
Statistical measures	Critical outcomes: Specificity Sensitivity Important outcomes: ROC curve or area under curve The GDG set the critical measure for decision making as sensitivity. The GDG set a level of 95% as an acceptable level for the sensitivity of the test (this level will be used to assess imprecision).
Other exclusions	Case-control studies
Search strategy	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only.
Review strategy	Subgroups where diagnostic tests may be more or less accurate – to investigate heterogeneity: Active drinkers and people who have ceased drinking (for the alcohol strata) (>80% with people still drinking; <80%) Appraisal of methodological quality: The methodological quality of each study will be assessed using the QUADAS-II checklist (per target condition). Extract data on the number of valid test readings for use in assessing the methodological quality. Synthesis of data: Diagnostic meta-analysis will be conducted where appropriate using hierarchical methods. If limited evidence is available for each aetiology we will, in order of preference: Consider evidence from conference abstracts and contact the authors Consider extrapolating evidence from another aetiology strata if evidence is available Consider evidence from studies reporting the accuracy in mixed aetiologies

Table 5Review protocol: Blood fibrosis test versus individual blood test

Component	Description
Review question	In people with suspected (or under investigation for) cirrhosis, is a blood fibrosis test more accurate compared to an individual blood test to identify whether the condition is present (as indicated by the reference standard, liver biopsy)?
Objectives	The first-line approach will be to review RCT test and treat studies. Patients are randomised to one test (with appropriate treatment for a positive result) versus another test (with appropriate treatment for a positive result) and look at patient outcomes. Patient outcomes for test-and-treat studies: Survival (time-to-event) or mortality at 5 years (dichotomous) Health-related quality of life (continuous) Incidence of a decompensating event: ascites, variceal bleeding, HCC, HRS (dichotomous) Adverse effects of testing (dichotomous) Referral to secondary or tertiary care (dichotomous) Need for liver transplant (dichotomous) The second-line approach will be to review diagnostic accuracy studies of each test compared to the reference standard, liver biopsy. Diagnostic accuracy studies will be reviewed unless RCT test and treat studies are available for all index tests.
Study design	RCTs (test and treat) Cross sectional studies, cohort studies, case series (including both retrospective and prospective analyses) Exclusions: case control studies
Population	Adults and young people >16 years with suspected (or under investigation for) cirrhosis. Stratify studies based on the underlying cause. Alcohol misuse disorders Hepatitis C Non-alcoholic fatty liver disease People with multiple aetiologies PBC or PSC (reported separately) Studies including mixed aetiologies which do not provide results subgrouped by aetiology will be excluded from the review. Exclusions: Patients under 16 years old General population or patients not suspected to have cirrhosis (not thought to be at-risk population and without signs or symptoms) Current diagnosis of cirrhosis (hepatic decompensation compatible with cirrhosis, encephalopathy, variceal bleeding, ascites) Patients with hepatitis B
Setting	Primary and secondary care
Index test	Individual blood tests: Albumin Platelets Prothrombin Time (INR) AST ALT Bilirubin γGT (alcohol/ cholestasis)
Reference standard/target condition	Cirrhosis diagnosed by liver biopsy using one of the following scoring systems: Knodell score (F4) Ishak fibrosis score (F5 or F6) METAVIR (F4) For NAFLD populations only, specific fibrosis scoring systems defined Kleiner 2005 and Brunt 2001 references. Liver biopsy should be at least 6 portal tracts or a length of 15 mm or more. Studies which do not specify this requirement will be excluded, unless no studies with this reference standard are identified (for each aetiology strata). A biopsy less than 25 mm or 10 portal tracts will reduce the accuracy of the reference standard test and the quality of evidence will be downgraded. Exclusions: Studies that used scoring systems other than METAVIR, Ishak and Knodell scores for diagnosis of cirrhosis or using different cut-off values to those specified above to indicate a positive test result. Liver biopsy length or number of portal tracts not stated or less than 15 mm and 6 portal tracts
Statistical measures	Critical outcomes: Specificity Sensitivity Important outcomes: ROC curve or area under curve The GDG set the critical measure for decision making as sensitivity. The GDG set a level of 95% as an acceptable level for the sensitivity of the test (this level will be used to assess imprecision).
Search strategy	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only.
Review strategy	Subgroups where diagnostic tests may be more or less accurate – to investigate heterogeneity: Active drinkers and people who have ceased drinking (for the alcohol strata) (>80% with people still drinking; <80%) Appraisal of methodological quality: The methodological quality of each study will be assessed using the QUADAS-II checklist (per target condition). Extract data on the number of valid test readings for use in assessing the methodological quality. Synthesis of data: Diagnostic meta-analysis will be conducted where appropriate using hierarchical methods. If limited evidence is available for each aetiology we will, in order of preference: Consider evidence from conference abstracts and contact the authors Consider extrapolating evidence from another aetiology strata if evidence is available Consider evidence from studies reporting the accuracy in mixed aetiologies.

