Non‐absorbable disaccharides versus placebo/no intervention and lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Summary of findings for the main comparison. Non‐absorbable disaccharides versus placebo/no intervention for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Non‐absorbable disaccharides versus placebo/no intervention for the prevention and treatment of hepatic encephalopathy in people with cirrhosis
Population: prevention and treatment of hepatic encephalopathy in people with cirrhosis Intervention: non‐absorbable disaccharides (lactulose and lactitol) Control: placebo/no intervention Setting: in‐hospital (overt hepatic encephalopathy) and outpatient (minimal hepatic encephalopathy and prevention trials) Duration of follow‐up: the duration depended on the type of encephalopathy with 5 days for acute, 74 days for chronic, 70 days for minimal, and 207 days for prevention of hepatic encephalopathy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Non‐absorbable disaccharides versus placebo/no intervention
Mortality	Study population		RR 0.59 (0.40 to 0.87) when including all RCTs; RR 0.63 (0.41 to 0.97) when including RCTs with a low risk of bias	1487 (24 studies)	⊕⊕⊕⊝ moderate¹	Trial Sequential Analysis: The Trial Sequential Analysis found a beneficial effect of the intervention including all RCTs, but when the analysis only included RCTs with a low risk of bias. Assessment method: Assessed based on the total number of participants who died.
	88 per 1000	49 per 1000 (32 to 75)
	Moderate
	20 per 1000	11 per 1000 (7 to 17)
Hepatic encephalopathy	Study population		RR 0.58 (0.5 to 0.69)	1415 (22 studies)	⊕⊕⊕⊝ moderate²	Trial Sequential Analysis: The Trial Sequential Analysis found a beneficial effect of the intervention including all RCTs, but when the analysis only included RCTs with a low risk of bias. Assessment method: Assessed based on the definitions in included RCTs (number of participants without a clinically relevant improvement of hepatic encephalopathy).
	469 per 1000	272 per 1000 (234 to 323)
	Moderate
	423 per 1000	245 per 1000 (211 to 292)
Serious adverse events	Study population		RR 0.47 (0.36 to 0.6)	1487 (24 studies)	⊕⊕⊕⊝ moderate³	Trial Sequential Analysis: The Trial Sequential Analysis found a beneficial effect of the intervention including all RCTs, but when the analysis only included RCTs with a low risk of bias. Assessment method: Assessed and defined as any untoward medical occurrence that led to death, was life threatening, or required hospitalisation or prolongation of hospitalisation (ICH‐GCP 2007).
	207 per 1000	97 per 1000 (75 to 124)
	Moderate
	142 per 1000	67 per 1000 (51 to 85)
Quality of life (secondary outcome)		No overall estimate available			⊕⊝⊝⊝ very low⁴	We were unable to combine the data into an overall analysis due to unacceptably high heterogeneity. Assessment method: Based on the quality of life questionnaires.
Non‐serious adverse events (secondary outcome)	Study population		RR 2.47 (1.24 to 4.93)	739 (9 studies)	⊕⊝⊝⊝ very low⁵	Assessment method: The outcome includes all adverse events that do not fulfil the criteria for 'serious' (ICH‐GCP 2007).
	106 per 1000	261 per 1000 (131 to 521)
	Moderate
	63 per 1000	156 per 1000 (78 to 311)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; RCT: randomised clinical trial; RR: risk ratio
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹Mortality is downgraded one level to 'moderate quality evidence' because the Trial Sequential Analysis found insufficient evidence when we limited the analysis to include only RCTs with a low risk of bias. ²Hepatic encephalopathy is downgraded one level to 'moderate quality evidence' because none of the RCTs had a low risk of bias in the overall assessment. ³Serious adverse events is downgraded one level to 'moderate quality evidence' because none of the RCTs had a low risk of bias in the overall assessment. ⁴Quality of life is downgraded three levels to 'very low quality evidence' because i) none of the included RCTs had a low risk of bias, ii) the heterogeneity was considerable, and iii) we were unable to combine the data in an overall analysis. ⁵Non‐serious adverse events is downgraded three levels to 'very low quality evidence' because i) none of the included RCTs had a low risk of bias, ii) the confidence intervals were wide (uncertainty), and iii) we were only able to include data from nine RCTs in our meta‐analysis.

Summary of findings 2. Lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis
Population: prevention and treatment of hepatic encephalopathy in people with cirrhosis Intervention: lactulose Control: lactitol Setting: in‐hospital (overt hepatic encephalopathy) and outpatient (minimal hepatic encephalopathy and prevention trials) Duration of follow‐up: the duration depended on the type of encephalopathy with 5 days for acute, 74 days for chronic, 70 days for minimal, and 207 days for prevention of hepatic encephalopathy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Lactulose versus lactitol
Mortality	Study population		RR 1.3 (0.59 to 2.85)	225 (8 studies)	⊕⊝⊝⊝ very low¹	Trial Sequential Analysis: The Trial Sequential Analysis found no evidence to support or refute a difference between the 2 interventions being compared. Assessment method: Assessed based on the total number of participants who died.
	71 per 1000	92 per 1000 (42 to 202)
	Moderate
	0 per 1000	0 per 1000 (0 to 0)
Hepatic encephalopathy	Study population		RR 1 (0.84 to 1.19)	194 (7 studies)	⊕⊝⊝⊝ very low¹	Trial Sequential Analysis: The Trial Sequential Analysis found no evidence to support or refute a difference between the 2 interventions being compared. Assessment method: Assessed based on the definitions in included RCTs (number of participants without a clinically relevant improvement of hepatic encephalopathy).
	286 per 1000	286 per 1000 (240 to 340)
	Moderate
	200 per 1000	200 per 1000 (168 to 238)
Serious adverse events	Study population		RR 1.56 (0.84 to 2.88)	245 (9 studies)	⊕⊝⊝⊝ very low¹	Trial Sequential Analysis: The Trial Sequential Analysis found no evidence to support or refute a difference between the 2 interventions being compared. Assessment method: Assessed based on the definitions in included RCTs (number of participants without a clinically relevant improvement of hepatic encephalopathy.
	106 per 1000	165 per 1000 (89 to 304)
	Moderate
	77 per 1000	120 per 1000 (65 to 222)
Quality of life (secondary outcome)	—	No data were available for this outcome	—	—	—	None of the included RCTs assessed quality of life.
Non‐serious adverse events (secondary outcome)	Study population		RR 1.55 (0.88 to 2.74)	169 (6 studies)	⊕⊝⊝⊝ very low²	Assessment method: The outcome includes all adverse events that do not fulfil the criteria for 'serious' (ICH‐GCP 2007).
	247 per 1000	383 per 1000 (217 to 677)
	Moderate
	246 per 1000	381 per 1000 (216 to 674)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; RCT: randomised clinical trial; RR: risk ratio
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹Mortality, hepatic encephalopathy, and serious adverse events are downgraded three levels to 'very low quality evidence' because i) the Trial Sequential Analysis found insufficient evidence to support or refute a difference between the intervention and control group, ii) the confidence intervals were wide, and ii) none of the included RCTs had a low risk of bias in the overall assessment of bias control. ²Non‐serious adverse events is downgraded three levels to 'very low quality evidence' because i) none of the included RCTs had a low risk of bias in the overall assessment of bias control, ii) only six RCTs reported the outcome, and iii) the confidence intervals were wide (uncertainty).

Background

Description of the condition

The term hepatic encephalopathy refers to a spectrum of neuropsychiatric changes occurring in people with liver disease. The joint guideline from the European and American Associations for the Study of the Liver defines hepatic encephalopathy as a brain dysfunction associated with liver insufficiency or portal systemic shunting (EASL and AASLD guideline 2014a; EASL and AASLD guideline 2014b). Clinically apparent or overt hepatic encephalopathy manifests as a neuropsychiatric syndrome encompassing a wide spectrum of mental and motor disorders (Weissenborn 1998; Ferenci 2002). Events such as gastrointestinal bleeding, infection, and alcohol misuse can trigger this so‐called acute or episodic hepatic encephalopathy. Fifty per cent of instances occur with no obvious cause. Episodes may recur. Between episodes, people may return to their baseline neuropsychiatric status or show clinical evidence of impairment (Bajaj 2010b). Less frequently, people present with persistent neuropsychiatric abnormalities, which are always present to some degree, but may vary in seriousness. Often people with persistent abnormalities have extensive spontaneous portal‐systemic shunting or else a surgically created or transjugular intrahepatic portosystemic shunt (TIPS).