Table 6Review protocol: Non-invasive tests versus blood fibrosis test

Component	Description
Review question	In people with suspected (or under investigation for) cirrhosis, is a combination of 2 non-invasive tests more accurate compared to a blood fibrosis test alone or an imaging test alone to identify whether cirrhosis is present (as indicated by the reference standard, liver biopsy)?
Objectives	The first-line approach will be to review RCT test and treat studies. Patients are randomised to one test (with appropriate treatment for a positive result) versus another test (with appropriate treatment for a positive result) and look at patient outcomes. Patient outcomes for test-and-treat studies: Survival (time-to-event) or mortality at 5 years (dichotomous) Health-related quality of life (continuous) Incidence of a decompensating event: ascites, variceal bleeding, HCC, HRS (dichotomous) Adverse effects of testing (dichotomous) Referral to secondary or tertiary care (dichotomous) Need for liver transplant (dichotomous) The second-line approach will be to review diagnostic accuracy studies of each test compared to the reference standard, liver biopsy. Diagnostic accuracy studies will be reviewed unless RCT test and treat studies are available for all index tests.
Study design	RCTs (for test and treat) Cross sectional studies, cohort studies, case series (including both retrospective and prospective analyses) Exclusions: case-control studies
Population/Target condition	Adults and young people >16 years with suspected (or under investigation for) cirrhosis. Stratify studies based on the underlying cause. Alcohol misuse conditions (narratively report the duration of abstinence before the test) Hepatitis C Non-alcoholic fatty liver disease People with multiple aetiologies PBC or PSC (reported separately) Studies including mixed aetiologies which do not provide results subgrouped by aetiology will be excluded from the review. Exclusions: Patients under 16 years old General population or patients not suspected to have cirrhosis (not thought to be at risk population and without signs or symptoms) Current diagnosis of cirrhosis (hepatic decompensation compatible with cirrhosis, encephalopathy, variceal bleeding, ascites) Patients with Hepatitis B
Setting	Primary and secondary care
Index test	Individual blood fibrosis test versus Individual imaging test versus diagnosis made on the basis of a combination of 2 non-invasive tests (a blood fibrosis test and an imaging test; 2 imaging tests; or 2 blood fibrosis tests) Only blood fibrosis tests that have been validated in an independent validation cohort for the aetiology will be included. The index test should be carried out according to the manufacturer's guidelines on performance standards (for example in the percentage of the transient elastography scan that needs to be successful for a valid scan).
Reference standard (could be more than one)	Cirrhosis diagnosed by liver biopsy using one of the following scoring systems: Knodell score (F4) Ishak fibrosis score (F5 or F6) METAVIR (F4) For NAFLD populations only, specific fibrosis scoring systems defined Kleiner 2005 and Brunt 2001 references. Liver biopsy should be at least 6 portal tracts or a length of 15 mm or more. Studies which do not specify this requirement will be excluded, unless no studies with this reference standard are identified (for each aetiology strata). A biopsy less than 25 mm or 10 portal tracts will reduce the accuracy of the reference standard test and the quality of evidence will be downgraded. Exclusions: Studies that used scoring systems other than METAVIR, Ishak and Knodell scores for diagnosis of cirrhosis or using different cut-off values to those specified above to indicate a positive test result. Liver biopsy length or number of portal tracts not stated or less than 15 mm and 6 portal tracts.
Statistical measures	Critical outcomes: Specificity Sensitivity Important outcomes: ROC curve or area under curve The GDG set the critical measure for decision making as the sensitivity. The GDG set a level of 95% as an acceptable level for the sensitivity of the test (this level will be used to assess imprecision).
Other exclusions	Case-control studies
Search strategy	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only.
Review strategy	Subgroups where diagnostic tests may be more or less accurate – to investigate heterogeneity: Active drinkers and people who have ceased drinking (for the alcohol strata) (>80% with people still drinking; <80%) Appraisal of methodological quality: The methodological quality of each study will be assessed using the QUADAS-II checklist (per target condition). Extract data on the number of valid test readings for use in assessing the methodological quality. Synthesis of data: Diagnostic meta-analysis will be conducted where appropriate using hierarchical methods. If limited evidence is available for each aetiology we will, in order of preference: Consider evidence from conference abstracts and contact the authors Consider extrapolating evidence from another aetiology strata if evidence is available Consider evidence from studies reporting the accuracy in mixed aetiologies