Changes in mental state range from subtle alterations in personality, intellectual capacity, and cognitive function to more profound alterations in consciousness leading to deep coma with decerebrate posturing. The changes in motor function may include rigidity, disorders of speech production, resting‐ and movement‐induced tremor, asterixis, delayed diadochocinetic movements, hyperreflexia, hyporeflexia, choreoathetoid movements, Babinsky's sign, and transient focal symptoms (Victor 1965; Weissenborn 1998; Cadranel 2001). Asterixis (flapping tremor) is the best known motor abnormality. Individuals with overt hepatic encephalopathy also show a wide spectrum of other abnormalities, including impaired psychometric performance (Schomerus 1998), disturbed neurophysiological function (Parsons‐Smith 1957; Chu 1997), altered cerebral neurochemical/neurotransmitter homeostasis (Taylor‐Robinson 1994), reductions in global and regional cerebral blood flow and metabolism (O'Carroll 1991), and changes in cerebral fluid homeostasis (Haussinger 2000). In general, the degree of impairment in these parameters increases as the clinical condition worsens. The term minimal hepatic encephalopathy (in the older literature subclinical or latent) refers to people with cirrhosis who are 'clinically normal', but who show abnormalities in neuropsychometric or neurophysiological performance (Ferenci 2002).

The diagnosis of hepatic encephalopathy may present no problems, but without the background information and an obvious precipitating event, it may go unrecognised. We have no gold standard for the diagnosis (Montagnese 2004), but techniques that we can use singly or in combination. The diagnosis or exclusion of overt hepatic encephalopathy should include a careful and detailed neuropsychiatric history and examination (Montagnese 2004), with particular attention paid to changes in memory, concentration, cognition, and consciousness. Clinicians and researchers often use the West Haven Criteria to grade mental state (Conn 1977), and the Glasgow Coma Score to grade the level of consciousness (Teasdale 1974). The neurological examination should be comprehensive, looking particularly for evidence of subtle motor abnormalities. The assessment should consider and exclude other potential causes of neuropsychiatric abnormalities including concomitant neurological disorders and metabolic abnormalities such as those associated with diabetes, renal failure, drug, or alcohol intoxication. People with hepatic encephalopathy have impaired psychometric performance (Montagnese 2004; Randolph 2009). Those with minimal hepatic encephalopathy show deficits in attention, visuo‐spatial abilities, fine motor skills, and memory while their other cognitive functions are relatively well preserved. People with overt hepatic encephalopathy show additional disturbances in psychomotor speed, executive function, and concentration. Psychometric test batteries to assess cognitive function form part of the evaluation. The Psychometric Hepatic Encephalopathy Score has a high specificity for the diagnosis (Schomerus 1998; Weissenborn 2001). The test employs five paper and pencil tests to assess attention, visual perception and visuo‐constructive abilities. Test scores have to be normalised to take account of factors such as age, gender, and educational level. At present, normative databases are available in Germany, Italy, Denmark, Spain, Mexico, Korea, India, and Great Britain.

People with hepatic encephalopathy may have a number of neurophysiological abnormalities (Guérit 2009). The electroencephalogram, which primarily reflects cortical neuronal activity, may show progressive slowing of the background activity and abnormal wave morphology. Recent advances in electroencephalogram analysis should provide better quantifiable and more informative data. Other potential diagnostic techniques include the Critical Flicker Fusion Frequency (Kircheis 2002), and the Inhibitory Control Test (Bajaj 2008). The tests need further validation. Studies using structural and functional cerebral imaging techniques have helped to unravel the pathophysiology of hepatic encephalopathy, but they currently offer little diagnostically (Grover 2006; Berding 2009).

Description of the intervention

The non‐absorbable disaccharides lactulose and lactitol are poorly absorbed sugars, which act as osmotic laxatives in the treatment of constipation (Johanson 2007; Miller 2014). Lactulose (Montgomery 1929) is dispensed as a syrup, which is contaminated with other sugars; a pure crystalline preparation is also available. Lactitol, a second‐generation disaccharide, is dispensed as a powder. The mode of administration is generally enteral.

How the intervention might work

The exact pathogenesis of hepatic encephalopathy is unknown. Ammonia plays a key role (Butterworth 2014). The main sources of ammonia include nitrogenous products in the diet, bacterial metabolism of urea and proteins in the colon, and deamination of glutamine in the small intestine. Non‐absorbable disaccharides lower ammonia levels through a number of mechanisms: (i) a laxative effect: the colonic metabolism of lactulose and lactitol results in an increase in intraluminal gas formation, an increase in intraluminal osmolality, a reduction in intraluminal pH, and an overall decrease in transit time; (ii) bacterial uptake of ammonia: the intraluminal changes in pH result in a leaching of ammonia from the circulation into the colon. The colonic bacteria use the released volatile fatty acids as substrate and proliferate. In doing so, they use the trapped colonic ammonia as a nitrogen source for protein synthesis. The increase in bacterial numbers additionally 'bulks' the stool and contributes to the cathartic effect; (iii) reduction of intestinal ammonia production: non‐absorbable disaccharides inhibit glutaminase activity and interfere with the intestinal uptake of glutamine and its subsequent metabolism to ammonia; (iv) beneficial effects on the gut microbiome: cirrhosis is associated with dysbiosis and changes in the colonic mucosal microbiome (Qin 2014). Further changes may be observed in patients with hepatic encephalopathy(Bajaj 2012). Non‐absorbable disaccharides can beneficially affect microbiota composition (Riggio 1990b; Bajaj 2012).

Why it is important to do this review

The prevalence of hepatic encephalopathy varies. About 10% to 14% have overt hepatic encephalopathy when first diagnosed with cirrhosis (Saunders 1981). In studies in people with decompensated cirrhosis, about 20% have overt hepatic encephalopathy (D'Amico 1986; de Jongh 1992; Zipprich 2012). The cumulated incidence of overt hepatic encephalopathy is as high as 40% (Randolph 2009; Bajaj 2011a). The prevalence of minimal hepatic encephalopathy varies in different studies, but it may be more than 50% or higher in people with previous overt hepatic encephalopathy (Sharma 2010; Lauridsen 2011). The presence of hepatic encephalopathy, whether minimal or overt, is associated with significant impairment in the performance of complex tasks, such as driving (Schomerus 1981; Bajaj 2009; Kircheis 2009). The condition is also associated with a detrimental effect on quality of life (Groeneweg 1998) and safety (Roman 2011). In addition, the presence of overt hepatic encephalopathy in people with cirrhosis awaiting liver transplantation has a detrimental effect on neurocognitive function following the procedure (Sotil 2009) and on overall survival (Bustamante 1999; D'Amico 2006; Stewart 2007; Bajaj 2011a; Patidar 2014). The survival probability in people with cirrhosis after their first episode of hepatic encephalopathy is 42% at one year and 23% at three years (Bustamante 1999). Thus, more than 50% die within one year and more than 75% within three years. Overt hepatic encephalopathy also poses a substantial burden for the caregivers of affected people (Bajaj 2011b), and a significant financial burden on healthcare systems (Poordad 2007; Stepanova 2012).