C.3. Severity risk tools

Table 7Review protocol: Severity risk tools

Component	Description
Review question	Which risk assessment tool is the most accurate and cost-effective for predicting the risk of future morbidity and mortality in people with compensated cirrhosis? When (at what severity score on the risk assessment tool) should people with cirrhosis be referred to specialist care?
Objectives	This review focuses on validation studies. The aims of the review are: To find the most accurate severity risk tool by assessing the discriminative ability (for example AUC) and calibration of the tools. To determine a threshold for low and high risk groups, that determines high risk people who should be referred to specialist care, based on: the predicted risk of the outcome at each score the sensitivity and specificity at given cut-off thresholds; for example, a lower threshold would mean additional cost of referral in people that will not have the event (high number of false positives, lower specificity), whereas a higher threshold would mean people who will have the event will not be referred (high number of false negatives, low sensitivity)
Population	Adults and young people >16 years with compensated cirrhosis (no prior decompensating event) Exclusions: People with decompensating cirrhosis (prior decompensating event) Prognosis of outcomes after transplant in patients with end-stage liver disease undergoing transplant. Prognosis of outcomes after TIPS in patients undergoing TIPS
Risks stratification tools	Model for end-stage liver disease (MELD) Child-Pugh (Child-Turcotte-Pugh) UK model for end-stage liver disease (UKELD) Transient elastography Modified risk tools by the addition of the following risk factors: Hepatovenous portal pressure gradient (HVPG) Na (for example MELD-Na) Delta-MELD MELD-EEG Transient elastography Nutrition
Event	Survival A decompensating event (hepatic encephalopathy; ascites; spontaneous bacterial peritonitis [SBP]; variceal bleeding; hepatorenal syndrome [HRS]; jaundice) or hepatocellular carcinoma (HCC) For both outcomes: report separately at different timepoints reported by study (minimum 3 months)
Outcomes (in terms of discrimination/calibration)	ROC area under the curve (of each risk tool for each outcome)/concordance c- statistic Sensitivity, specificity, predictive values Predicted risk, observed risk/calibration plot (reproduced with author permissions) (that is, predicted x-year mean risk % verses Kaplan-Meier x-year event rate). Narrative of agreement between observed and predicted risk and whether underestimation/overestimation of predicted risk) Other outcomes: D statistics, R2 statistic and Brier score
Study design	Cohort (prospective or retrospective). Only include external validation studies (not the development/derivation or internal validation studies).
How the information will be searched	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only. No date restriction will be applied.
The review strategy	Meta-analysis may be considered, if appropriate. If no external validation studies are available, then include internal validation studies but as long as the patients are different (spatially or temporally).

C.4. Surveillance for the early detection of hepatocellular carcinoma (HCC)

Table 8Review protocol: surveillance for the early detection of hepatocellular carcinoma (HCC)

Review question	When and how frequently should surveillance testing be offered for the early detection of hepatocellular carcinoma (HCC) in people with cirrhosis?
Population	Adults and young people (16 and over) with confirmed cirrhosis, without HCC at the start of surveillance, or with a history of HCC prior to surveillance. Population strata (that will not be combined in analysis): None Exclusions: People without cirrhosis (exclude studies recruiting >15% of people without cirrhosis, that is with other stages of fibrosis or risk factors for HCC) People whose cirrhosis is diagnosed before 16 years old People with hepatitis B (exclude studies with mixed aetiologies and >15% of people with hepatitis B) HCC at the start of surveillance or a history of HCC prior to surveillance
Intervention	Intervention: No surveillance Surveillance with ultrasound, with or without serum AFP assay: yearly 6-monthly 3-monthly Exclusions: Studies that evaluate one-time screening instead of surveillance
Comparison	No surveillance versus surveillance Different frequencies of surveillance
Outcomes	Critical outcomes: Transplant-free survival (time-to-event) or mortality at 5 years Health-related quality of life Important outcomes: HCC occurrence Lesion of HCC less than or equal to 3 cm, greater than 3 cm Number of lesions (if multiple lesions) Liver cancer staging (according to Barcelona Clinic Liver Cancer [BCLC] system) Liver transplant
Search	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only.
Study design	RCTs, systematic reviews of RCTs, observational studies, systematic reviews of observational studies.
Review strategy	A meta-analysis will be conducted on RCTs with appropriate outcome data. Subgrouping will occur if there is statistical heterogeneity in meta-analysis results. Subgroups include: Subgroup by aetiology (different risks of HCC depending on the underlying cause) Severity of underlying liver disease: Child-Pugh A or B versus Child-Pugh C Treatment/prior treatment for underlying condition versus not on treatment (for example, if the hepatitis C virus has been treated or not) Minimally important differences – none identified If no evidence is identified from RCT studies, evidence will be considered from observational studies, to investigate the predictive ability of surveillance at different frequencies or no surveillance on patient outcomes, using multivariable analysis adjusting for other confounders. Confounding factors (must be taken into account at analysis or design stage): Age Severity of cirrhosis Aetiology of the liver disease: hepatitis C versus other non-viral causes of cirrhosis Co-existing morbidities Progression of liver disease, treatment of underlying liver disease (for example, abstinence from alcohol or antiviral therapy) Exclusions: Studies not taking into account all the essential confounding factors at analysis (in multivariate analysis) or design stage will be excluded. Studies not taking into account all the confounding factors will be considered if no other evidence is available for each comparison. Studies with univariate analyses will be excluded. Studies with univariate analysis will be considered if studies with multivariable analysis are not available for each comparison. Evidence from studies in people with cirrhosis and a proportion of people with HBV >15% will only be considered if there is no evidence identified using the criteria above.

C.5. Surveillance for the detection of varices

Table 9Review protocol: surveillance for the detection of varices

Review question	How frequently should surveillance testing using endoscopy be offered for the detection of oesophageal varices and isolated gastric varices in people with cirrhosis?
Population	Adults and young people (16 and over) with confirmed cirrhosis, without varices and who have not already been started on primary prophylactic therapy for the prevention of variceal bleeding. Population strata (that will not be combined in analysis): Severity of the underlying liver disease: Child-Pugh A Child-Pugh B and C Exclusions: People whose cirrhosis is diagnosed before 16 years old Oesophageal or gastric varices already present, or on primary prophylaxis for the prevention of variceal bleeding or taking beta-blockers
Intervention	Intervention: endoscopy at: Baseline only Yearly Every 2 years Every 3 years
Comparison	Comparison: endoscopy at: Baseline only Yearly Every 2 years Every 3 years Exclusions: Surveillance endoscopy versus no surveillance endoscopy
Outcomes	Critical outcomes: Survival (time-to-event) or mortality at 5 years Free from variceal bleeding (time-to-event) or variceal bleeding at 5 years Health-related quality of life Important outcomes: Free from varices (time-to-event) Occurrence of moderate or large varices Size of varices Number receiving prophylactic treatment (beta-blockers or EVL)
Search	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only.
Study design	RCTs, systematic reviews of RCTs, observational studies, systematic reviews of observational studies
Review strategy	A meta-analysis will be conducted on RCTs with appropriate outcome data. Subgrouping will occur if there is statistical heterogeneity in meta-analysis results. Subgroups include: Primary biliary cholangitis and primary sclerosing cholangitis versus other aetiologies Alcohol-related cirrhosis versus non-alcohol related cirrhosis Presence of portal hypertension: hepatic venous pressure gradient (HVPG) of <10 mmHg versus HVPG of ≥10 mmHg Treatment/prior treatment for underlying condition versus not on treatment Minimally important differences – none identified. If no evidence is identified from RCT studies, evidence will be considered from observational studies to investigate the predictive ability of surveillance at different frequencies on patient outcomes, using multivariable analysis adjusting for other confounders. Confounding factors (must be taken into account at analysis or design stage): Age Severity of cirrhosis Aetiology of the liver disease Portal hypertension Co-existing morbidities Progression of liver disease, treatment of underlying liver disease (for example, abstinence from alcohol or antiviral therapy)

C.6. Prophylaxis of variceal haemorrhage

Table 10Review protocol: primary prevention of bleeding in people with oesophageal varices due to cirrhosis

Review questions	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? 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? 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?
Objectives	To determine whether non-selective beta-blockers, endoscopic band ligation, or placebo or no intervention is more effective for the primary prevention of bleeding in people with oesophageal varices due to cirrhosis
Review population	Adults and young people (16 years and over) with endoscopically verified oesophageal varices that have never bled, with cirrhosis as the underlying cause.
Interventions and comparators: generic/class; specific/drug	Oral non-selective beta-blockers; carvedilol Oral non-selective beta-blockers; propranolol Band ligation; conventional Band ligation; multiband Placebo No intervention Comparisons: Oral non-selective beta-blockers versus placebo or no intervention Band ligation versus no intervention Oral non-selective beta-blockers versus band ligation Exclusions: Nadolol (not licenced or widely used in the UK for this indication)
Outcomes	Critical Health-related quality of life at end of study (continuous) Survival (with or without transplant) at end of study (time to event) Free from primary variceal bleeding at end of study (time to event) Important Hospital admission at end of study (dichotomous) Hospital length of stay at end of study (continuous) Primary upper gastrointestinal bleeding (irrespective of bleeding source) at end of study (dichotomous) Bleeding-related mortality at end of study (dichotomous) Adverse events: fatigue at end of study (dichotomous)
Study design	Systematic review RCT
Unit of randomisation	Patient
Crossover study	Not permitted
Minimum duration of study	Not defined
Other exclusions	People with current or previous variceal bleeding/variceal haemorrhage/upper gastrointestinal bleeding (as determined by endoscopy) People without cirrhosis who have another cause of varices People with gastric varices
Population stratification	Size of varices (small) Size of varices (medium or large)
Reasons for stratification	Effectiveness of beta-blockers and band ligation expected to be different in people with small varices compared to people with medium or large varices.
Other stratifications	Drugs will be combined within the same drug class irrespective of dose or duration of intervention.
Subgroup analyses if there is heterogeneity	Severity of underlying liver disease at the time of intervention (measured by Child-Pugh score) (Child-Pugh score A; Child-Pugh score B or C): intervention expected to be less effective in people with more severe cirrhosis Age of patient (65 years and under; over 65 years): increased age may reduce effectiveness of intervention
Search criteria	Databases: Medline, Embase and the Cochrane Library Date limits for search: no date restriction Language: studies will be restricted to English language only