Since 1966 (Bircher 1966), when lactulose was first introduced into clinical practice, several RCTs have evaluated non‐absorbable disaccharides for hepatic encephalopathy. Previous meta‐analyses have found that lactitol may be more beneficial than lactulose (Blanc 1992), or that lactulose and lactitol had comparable effects (Camma 1993). The previous versions of this review did not find sufficient evidence to recommend lactulose or lactitol for routine clinical use in people with cirrhosis and hepatic encephalopathy (Als‐Nielsen 2000; Als‐Nielsen 2004a; Als‐Nielsen 2004b; Als‐Nielsen 2005). Methodological issues including unclear bias control and lack of statistical power weakened the strength of the conclusions. A subsequent guideline from the European and American Association for the Study of Liver Diseases recommended lactulose as the intervention of choice for overt hepatic encephalopathy and its secondary prevention after an index event (EASL and AASLD guideline 2014a; EASL and AASLD guideline 2014b). The guideline did not recommend primary prevention of encephalopathy nor the routine treatment of minimal hepatic encephalopathy. Clinicians may consider treating minimal hepatic encephalopathy on a case by case basis under certain circumstances such as impaired driving skills, work performance, quality of life issues, or cognitive impairment. The original Cochrane review and the current European and American Associations for the Study of the Liver guidelines provide discrepant views about the role of lactulose. We therefore conducted this updated review.

Objectives

To assess the beneficial and harmful effects of i) non‐absorbable disaccharides versus placebo/no intervention and ii) lactulose versus lactitol in people with cirrhosis and hepatic encephalopathy.

To avoid overlap with another planned Cochrane review, we did not evaluate non‐absorbable disaccharides versus antibiotics (Kimer 2015).

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs, regardless of publication status, language, or blinding.

Types of participants

We included people with cirrhosis from RCTs on the prevention (primary or secondary) or treatment of hepatic encephalopathy, regardless of sex, age, aetiology of the underlying liver disease, type of hepatic encephalopathy, or precipitating factors.

Types of interventions

The intervention comparisons were i) non‐absorbable disaccharides (lactulose or lactitol) versus placebo/no intervention and ii) lactulose versus lactitol. We included RCTs, irrespective of the doses, treatment durations, and modes of administration and allowed co‐interventions if administered equally to allocation trial arms.

Types of outcome measures

We assessed all outcomes at the maximum duration of follow‐up (Gluud 2015).

Primary outcomes

Mortality.
Hepatic encephalopathy. We based our assessment of hepatic encephalopathy on the definitions in included RCTs.
Serious adverse events defined as any untoward medical occurrence that led to death, was life threatening, or required hospitalisation or prolongation of hospitalisation (ICH‐GCP 2007). We analysed serious adverse events as a composite outcome (Gluud 2015).

Secondary outcomes

Quality of life.
Non‐serious adverse events: all adverse events that did not fulfil the criteria for a serious adverse event.
Surrogate outcomes: Number Connection Test results and blood ammonia concentrations.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Hepato‐Biliary Group Controlled Trials Register, Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, and Science Citation Index Expanded using the strategies described in Appendix 1. The last search update was 19 October 2015.

Searching other resources

We scanned the reference lists of relevant articles and proceedings from meetings of the British Society for Gastroenterology (BSG), the British Association for the Study of the Liver (BASL), the European Association for the Study of the Liver (EASL), the United European Gastroenterology Week (UEGW), the American Gastroenterological Association (AGA), the American Association for the Study of Liver Diseases (AASLD), and the International Society for Hepatic Encephalopathy and Nitrogen Metabolism (ISHEN). We wrote to the principal authors of RCTs and the pharmaceutical companies involved in the production of non‐absorbable disaccharides for additional information about completed RCTs and for information about any ongoing RCTs, and searched the database ClinicalTrials.gov (clinicaltrials.gov/) and the World Health Organization (WHO) online trial meta‐register (apps.who.int/trialsearch/).

Data collection and analysis

Selection of studies

Two review authors (Lise L Gluud and Marsha Y Morgan) read the electronic searches, performed additional manual searches, and listed potentially eligible RCTs. All authors read the potentially eligible trial reports and participated in the final selection of those to be included in the analyses. We reached the final selection through consensus. For RCTs reported in more than one publication, we selected the paper reporting the longest duration of follow‐up as the primary reference. We listed details of all included RCTs (Characteristics of included studies) and excluded studies (Characteristics of excluded studies).

Data extraction and management

Two review authors (Lise L Gluud and Marsha Y Morgan) independently collected data and resolved contrary opinions through discussion. The collected data included information on: i) RCTs: design (cross‐over or parallel), settings (number of clinical sites; outpatient or inpatient; inclusion period), country of origin; ii) participants: mean age, proportion of men, aetiology of cirrhosis, type of hepatic encephalopathy, previous history of hepatic encephalopathy and iii) interventions: type, dose, duration of therapy, mode of administration. We gathered the primary and secondary outcome data, including the criteria used in the assessment of hepatic encephalopathy, and bias control information. A commercial translation services or medical personnel fluent in the language translated foreign language (non‐English) papers (Acknowledgements). We requested missing data and other information from authors of included RCTs.

Assessment of risk of bias in included studies

We assessed bias control using the domains described in the Cochrane Hepato‐Biliary (CHB) module and classified the risk of bias for each domain as high, unclear, or low and the overall assessment as high or low (Gluud 2015).

Allocation sequence generation

Low risk of bias: sequence generation achieved using computer random number generation or a random number table. Drawing lots, tossing a coin, shuffling cards, or throwing dice are 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.

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 (for example, 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 have been 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.

Blinding of participants, personnel, and outcome assessors

Low risk of bias: blinding was performed adequately. We defined lack of blinding (detection and performance bias) as not likely to affect the assessment of the outcome mortality.
Unclear risk of bias: there was insufficient information to assess whether blinding was likely to induce bias in the results.
High risk of bias: no blinding or incomplete blinding, and the assessment of outcomes 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. The investigators used sufficient methods, such as intention‐to‐treat analyses with multiple imputations or carry‐forward analyses 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 induced bias in 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 clinically relevant outcomes (mortality, hepatic encephalopathy, and serious adverse events). If we had access to the original trial protocol, the outcomes selected were those called for in that protocol. If we obtained information from a trial registry (such as www.clinicaltrials.gov), we only used that information if the investigators registered the trial before inclusion of the first participant.
Unclear risk of bias: not all pre‐defined outcomes were reported fully, or it was unclear whether data on these outcomes were recorded or not.
High risk of bias: one or more predefined outcomes were not reported.

For‐profit bias

Low risk of bias: the trial was free of industry sponsorship or other type of for‐profit support that may influence the trial design, conduct, or results.
Unclear risk of bias: no information on clinical trial support or sponsorship was available.
High risk of bias: the trial was sponsored by industry, received support in the form of lactulose, lactitol, or placebo, or received any other type of support.

Other bias

Low risk of bias: the trial appeared to be free of other biases including: medicinal dosing problems or follow‐up (as defined below).
Unclear risk of bias: the trial may or may not have been free of other domains that could put it at risk of bias.
High risk of bias: there were other factors in the trial that could put it at risk of bias such as the administration of inappropriate treatments being given to the controls (e.g. an inappropriate dose) or follow‐up (e.g. the trial included different follow‐up schedules for participants in the allocation groups).

Overall bias assessment

Low risk of bias: all domains were low risk of bias using the definitions described above.
High risk of bias: one or more of the bias domains were of unclear or high risk of bias.

Measures of treatment effect

We used risk ratios (RR) for dichotomous outcomes and the mean differences (MD) for continuous outcomes, both with 95% confidence intervals (CI). For primary outcomes, we calculated the number needed to treat to benefit (NNTB) as 1/ risk difference (RD) based on the highest quality evidence (RCTs with a low risk of bias where available).

Unit of analysis issues

We included data from the first treatment period of cross‐over trials (Higgins 2011a).

Dealing with missing data

We extracted data on all randomised participants in order to allow intention‐to‐treat analyses. To evaluate the importance of missing data, we conducted a worst‐case scenario analysis with simple imputation (Higgins 2008), with inclusion of missing outcomes as treatment failures. We also conducted an 'extreme' worst‐case scenario analysis in which we included missing outcome data as treatment failures (intervention group) or successes (control group).