C.7. Primary prevention of bacterial infections in cirrhosis and upper gastrointestinal bleeding

Table 11Review protocol: Prevention of bacterial infections in people with confirmed cirrhosis and upper gastrointestinal bleeding

Review question	What is the most clinically- and cost-effective prophylactic antibiotic for the primary prevention of bacterial infections in people with cirrhosis and upper gastrointestinal bleeding?
Guideline condition and its definition	Cirrhosis
Objectives	To determine the most effective antibiotic for primary prevention of bacterial infections in people with cirrhosis and upper gastrointestinal bleeding
Review population	People with cirrhosis and upper gastrointestinal bleeding Adults and young people (16 years and over)
Interventions and comparators: generic/class; specific/drug (All interventions will be compared with each other)	IV: Penicillin (beta-lactams); Amoxicillin IV: Penicillin (beta-lactams); Co-Amoxiclav (Amoxicillin and clavulanic acid [Augmentin]) IV: Penicillin (beta-lactams); Ampicillin IV: Penicillin (beta-lactams); Tazocin IV: Cephalotin (beta-lactams); Cephalotin IV: third generation Cephalosporins (beta-lactams); Cefotaxime IV: third generation Cephalosporins (beta-lactams); Ceftazidime IV: third generation Cephalosporins (beta-lactams); Ceftriaxone IV: Aminoglycoside; Gentamicin IV: Aminoglycoside; Tobramycin IV: Aminoglycoside; Amikacin IV: Quinolones; Ciprofloxacin IV: Quinolones; Pefloxacin IV: Quinolones; Ofloxacin IV: Quinolones; Floxacin IV: Carbopenums; Meropenum IV: Carbopenums; Ertapenem IV: Carbopenums; Impenem IV: Glycopeptide; Vancomycin IV: Glycylcycline; Tigecycline Oral: Quinolones; Ciprofloxacin Oral: Quinolones; Norfloxacin Oral: Quinolones; Pefloxacin Oral: Quinolones; Ofloxacin Oral: Quinolones; Floxacin Oral: Quinolones; Levofloxacin Oral: Quinolones; Moxifloxacin Oral: Penicillin; Amoxycilin Oral: Penicillin; Co-amoxiclav [Augmentin] Oral: Penicillin; Phenoxymethylpenicillin (Penicillin V) Oral: Sulfonamides Trimethoprim Oral: Sulfonamides Trimethoprim/Sulphamethoxazole [Septrin] Oral: Sulfonamides; Co-trimoxazole Oral: third generation Cephalosporin; Cefalexin Oral: Clarythromycin Oral: Erythromycin Oral: Colistin Oral: Clindamycin Oral: Doxycycline Oral: Azithromycin Oral: Metronidazole Combinations; Ceftriaxone (IV) and norfloxacin (oral) (any other combinations of the above)
Comparisons	IV versus oral IV versus IV Oral versus oral Any combinations of drugs above (that is, IV + oral combination versus monotherapy) Exclusions: Placebo/no treatment
Outcomes	Critical outcomes: Occurrence of bacterial infections at end of study (dichotomous) Quality of life at end of study (continuous) All-cause mortality (time to event) Important outcomes: Renal failure at end of study (dichotomous) Length of hospital stay at end of study (continuous) Readmission rate at end of study (continuous) Antibiotic complications (for example Clostridium difficile, diarrhoea) (no minimally important differences identified)
Study design	Systematic review of RCTs RCT
Unit of randomisation	Patient
Crossover study	Not permitted
Minimum duration of study	Not defined
Other exclusions	Bleeding from non-cirrhotic portal hypertension (that is portal vein thrombosis) People with nephrotic syndrome People whose cirrhosis is diagnosed before 16 years of age Other routes of administration other than that specified above Placebo as a comparator Conference abstracts
Subgroup analyses if there is heterogeneity	Severity of the underlying liver disease (Child Pugh A (score 5, 6) – normal decompensation; Child Pugh B (score 7,8,9) – moderate decompensation; Child Pugh C (score 10–15) – decompensated liver disease; MELD categories; Child Pugh mixed categories): degree of underlying liver decompensation at time of haemorrhage may impact on the effectiveness of antibiotics. Different modes of administration (IV administration; IV, then oral administration; oral; other; IV and oral): must give IV initially due to oral bleeding but can then switch to oral antibiotics. They may not be as effective.
Search criteria	Databases: Medline, Embase, The Cochrane Library. Date limits for search: from 2010 onwards (date of Cochrane review search) Language: English language only Systematic review and RCT search filters will be applied.
Review strategy (further details)	A meta-analysis will be conducted on RCTs with appropriate outcome data. If no RCT evidence is identified in full-text publications, conference abstracts will be considered.

C.8. Transjugular intrahepatic portosystemic shunt (TIPS) versus large-volume paracentesis (LVP) for ascites

Table 12Review protocol: TIPS versus LVP

Review question	What is the clinical- and cost-effectiveness of transjugular intrahepatic portosystemic shunt (TIPS) compared with large-volume paracentesis (LVP) with albumin in the management of diuretic-resistant ascites due to cirrhosis?
Guideline condition	Cirrhosis
Objectives	To determine whether TIPS or LVP is more effective in the management of diuretic-resistant ascites due to cirrhosis.
Review population	Adults and young people (16 years and over) with confirmed cirrhosis and diuretic-resistant (or refractory) ascites. Exclude: Patients whose cirrhosis is diagnosed before 16 years old Patients with ascites from causes other than cirrhosis (that is, peritoneum malignancy, heart failure, tuberculosis, pancreatitis, nephrotic syndrome, other causes).
Interventions and comparators: (All interventions will be compared with each other, unless otherwise stated)	TIPS LVP with albumin infusion (includes sequential LVP) Note: TIPS interventions will be considered alone or followed by diuretic treatment. TIPS using either coated or uncoated stents will be considered. Data will be extracted on any concomitant diuretic therapies and the details of the TIPS intervention (for example diameter). Exclusions: LVP without albumin infusion No intervention Placebo
Outcomes	Critical outcomes: Re-accumulation of ascites at end of study (dichotomous) Health-related quality of life at end of study (continuous) Transplant-free survival at 12 months (time to event) Important outcomes: Spontaneous bacterial peritonitis at end of study (dichotomous) Renal failure at end of study (dichotomous) Hepatic encephalopathy at end of study (dichotomous) Length of stay at end of study (continuous) Readmission rate at end of study (dichotomous)
Study design	Systematic review RCT
Unit of randomisation	Patient
Crossover study	Not permitted
Minimum duration of study	None
Subgroup analyses if there is heterogeneity	Severity of underlying liver disease at the time of intervention (measured by MELD) (MELD score <15; MELD score ≥ 15): TIPS intervention expected to be less effective in people with more severe cirrhosis. Age of patient (65 years and under; over 65 years): increased age may reduce effectiveness of TIPS intervention. Current or past encephalopathy (current encephalopathy; past encephalopathy; no encephalopathy): current or past encephalopathy may reduce the effectiveness of TIPS. Type of TIPS stent (coated stents; uncoated stents): TIPS intervention expected to be more effective with interventions using coated stents.
Search criteria	Databases: Medline, Embase and the Cochrane Library Date limits for search: no date restriction Language: studies will be restricted to English language only.

C.9. Primary prevention of spontaneous bacterial peritonitis (SBP) in people with cirrhosis and ascites