Assessment of heterogeneity

We expressed heterogeneity as I² values using the following thresholds: 0% to 40% (unimportant), 40% to 60% (moderate), 60% to 80% (substantial), and > 80% (considerable). This information is included in the 'Summary of findings' tables (GRADEpro).

Assessment of reporting biases

For meta‐analyses with at least 10 RCTs, we assessed reporting biases through regression analyses using the Harbord test (Harbord 2006), which regresses Z/sqrt(V) against sqrt(V), where Z is the efficient score and V is Fisher's information (the variance of Z under the null hypothesis). All meta‐analyses of continuous outcomes included fewer than 10 RCTs.

Data synthesis

We performed the analyses in Review Manager 5 (RevMan 2014), STATA (Stata), and Trial Sequential Analysis (Thorlund 2011; TSA 2011).

Meta‐analysis

We undertook random‐effects and fixed‐effect meta‐analyses. Although the conclusion of the two models concurred, the random‐effects meta‐analysis provides the most conservative estimate of intervention effects. Therefore, we report the random‐effects meta‐analyses in our results.

Trial Sequential Analysis

We performed a Trial Sequential Analysis (Higgins 2008; Thorlund 2011), and defined the required information size (also known as the heterogeneity adjusted required information size) as the number of participants needed to detect or reject an intervention effect based on the relative risk reduction (RRR) and CGR. The analyses show firm evidence if the Z‐curve crosses the monitoring boundary (also known as the trial sequential monitoring boundary) before reaching the required information size. We constructed futility boundaries to evaluate the uncertainty of obtaining a chance negative finding and performed the analyses with alpha set to 5%, power to 80%, and model‐based diversity. Based on previous evidence (Thorlund 2011; EASL and AASLD guideline 2014a; EASL and AASLD guideline 2014b), we set the relative risk reduction (RRR) to 30% and the CGR to 15% (mortality), 45% (hepatic encephalopathy), and 30% (serious adverse events). In the analysis of mortality, we conducted the analysis with inclusion of i) all RCTs and ii) RCTs with a low risk of bias (only possible in mortality analyses). We repeated the analyses with the RRR reduced to 20% and with diversity increased by 20% (from 0% to 20% in the analyses of mortality and serious adverse events and from 30% to 50% in the analysis of hepatic encephalopathy).

Subgroup analysis and investigation of heterogeneity

We undertook subgroup analyses to investigate the effect of non‐absorbable disaccharides in RCTs evaluating the prevention or treatment of hepatic encephalopathy. We also evaluated heterogeneity based on stratification of RCTs by:

primary or secondary prevention of hepatic encephalopathy;
overt or minimal hepatic encephalopathy;
acute or chronic hepatic encephalopathy.

Sensitivity analysis

We performed a sensitivity analysis including only RCTs with a low risk of bias (as described above) and worse‐case scenario analysis as described above.

'Summary of findings' tables

We used the GRADE system to evaluate the quality of the evidence for outcomes reported in the review considering the within‐study risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimate, and risk of publication bias (GRADEpro).

Results

Description of studies

We included 38 RCTs in our qualitative analyses (Characteristics of included studies) and excluded 24 studies (Characteristics of excluded studies). We were able to gather data for our quantitative analyses from 34 RCTs.

Results of the search

We identified 1378 potentially relevant references in electronic databases and 10 additional records through manual searches (Figure 1). After removing duplicates and references that were clearly irrelevant, we identified 38 RCTs described in 56 references that fulfilled our inclusion criteria (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014).

Figure 1

Trial flow diagram.

We were unable to obtain outcome data from four RCTs (Elkington 1969; Brown 1971; Rodgers 1973; Shi 1997), and we included the remaining 34 RCTs, all published as full paper articles, in our quantitative analyses (Simmons 1970; Germain 1973; Corazza 1982; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014).

The countries of origin were India (Dhiman 2000; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013), the USA (Elkington 1969; Simmons 1970; Brown 1971; Rodgers 1973; McClain 1984), China (Shi 1997; Xing 2003; Zeng 2003; Wen 2013; Li 1999; Yao 2014), Italy (Corazza 1982; Riggio 1989; Grandi 1991; Riggio 2005), the United Kingdom (Morgan 1987a; Morgan 1987b; Morgan 1989), Spain (Heredia 1987; Heredia 1988), Mexico (Uribe 1987a; Uribe 1987b), Belgium (Horsmans 1997), Egypt (Ziada 2013), France (Germain 1973), Holland (Quero 1997), Pakistan (Raza 2004), Serbia (Jankovic 1996), and Taiwan (Pai 1995).

Included studies

Participants

The total number of participants was 1828. Their mean age ranged from 41 to 67 years and the proportion of men from 11% to 100%. The proportion of participants with cirrhosis secondary to hepatitis B/C infection ranged from 0% to 81%, while the proportion with alcohol‐related cirrhosis ranged from 0% to 100%.

Seven RCTs evaluated the prevention of hepatic encephalopathy. Three RCTs evaluated primary (Sharma 2012), or secondary prevention of hepatic encephalopathy (Sharma 2009; Agrawal 2012), in participants with no obvious risks. Four included participants with an increased risk of hepatic encephalopathy due to gastrointestinal bleeding (Sharma 2011; Wen 2013), recent insertion of a transjugular intrahepatic portosystemic shunt (Riggio 2005), or portosystemic shunt surgery (Riggio 1989). In 16 RCTs, participants had overt hepatic encephalopathy (Table 1) classed as acute (Simmons 1970; Heredia 1987; Morgan 1987a; Uribe 1987a; Pai 1995; Jankovic 1996; Raza 2004), or chronic (Elkington 1969; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; Morgan 1987b; Uribe 1987b; Heredia 1988; Grandi 1991). In 15 RCTs, participants had minimal hepatic encephalopathy (McClain 1984; Morgan 1989; Horsmans 1997; Quero 1997; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Prasad 2007; Mittal 2011; Jain 2013; Ziada 2013; Yao 2014).

See: Summary of findings for the main comparison Non‐absorbable disaccharides versus placebo/no intervention for the prevention and treatment of hepatic encephalopathy in people with cirrhosis; Summary of findings 2 Lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Table 1. Definitions and assessment of overt hepatic encephalopathy with corresponding recommended definitions in the EASL/AASLD guidelines

Trial	Definition in trial publication	Definition based on classification in EASL/AASLD guidelines	Assessment of hepatic encephalopathy
Elkington 1969	Chronic persistent hepatic encephalopathy	Persistent	Mental status assessed using Parsons‐Smith criteria Arterial blood ammonia concentrations Electroencephalogram
Simmons 1970	Acute, acute remittent, and chronic remittent hepatic encephalopathy	Episodic (81%) Recurrent (19%)	Mental status assessed on a scale similar to but more extensive than the West Haven Criteria Venous blood ammonia concentrations
Brown 1971	Chronic persistent hepatic encephalopathy	Persistent	Mental status Blood ammonia concentrations Electroencephalogram*
Germain 1973	Chronic persistent hepatic encephalopathy	Persistent	Mental status assessed using Parson‐Smith criteria Psychometric tests Venous blood ammonia concentrations Electroencephalogram
Rodgers 1973	Chronic persistent hepatic encephalopathy	Persistent	Clinical assessment of mental status Blood ammonia concentrations Electroencephalogram*
Corazza 1982	Chronic persistent hepatic encephalopathy	Persistent	Encephalopathy Intensity Score Plasma ammonia concentrations
Heredia 1987	Acute hepatic encephalopathy	Episodic/recurrent	Conn score Number Connection Test Blood ammonia concentrations Electroencephalogram
Morgan 1987a	Acute hepatic encephalopathy	Episodic	Portal Systemic Encephalopathy Sum and Index
Morgan 1987b	Chronic persistent hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index
Uribe 1987a	Acute hepatic encephalopathy	Episodic	Portal Systemic Encephalopathy Sum and Index
Uribe 1987b	Chronic persistent hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index
Heredia 1988	Chronic recurrent hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index*
Grandi 1991	Chronic hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index modified by omitting the electroencephalogram
Pai 1995	Acute hepatic encephalopathy	Episodic	Portal Systemic Encephalopathy Sum and Index
Jankovic 1996	Acute hepatic encephalopathy	Episodic	Mental status using West Haven criteria Number connection Test A Electroencephalogram*
Raza 2004	Acute hepatic encephalopathy	Episodic	Clinical scoring Modified Portal Systemic Encephalopathy Sum and Index with electroencephalogram omitted and Digit Symbol test replacing Number Connection Test A

*The trial is not included in the analysis of hepatic encephalopathy, because we were unable to extract data on the number of participants with (or without) an overall improvement.