Table 13Review protocol: SBP prevention in people with cirrhosis and ascites

Review question	What is the clinical- and cost-effectiveness of antibiotics compared with placebo for the primary prevention of spontaneous bacterial peritonitis (SBP) in people with cirrhosis and ascites?
Guideline condition and its definition	Cirrhosis
Objectives	To estimate the clinical effectiveness of prophylactic oral antibiotics for the primary prevention of SBP in patients with confirmed cirrhosis and ascites.
Review population	Patients with cirrhosis and ascites
	Adults and young people (16 years and over)
Interventions and comparators:	Oral: Quinolones: Ciprofloxacin Oral: Quinolones: Norfloxacin Oral: Quinolones: Pefloxacin Oral: Quinolones: Ofloxacin Oral: Quinolones: Floxacin Oral: Penicillin: Amoxycillin Oral: Penicillin: Co-amoxiclav Oral: Sulfonamides: Co-trimoxazole (Trimethoprim+Sulphamethoxazole) Oral: third generation Cephalosporin: Cefalexin Placebo No intervention Comparisons: Any oral antibiotic (mono-therapy; all classes of antibiotics pooled together) versus placebo/no intervention
Outcomes	Critical: Occurrence of SBP at end of study (dichotomous) All-cause mortality (time to event) Quality of life at end of study (continuous) Important: Incidence of resistant organisms at end of study (dichotomous) Renal failure at end of study (dichotomous) Liver failure at end of study (dichotomous) Length of hospital stay at end of study (continuous) Readmission rate at end of study (dichotomous)
Study design	Systematic review RCT
Unit of randomisation	Patient
Crossover study	Not permitted
Minimum duration of study	None
Other exclusions	People with nephrotic syndrome People whose cirrhosis is diagnosed before 16 years of age People with previous SBP; studies which included more than 15% of patients who had previously had SBP People with variceal bleeding
Subgroup analyses if there is heterogeneity	Severity of the underlying liver disease (Child Pugh 9 or less; Child Pugh >9): severity of underlying liver disease may impact on the effectiveness of antibiotics. Risk of SBP (high risk: ascitic protein level <15 g/litre [1.5 g/dl]; low risk: ascitic protein level ≥15 g/litre [1.5 g/dl]): those at higher risk of SBP are more likely to have the outcome and may be more likely to see an effect of antibiotics. Antibiotic class (Penicillins; Quinolones; third generation Cephalosporins; Sulfonamides): different antibiotic classes may have different effectiveness.
Search criteria	Databases: Medline, Embase, The Cochrane Library. Date limits for search: from 2010 onwards (date of Cochrane review search) Language: English language only Systematic review and RCT search filters will be applied.

C.10. Volume replacers in hepatorenal syndrome

Table 14Volume replacers in hepatorenal syndrome

Review question	Which is the most clinically and cost-effective volume replacer for patients with hepatorenal syndrome due to cirrhosis who are also receiving vasoactive drugs?
Objectives	To estimate the clinical effectiveness and cost-effectiveness of volume replacers in the management of patients with hepatorenal syndrome due to cirrhosis who are also receiving vasoactive drugs.
Population	Adults and young people (16 and over) with confirmed cirrhosis and hepatorenal syndrome. Hepatorenal syndrome is defined as reversible renal dysfunction occurring in patients with cirrhosis (with a serum creatinine 133 micromol/litre and an absence of other identifiable causes of renal failure). People who are also receiving vasoconstrictors (vasopressin, ornipressin, terlipressin, octreotide, midodrine, noradrenaline, norepinephrine, dopamine). Population strata (that will not be combined in analysis): No population strata (type I and type II hepatorenal syndrome will be grouped together in the analysis). Exclusions: People whose cirrhosis is diagnosed before 16 years old Renal failure due to hypovolaemia as defined by sustained improvement of renal function (creatinine decreasing to <133 micromol/litre) following at least 2 days of diuretic withdrawal (if on diuretics), and volume expansion with albumin at 1 g/kg/day up to a maximum of 100 g/day Renal failure due to current or recent treatment with nephrotoxic drugs Renal failure due to parenchymal renal disease People receiving vaptans
Intervention	IV albumin IV crystalloids (Ringer's lactate solution, 0.9% sodium chloride (saline), Hartmann's solution, dextrose) IV polygel, plasma or colloid expanders (group all polygel, plasma or colloid expanders together, for example haemocel, gelofusion/gelofusine, dextran, manitol, voluven)
Comparisons	IV albumin versus IV crystalloids IV albumin versus polygel, plasma or colloid expanders IV crystalloids versus polygel, plasma or colloid expanders Interested in the effect of the volume replacer, therefore the vasoconstrictor type and dose should be the same within both arms of the study.
Outcomes	Critical outcomes: Survival (time-to-event) or mortality at 3 months Health-related quality of life (continuous) Reversal of hepatorenal syndrome or improved renal function (dichotomous – as defined by the study) at 3 months (reduction of serum creatinine below 133 micromol/litre, creatinine clearance, renal function returning to functioning kidneys without the requirement for drugs) Important outcomes: Time to discharge from hospital (time to event) Readmission to hospital (dichotomous) Adverse events of volume replacement (infection) Adverse events of volume replacement (heart failure)
Search	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only. Systematic review and RCT search filters will be applied.
Study designs	RCTs Systematic reviews
Review strategy	A meta-analysis will be conducted on RCTs with appropriate outcome data. Subgrouping will occur if there is statistical heterogeneity in meta-analysis results. Subgroups include: Length of time in established hepatorenal syndrome (less than 24 hours versus more than 24 hours) Aetiology of liver injury (alcohol-related versus non-alcohol related) Albumin (high dose >40 g/day versus low dose <40 g/day) Severity of the underlying liver disease/degree of liver decompensation at the time of hepatorenal syndrome Child-Pugh B (score 7, 8, 9) /moderate decompensation Child-Pugh C (score 10–15) /severe decompensation liver disease Minimally important differences – none identified. If no RCT evidence is identified in full-text publications, conference abstracts will be considered.
Exclusion	Crossover studies, observational studies