Interventions

Twenty‐nine RCTs assessed non‐absorbable disaccharides versus placebo/no intervention (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; McClain 1984; Uribe 1987a; Uribe 1987b; Horsmans 1997; Quero 1997; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014). Of these, 25 assessed lactulose (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; McClain 1984; Horsmans 1997; Quero 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014), and four assessed lactitol (Uribe 1987a; Uribe 1987b; Shi 1997; Riggio 2005).

Nine RCTs compared lactulose versus lactitol (Heredia 1987; Morgan 1987a; Morgan 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996).

Outcomes

We were unable to extract outcome data from four RCTs with 64 participants (Elkington 1969; Brown 1971; Rodgers 1973; Shi 1997).

In total, our quantitative analyses included 34 RCTs with 1764 participants (Simmons 1970; Germain 1973; Corazza 1982; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Yao 2014).

Thirty‐one RCTs followed participants to the end of the intervention (Simmons 1970; Germain 1973; Corazza 1982; McClain 1984; Heredia 1987; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Horsmans 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014). Three parallel‐arm RCTs followed participants for an additional 13 days (Jankovic 1996), one month (Morgan 1987a), or three months after the end of treatment (Quero 1997). The duration of the intervention depended on the type of hepatic encephalopathy. Overall, the RCTs followed participants for 89 days (range 4 to 360 days) after randomisation. In prevention RCTs, the duration was 207 days (range 5 to 360 days). For participants with overt hepatic encephalopathy, the mean duration was 49 days (range 4 to 360) with a shorter duration in RCTs on acute (mean 5 days; range 4 to 7 days) and chronic hepatic encephalopathy (mean 74 days; range 10 to 360 days). The mean duration was 70 days in RCTs on minimal hepatic encephalopathy (range 14 to 180).

Investigators assessed overt (Table 1) and minimal hepatic encephalopathy using several different neuropsychiatric assessments and variables (Characteristics of included studies). Eight RCTs used the Portal Systemic Encephalopathy Index and Ratio (Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Riggio 1989; Pai 1995; Riggio 2005), which comprises mental status (West Haven Criteria), asterixis, Number Connection Test A results, blood ammonia concentrations, and the electroencephalogram mean cycle frequency. Two RCTs used a modified version of the test without the electroencephalogram (Grandi 1991; Raza 2004), while one additionally replaced Number Connection Test A with the Digit Symbol test (Raza 2004).

Ten of the remaining RCTs also used West Haven Criteria to assess mental status (Jankovic 1996; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013). Three RCTs used the Conn Score, which is similar to the West Haven Criteria (Heredia 1987; Morgan 1989; Watanabe 1997). Thirty‐two RCTs employed the Number Connection Test (Germain 1973; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014). Twenty‐five RCTs measured blood ammonia in plasma, venous, or arterial blood (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Corazza 1982; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Riggio 1989; Grandi 1991; Pai 1995; Quero 1997; Shi 1997; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Mittal 2011; Sharma 2011; Agrawal 2012; Jain 2013; Ziada 2013), and 22 assessed the electroencephalogram mean cycle frequency (Elkington 1969; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Agrawal 2012).

Excluded studies

We excluded four RCTs and 20 observational studies (Characteristics of excluded studies). Three RCTs compared lactulose versus probiotics (Sharma 2008), polyethylene glycol followed by lactulose (Rahimi 2014), or a carbon adsorbent (Pockros 2009), while one RCT compared mannitol lavage versus a combination of lactulose and the antibiotic kanamycin (Quinton 1982). Five case series described the effects of lactulose on minimal (Salerno 1994) or recurrent hepatic encephalopathy (Brown 1970; Rorsman 1970; Zeegen 1970; Bircher 1971). One additional study looked at the differential effects of lactitol and lactulose on chronic hepatic encephalopathy (Lanthier 1985), while another looked at the effect of lactulose in preventing hepatic encephalopathy following insertion of a transjugular intrahepatic portosystemic shunt (Piotraschke 1996). Three studies of participants with cirrhosis described compliance with non‐absorbable disaccharides, the predictors of recurrence of hepatic encephalopathy, and the predictors of response (Bajaj 2010b; Sharma 2009a; Sharma 2010). Three studies describe the prevalence and characteristics of participants with overt or minimal hepatic encephalopathy (Schomerus 1993; Sharma 2010a), or young people admitted with overt hepatic encephalopathy (Sharma 2011a). Six studies describe the effects of non‐absorbable disaccharides on cerebral blood flow and metabolism (James 1971), fat excretion (Merli 1992), terminal ileal and colonic pH (Patil 1987), blood ammonia, atrial natriuretic peptide and amino acid concentrations (Trovato 1995), blood ammonia, Number Connection Test results and lymphocyte subpopulations (Vendemiale 1992), and benzodiazepine‐like compounds (Venturini 2005).

Risk of bias in included studies

We based our bias assessment on the published descriptions combined with additional information from investigators (Figure 2).

Figure 2

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

Allocation

In 26 RCTs, investigators generated the allocation sequence based on a table of random numbers or computer‐generated random numbers (Simmons 1970; Germain 1973; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Pai 1995; Horsmans 1997; Quero 1997; Watanabe 1997; Dhiman 2000; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Yao 2014).

In 28 RCTs, the allocation concealment involved randomisation via a central independent unit, serially numbered, opaque, sealed envelopes, or blinded administration of identically appearing drug containers (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Horsmans 1997; Quero 1997; Shi 1997; Watanabe 1997; Dhiman 2000; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013).

We classified 23 RCTs as having low risk of selection bias (Simmons 1970; Germain 1973; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Horsmans 1997; Quero 1997; Watanabe 1997; Dhiman 2000; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013).

We classified 15 RCTs as having unclear risk of selection bias (Elkington 1969; Brown 1971; Rodgers 1973; Corazza 1982; Grandi 1991; Pai 1995; Jankovic 1996; Shi 1997; Li 1999; Xing 2003; Zeng 2003; Raza 2004; Wen 2013; Ziada 2013; Yao 2014).

Blinding

We classified five single‐blind RCTs with blinded outcome assessment as having low risk of detection bias (Morgan 1989; Riggio 1989; Pai 1995; Riggio 2005; Wen 2013), and 14 double‐blind RCTs as having low risk of performance and detection bias (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; McClain 1984; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Horsmans 1997; Quero 1997; Shi 1997).

The remaining 19 RCTs were open and we classified them as having high risk of performance and detection bias (Heredia 1987; Heredia 1988; Grandi 1991; Jankovic 1996; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Ziada 2013; Yao 2014).

Incomplete outcome data

In 12 trials, the authors described missing outcome data and excluded participants who were dropouts or withdrawals from their analyses (Brown 1971; Rodgers 1973; McClain 1984; Uribe 1987b; Heredia 1988; Pai 1995; Jankovic 1996; Quero 1997; Watanabe 1997; Jain 2013; Wen 2013; Ziada 2013). We classified these RCTs as having high risk of attrition bias and four RCTs as having unclear risk of attrition bias because the trial reports did not describe dropouts or withdrawals or the handling of missing outcome data in the analyses (Elkington 1969; Corazza 1982; Shi 1997; Raza 2004).

The remaining 22 RCTs had no missing outcome data and the analyses included all participants based on the intention‐to‐treat principle using adequate methods including last observation carried forward or multiple imputation (Simmons 1970; Germain 1973; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Morgan 1989; Riggio 1989; Grandi 1991; Horsmans 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Yao 2014). We classified these RCTs as having low risk of attrition bias.

Selective reporting

Thirty‐two RCTs reported predefined, clinically relevant outcome measures suggesting a low risk of selective reporting (Elkington 1969; Simmons 1970; Germain 1973; Corazza 1982; McClain 1984; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Wen 2013; Yao 2014).

One trial reported different primary and secondary outcomes in the electronic trial register (Jain 2013). The remaining five RCTs did not report mortality (Brown 1971; Rodgers 1973; Heredia 1988; Shi 1997; Ziada 2013). We therefore classed these six RCTs as having a high risk of selective reporting.

For‐profit funding

Twenty RCTs did not receive funding or had other involvement with for‐profit companies (Corazza 1982; Heredia 1987; Pai 1995; Jankovic 1996; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013).

In 10 RCTs, investigators received lactitol, lactulose, or placebo from a pharmaceutical company (Simmons 1970; McClain 1984; Morgan 1987a; Morgan 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Horsmans 1997; Raza 2004).

Seven RCTs received financial or other support from a pharmaceutical company (Brown 1971; Elkington 1969; Germain 1973; Quero 1997; Rodgers 1973; Uribe 1987a; Uribe 1987b).

One RCT did not report funding (Yao 2014).

Other potential sources of bias

We found no other potential sources of bias and therefore classified all RCTs as having low risk of bias for this domain (Elkington 1969; Simmons 1970; Brown 1971; Germain 1973; Rodgers 1973; Corazza 1982; McClain 1984; Heredia 1987; Morgan 1987a; Morgan 1987b; Uribe 1987a; Uribe 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996; Horsmans 1997; Quero 1997; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000; Xing 2003; Zeng 2003; Raza 2004; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013; Wen 2013; Ziada 2013; Yao 2014).

Overall bias assessment

We classified eight RCTs as having low risk of bias in the assessment of mortality (Dhiman 2000; Riggio 2005; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012), and none of the RCTs as having low risk of bias in the assessment of the remaining outcomes.

Effects of interventions

Non‐absorbable disaccharides versus placebo/no intervention

Primary outcomes

Our meta‐analysis of mortality included 24 RCTs with 1487 participants (Analysis 1.1). Compared with placebo/no intervention, non‐absorbable disaccharides were associated with a beneficial effect on mortality when including all randomised clinical trials (risk ratio (RR) 0.59, 95% confidence interval (CI) 0.40 to 0.87; I² = 0%) or the eight RCTs with a low risk of bias (RR 0.63, 95% CI 0.41 to 0.97; number needed to treat to benefit (NNTB) 19; Analysis 1.2).

Our meta‐analysis of hepatic encephalopathy included 22 RCTs with 1415 participants (Analysis 1.3) and showed that compared with placebo/no intervention, non‐absorbable disaccharides were associated with a beneficial effect on hepatic encephalopathy (RR 0.58, 95% CI 0.48 to 0.69; I² = 43%; NNTB six participants). Twenty‐four RCTs with 1487 participants reported serious adverse events (Analysis 1.4) that reflected liver‐related morbidity such as liver failure, hepatorenal syndrome, and variceal bleeding (Table 2). Non‐absorbable disaccharides had a beneficial effect on serious adverse events (RR 0.47, 95% CI 0.36 to 0.60; I² = 0%; Analysis 1.4). None of the RCTs evaluating hepatic encephalopathy or serious adverse events had a low risk of bias.

Table 2. Liver‐related serious adverse events

Event	Non‐absorbable disaccharides	Placebo/no intervention
Variceal bleeding	19/438 (4%)	17/336 (5%)
Hepatorenal syndrome	10/196 (5%)	7/153 (5%)
Spontaneous bacterial peritonitis	10/140 (7%)	16/138 (12%)
Liver failure	9/189 (5%)	7/117 (6%)

The overall risk of serious adverse events is analysed as one of the primary outcomes.

We conducted the Trial Sequential Analyses of primary outcomes with the relative risk reduction (RRR) downgraded to 30%. In the analysis of mortality, we set the CGR to 15%. When including all 24 RCTs (Figure 3), the cumulative Z‐curve crossed the monitoring boundary after 1037 participants before reaching the heterogeneity adjusted information size. The cumulative Z‐curve did not cross the monitoring boundary when we reduced the RRR to 20% and increased the diversity to 20%, or when we only included RCTs with a low risk of bias (Figure 4). When we conducted the Trial Sequential Analysis for the outcome hepatic encephalopathy, we initially set the CGR to 45% (Figure 5). The analysis found that the Z‐curve crossed the monitoring boundary before reaching the information size of 581 participants and the analysis was confirmed when we decreased the RRR to 20% (information size 1337 participants) and increased diversity from 30% (model based) to 50% (information size 814 participants). Likewise, when analysing serious adverse events with the CGR set to 30%, the Z‐curve crossed the monitoring boundary before reaching the required information size (737 participants; Figure 6). We confirmed the result in an analysis with RRR of 20% and diversity 20% (information size 1719 participants).

Figure 3

Trial Sequential Analysis of mortality in 24 RCTs evaluating non‐absorbable disaccharides versus placebo/no intervention. The primary meta‐analysis found a RR of 0.59 (95% CI 0.40 to 0.87). When we set the RRR to 30% and CGR to 15%, (power 80%, alpha 5%, and diversity 0%), the cumulative Z‐curve (the green line) crossed the monitoring boundary (inward sloping line) after 1037 participants before reaching the heterogeneity adjusted information size. The cumulative Z‐curve did not cross the monitoring boundary when we increased the diversity to 20% and reduced the RRR to 20%.

Figure 4

Trial Sequential Analysis of mortality in 8 RCTs with a low risk of bias. The RCTs compare non‐absorbable disaccharides versus placebo/no intervention and the primary meta‐analysis found an effect of non‐absorbable disaccharides with a RR of 0.63 (95% CI 0.41 to 0.97). When we set the RRR to 30% and CGR to 45% (power 80%, alpha 5%, and diversity 0%), the cumulative Z‐curve (the green line) did not cross the monitoring boundary (inward sloping line). The heterogeneity adjusted information size was 1725 participants.

Figure 5

Trial Sequential Analysis of hepatic encephalopathy in 22 RCTs evaluating non‐absorbable disaccharides versus placebo/no intervention. A meta‐analysis including all trials found a RR of 0.58 (95% CI 0.48 to 0.69). The analysis includes a RRR of 30% and CGR of 45% (power 80%, alpha 5%, and diversity 30%). The analysis found that the Z‐curve (green line) crossed the monitoring boundary (inward sloping black line) before reaching the information size of 581 participants. None of the RCTs were low risk of bias in the overall assessment. The Z‐curve crossed the monitoring boundary before reaching the information size when we decreased the RRR to 20% (information size 1337 participants) and when we increased diversity to 50% (814 participants).

Figure 6

Trial Sequential Analysis of serious adverse events including 24 RCTs evaluating non‐absorbable disaccharides versus placebo/no intervention. The primary meta‐analysis found a beneficial intervention effect with a RR of 0.47 (95% CI 0.36 to 0.60). None of the included RCTs had a low risk of bias in the overall assessment. When conducting the Trial Sequential Analysis with RRR 30%, CGR 30%, power 80%, alpha 5%, and diversity 0%, the Z‐curve crossed the monitoring boundary before reaching the required information size of 737 participants. The Z‐curve also crossed the monitoring boundary before reaching the required information size when we reduced the RRR to 20% (information size 1719 participants) and when we increased diversity to 20% (information size 921 participants).