C.11. Management of an episode of acute hepatic encephalopathy

Table 15Review protocol: acute hepatic encephalopathy

Review question	What is the most clinically and cost-effective intervention for the first-line treatment of an episode of acute hepatic encephalopathy in people with cirrhosis?
Objectives	To investigate the most clinically and cost-effective intervention for the first-line treatment of an episode of acute encephalopathy. A network meta-analysis (NMA) will be considered.
Population	Adults and young people (16 and over) with confirmed cirrhosis, presenting at their GP or emergency care with an episode of acute hepatic encephalopathy. We will only consider patients in whom hepatic encephalopathy is associated with cirrhosis Hepatic encephalopathy is diagnosed based on clinical observation of a change in mental state associated with known chronic liver disease/cirrhosis based on either biopsy or relevant clinical tests and imaging, with the exclusion of other causes of confusion. Acute hepatic encephalopathy stages 1, 2, 3 and 4 (West Haven Criteria) will be included. Population strata (that will not be combined in analysis): None Exclusions: People whose cirrhosis is diagnosed before 16 years old People with minimal hepatic encephalopathy (sometimes called latent or subclinical) People with chronic hepatic encephalopathy (if acute is not stated in the research paper, there is no definition for when acute hepatic encephalopathy becomes chronic. Inclusion for acute hepatic encephalopathy should be based on the first-line treatment on admission with acute symptoms) Primary or secondary prevention of hepatic encephalopathy Patients in whom hepatic encephalopathy is caused by acute liver failure (may be described as fulminant hepatic failure, sub-acute liver failure) Patients with another underlying cause of confusion/impaired mental state (for example heart failure, hyponatraemia, renal failure, hypoglycaemia)
Intervention	Non-absorbable disaccharides (combined within drug class): Lactulose (including different routes of administration, for example enema) Lactitol Oral non-absorbable antibiotics (with or without sorbitol) (individual drug level, not combined within drug class): Aminoglycosides (Neomycin) Rifaximin Vancomycin Other oral antibiotics (Metronidazole) Phosphate enemas (combined within drug class) Polyethylene gycol electrolyte solution, PEG 3350 Amino acids (IV or oral): l-ornithine-l-aspartate (LOLA) branch chain amino acids (combined within drug class) IV flumazenil Oral probiotics (combined within drug class) Sodium benzoate Oral zinc MARS Combination therapy (any combinations of the above) Placebo/no treatment Exclusions: Second-line treatment Dopaminergic agonists (used for chronic hepatic encephalopathy treatment) Liver dialysis Mannitol enema (not widely used in the UK) Paromomycin (not licenced in the UK) Lactitol versus lactulose studies (as non-absorbable disaccharides will be combined within drug class)
Comparisons	Any head to head comparison (combination or mono therapy) Any intervention versus placebo/no treatment Duration of treatment up to 2 weeks (exclude studies with duration of treatment >2 weeks as this will not be treatment of the acute episode). Note: Drugs will be combined within drug class as defined above Doses as per standard doses in the BNF Different doses and durations of treatment will be combined
Outcomes	Critical outcomes: Survival (time-to-event) No improvement in hepatic encephalopathy (time to event outcome or dichotomous if time to event not reported; improvement defined as a partial or complete resolution of clinical symptoms of hepatic encephalopathy. Some studies may assess improvement using electrophysiological or psychometrical testing, PSE score, or blood plasma ammonia levels) Health-related quality of life (continuous) Important outcomes: Time to discharge from hospital (time to event) Adverse events (diarrhoea, flatulence, abdominal pain, nausea, GI bleeding, renal failure) Note: If performing an NMA, one network will be performed per outcome so limit to 2 critical outcomes (survival and ‘no improvement in hepatic encephalopathy’ outcomes). For other outcomes, direct pairwise comparisons will be presented.
Search	The databases to be searched are Medline, Embase, The Cochrane Library. Studies will be restricted to English language only. Systematic review and RCT search filters will be applied.
Study designs	RCTs and systematic reviews of RCTs Exclusions: Observational studies Crossover studies
Review strategy	A meta-analysis will be conducted on RCTs with appropriate outcome data. Sub-grouping will occur if there is statistical heterogeneity in meta-analysis results. Subgroups include: Grade of acute hepatic encephalopathy (grade 1–2 versus grade 3–4): people with grade 4 hepatic encephalopathy are not able to take oral drugs so the intervention is expected to be less effective. Severity of the underlying liver disease (Child-Pugh A versus Child-Pugh B/C): interventions expected to be more effective in people with less severe underlying liver disease. Minimally important differences – none identified. If no RCT evidence is identified in full-text publications, conference abstracts will be considered.

Bookshelf ID: NBK385230

Contents