Worst‐case scenario analyses (missing outcome data counted as failures) showed that the non‐absorbable disaccharides were associated with a beneficial effect on mortality (RR 0.61, 95% CI 0.42 to 0.88; Analysis 1.10), hepatic encephalopathy (RR 0.59, 95% CI 0.50 to 0.69; Analysis 1.11), and serious adverse events (RR 0.47, 95% CI 0.37 to 0.61; Analysis 1.12). The 'extreme worst‐case scenario' analyses (missing outcome data counted as failures in the non‐absorbable disaccharide group and successes in the control group) reached the same conclusions (Analysis 1.10, Analysis 1.11, and Analysis 1.12).

Regression analyses and funnel plots showed no evidence of small study effects in the analysis of mortality (P value = 0.73), hepatic encephalopathy (P value = 0.93), or serious adverse events (P value = 0.96).

Secondary outcomes

Six RCTs included quality of life assessments (McClain 1984; Quero 1997; Watanabe 1997; Zeng 2003; Prasad 2007; Mittal 2011). Three RCTs, Quero 1997, Prasad 2007 and Mittal 2011, evaluated 160 participants with minimal hepatic encephalopathy using the Sickness Impact Profile (Table 3; Table 4; Table 5), which includes 136 questions about health‐related dysfunction (Gilson 1975; SF 36 questionnaire). The responses to these questions are divided into 12 categories: ambulation, body care/movement, mobility, emotional behaviour, social interaction, alertness behaviour, communication, work, sleep and rest, eating, home management, and recreation/pastimes. These, in turn, are used to inform the two major summative domains physical and psychosocial health. Two RCTs defined the alteration in the total score after treatment as the change in the overall quality of life (Prasad 2007; Mittal 2011). The third trial compared the end of treatment values (Quero 1997). The three RCTs individually found a beneficial effect of lactulose. However, the heterogeneity between RCTs was considerable so we did not conduct a meta‐analysis (Analysis 1.5).

Table 3. Quero 1996: Sickness Impact Profile selected subscores

End of treatment	Control (n = 21)		Lactulose (n = 19)
	Mean	Standard deviation	Mean	Standard deviation
Psychological subscore	8.0	11	10.9	14
Physical subscore	2.8	4	4.8	6

Table 4. Prasad 2007: Sickness Impact Profile selected subscores

Change from baseline	Control (n = 20)		Lactulose (n = 25)
	Mean	Standard deviation	Mean	Standard deviation
Psychosocial scales
Social interactions	0.5	0.68	8.5	1.35
Alertness	‐0.75	1.13	10.43	1.73
Emotional behaviour	2.76	1.83	8.98	1.55
Communication	0.75	1.19	2.66	1.22
Total psychological subscore	0.77	0.41	8.47	0.98
Physical scales
Ambulation	‐1.89	1.12	3.67	0.80
Mobility	1.22	1.18	5.36	1.35
Body care and movements	0.72	0.42	1.62	0.55
Total physical subscore	0.01	0.52	2.99	0.56
Independent scales
Sleep and rest	2.29	1.35	9.04	1.95
Work	‐0.06	1.44	15.83	4.45
Home management	0.94	1.19	12.64	2.71
Recreation and pastimes	‐0.28	1.11	11.59	1.97
Eating	‐0.56	1.31	3.88	1.21

Table 5. Mittal 2009: Sickness Impact Profile selected subscores

Change from baseline	Control (n = 31)		Lactulose (n = 35)
	Mean	Standard deviation	Mean	Standard deviation
Subscores
Sleep and rest	2.87	6.5	11.64	5.5
Emotional behaviour	0.40	4.1	9.84	4.8
Body care and movements	–0.38	1.9	3.20	2.4
Home management	–0.25	5.7	6.34	5.20
Mobility	0.59	5.5	4.64	4.3
Social interaction	1.63	3.2	3.88	2.8
Alertness	0.18	2.4	3.63	2.2
Ambulation	–0.18	2.9	5.10	4.2
Communication	0.80	3.3	2.07	5.1
Work	0.64	2.5	9.46	15.7
Recreation and pastime	3.06	4.4	7.74	5.7
Eating	1.12	3.1	2.48	3.1
Psychosocial	1.13	2.4	5.17	2.9
Physical	–0.05	2.0	3.59	2.1

One trial, Zeng 2003, used an abbreviated version of the World Health Organization quality of life 100 questionnaire (WHOQOL 1998), which evaluates the domains: physical health, psychological health, social relationships, and environment. The trial report includes a table showing a selection of subscores from the questionnaire (Table 6). The analyses showed that lactulose improved the domains of physical and psychological health, and social relationships (P value < 0.05 for all subscores).

Table 6. Zeng 2003: WHO‐Bref selected subscores

End of treatment	Control (n = 20)		Short term lactulose (n = 20)		Long‐term lactulose (n = 20)
	Mean	Standard deviation	Mean	Standard deviation	Mean	Standard deviation
Physical health	28	19	37	18	54	19
Psychological health	42	14	44	15	58	15
Social relationships	38	16	42	15	60	17
Environment	51	18	53	15	51	13

One trial described the effect of lactulose on the quality of life without specifying the assessment method (Watanabe 1997). The abstract states that lactulose improved the quality of life without providing quantitative data. One further trial, McClain 1984, assessed quality of life using the Katz functioning scale (Katz 1963), which evaluates the adjustment and social behaviour in the community. The investigators state that there were no differences between the intervention groups before or after treatment, but do not provide quantitative data.

The non‐absorbable disaccharides increased the risk of gastrointestinal non‐serious adverse events (RR 2.47, 95% CI 1.24 to 4.93; 739 participants; nine RCTs; I² = 64%; Analysis 1.6), including diarrhoea, bloating, flatulence, and nausea. Participants allocated to placebo/no intervention had a higher risk of constipation.

The surrogate outcomes included Number Connection Test results (mean difference (MD) ‐5.56, 95% CI ‐11.59 to 0.47; Analysis 1.7) and blood ammonia concentrations assessed at the end of the trials (MD ‐11.64, 95% CI ‐21.14 to ‐2.14; Analysis 1.8) and as the change from baseline to the end of follow‐up (MD 18.97, 95% CI 8.86 to 29.09; Analysis 1.9). The analyses included a small number of participants and considerable heterogeneity.

Prevention RCTs

The meta‐analysis evaluating primary or secondary prevention showed a beneficial effect on mortality when including all six RCTs (RR 0.63, 95% CI 0.40 to 0.98; 668 participants; Analysis 2.1), or the five RCTs with a low risk of bias (RR 0.64, 95% CI 0.41 to 0.99; 538 participants; Analysis 2.2). The non‐absorbable disaccharides also had beneficial effects on the prevention of hepatic encephalopathy (RR 0.47, 95% CI 0.33 to 0.68; Analysis 2.3), and serious adverse events (RR 0.48, 95% CI 0.33 to 0.70, Analysis 2.4). Additional analyses including four RCTs showed that non‐absorbable disaccharides increased the risk of non‐serious adverse events (RR 2.78, 95% CI 1.50 to 5.13; 548 participants; Analysis 2.5).

Treatment RCTs

The meta‐analysis evaluating the treatment of overt or minimal hepatic encephalopathy showed no effect of non‐absorbable disaccharides on mortality when including all 18 RCTs (RR 0.49, 95% CI 0.23 to 1.05; 819 participants; Analysis 3.1), or the three RCTs with a low risk of bias (RR 0.56, 95% CI 0.12 to 2.68; 167 participants; three RCTs; Analysis 3.2). The analyses showed beneficial effect of non‐absorbable disaccharides on mortality in RCTs evaluating acute, overt hepatic encephalopathy (RR 0.36, 95% CI 0.14 to 0.94; 172 participants; six RCTs), but not in RCTs evaluating minimal hepatic encephalopathy (RR 0.82, 95% CI 0.24 to 2.86; 647 participants; 12 RCTs). No events occurred in RCTs evaluating chronic hepatic encephalopathy (Analysis 3.3).

The non‐absorbable disaccharides had beneficial effects on overt and minimal hepatic encephalopathy (RR 0.63, 95% CI 0.53 to 0.74; 747 participants; 16 RCTs; Analysis 3.4). The effect was similar in RCTs evaluating acute or chronic hepatic encephalopathy (Analysis 3.5). Non‐absorbable disaccharides had a beneficial effect on serious adverse events (RR 0.42, 95% CI 0.26 to 0.69; 819 participants; 18 RCTs; Analysis 3.6) with no difference between the acute and chronic hepatic encephalopathy subgroups (Analysis 3.7). Non‐absorbable disaccharides did not increase the risk of non‐serious adverse events (RR 2.12, 95% CI 0.62 to 7.28; 191 participants; five RCTs; Analysis 3.7).

Lactulose versus lactitol

Meta‐analyses showed no difference between lactulose versus lactitol in the assessment of mortality (RR 1.30, 95% CI 0.59 to 2.85; 225 participants; eight RCTs; I² = 0%; Analysis 4.1), hepatic encephalopathy (RR 1.00, 95% CI 0.84 to 1.19; Analysis 4.2), or serious adverse events (RR 1.56, 95% CI 0.84 to 2.88; Analysis 4.3). All Trial Sequential Analyses ignored the monitoring boundaries because the information size was insufficient. None of the RCTs assessed the quality of life. The non‐serious adverse events were mainly gastrointestinal (RR 1.55, 95% CI 0.88 to 2.74; Analysis 4.4). We found no differences between interventions for the surrogate outcomes Number Connection Test (end of treatment Analysis 4.5 or change from baseline Analysis 4.6), or blood ammonia concentrations (end of treatment Analysis 4.7 or change from baseline Analysis 4.8). We found no differences between subgroups for any outcomes. We only found evidence of missing outcome data in two RCTs (Pai 1995; Jankovic 1996). The trials did not provide information about the number of participants in the two groups (lactulose or lactitol) with missing outcome data. Therefore, we were unable to conduct worst‐case scenario or extreme worst‐case scenario analyses.

'Summary of findings' tables

In the analyses comparing non‐absorbable disaccharides versus placebo/no intervention (Summary of findings table 1), we downgraded the quality of the evidence to 'moderate' for the outcome mortality because the Trial Sequential Analysis of RCTs with a low risk of bias found no evidence to support or refute an intervention effect. Likewise, we downgraded the quality of evidence for the outcomes hepatic encephalopathy and serious adverse events one level to 'moderate' because none of the included RCTs had a low risk of bias. We downgraded the outcome quality of life three levels to 'very low quality evidence' because none of the included RCTs had a low risk of bias, the heterogeneity was considerable, and we were unable to combine the data in an overall analysis. We also downgraded the outcome non‐serious adverse events three levels to 'very low quality evidence' because none of the included RCTs had a low risk of bias, the confidence intervals were wide, and we were only able to include data from nine RCTs in our meta‐analysis.

In the analyses comparing lactulose versus lactitol (Summary of findings table 2), we downgraded the evidence three levels to 'very low quality' due to imprecision, uncertainty, and a methodological quality (none of the included RCTs had a low risk of bias).

Discussion

Summary of main results

This review includes descriptive information from 38 randomised clinical trials (RCTs) with 1828 participants and quantitative data from 34 RCTs with 1764 participants. The primary analyses show that use of the non‐absorbable disaccharides, lactulose and lactitol, is associated with reduced mortality compared with placebo/no intervention when including all RCTs and when including the RCTs with a low risk of bias. In subgroup analyses, we found no statistical differences between RCTs stratified by the type of hepatic encephalopathy. We found a beneficial effect on mortality in RCTs evaluating prevention and RCTs evaluating acute hepatic encephalopathy, but not in RCTs evaluating chronic or minimal hepatic encephalopathy (where the mortality rates overall were extremely low). The quality of the evidence was moderate.

Use of non‐absorbable disaccharides is associated with a beneficial effect on the prevention and treatment of hepatic encephalopathy (moderate quality evidence). Additional analyses showed that non‐absorbable disaccharides can help to reduce serious adverse events associated with the underlying liver disease including liver failure, variceal bleeding, and hepatorenal syndrome (moderate quality evidence). Six RCTs suggested a beneficial effect on quality of life, but we were unable to combine the results in a meta‐analysis (very low quality evidence). As expected, the non‐absorbable disaccharides increased the risk of non‐serious gastrointestinal adverse events (very low quality evidence). None of the RCTs comparing lactulose versus lactitol assessed quality of life. Analyses of the remaining outcomes found no differences between the two interventions (very low quality evidence).

Overall completeness and applicability of evidence

The most important outcomes for people with cirrhosis and hepatic encephalopathy are mortality, morbidity, adverse events, and quality of life (Bajaj 2011a). We included information on all of these outcomes. The RCTs evaluated improvement in hepatic encephalopathy using a variety of methods. This partly reflects that fact that the included RCTs were conducted between 1969 and 2014 during which time diagnostic criteria changed on more than one occasion. The included RCTs often used clinical or composite scoring systems and a categorical approach to define improvement (or lack thereof). The investigators did not use the same thresholds to define improvement, so we chose to use the definitions that they defined as clinically relevant. The diagnostic classification of hepatic encephalopathy also changed during the time period (EASL and AASLD guideline 2014a; EASL and AASLD guideline 2014b). Thus, we made a decision a priori to utilise the individual primary investigators' classification of the type of hepatic encephalopathy and the outcome criteria for hepatic encephalopathy, based on the argument that these decisions will have been made using the criteria that were most clinically relevant when the investigators conducted the trial.

The older RCTs often used co–interventions such as dietary protein restriction. Although the RCTs did not use the co‐interventions consistently, participants randomised to experimental or control groups within a given RCT would have had equal access to them. This might result in heterogeneity, but not in systematic differences between groups.

Hepatic encephalopathy varies widely in its manifestations. The RCTs included in our review represent the entire spectrum of the syndrome encountered in people with cirrhosis. Thus, RCTs included people experiencing an acute episode of hepatic encephalopathy, chronic hepatic encephalopathy associated with advanced liver disease, spontaneous or surgically created portal‐systemic shunts, and minimal hepatic encephalopathy. In addition, the included RCTs explored the use of non‐absorbable disaccharides for primary and secondary prevention of hepatic encephalopathy. The fact that the RCTs address all the objectives of the review strengthens the completeness of the evidence. We included all RCTs with extractable data in our primary analyses. We also conducted subgroup, sensitivity, and regression analyses to determine the differential effects of intervention on the clinical variants. Our analyses showed that non‐absorbable disaccharides are associated with stable beneficial effects on clinically important outcomes across the different groups. This supports the external validity of our findings.

This review includes the two commercially available disaccharides, lactulose and lactitol. However, only four of the 28 RCTs of non‐absorbable disaccharides versus placebo/no treatment utilised lactitol (Uribe 1987a; Uribe 1987b; Shi 1997; Riggio 2005). Nine RCTs with a total of 248 participants compared lactulose versus lactitol (Heredia 1987; Morgan 1987a; Morgan 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995; Jankovic 1996). We found no differences between the two interventions, but the statistical power was insufficient.

People with non‐cirrhotic portal hypertension and those with fulminant hepatic failure may also develop hepatic encephalopathy. They are encountered much less frequently in clinical practice and were not represented in the included trials. There is no reason to suppose that our results cannot be extrapolated to people with hepatic encephalopathy associated with non‐cirrhotic portal hypertension, e.g. portal vein block. However, the situation in people with fulminant hepatic failure is much more complex and the result may not be directly applicable.

Episodes of hepatic encephalopathy often develop in response to a precipitating event such as infection, gastrointestinal bleeding, alcohol misuse, or electrolyte disturbances. Identification and treatment of these precipitating factors is key to the management of affected individuals although no obvious precipitating factor is identified in 50% of instances (EASL and AASLD guideline 2014a; EASL and AASLD guideline 2014b). Avoiding likely precipitants such as constipation, dietary indiscretion, and certain medications can also reduce the risk of developing hepatic encephalopathy in the longer term. It is not clear whether use of non‐absorbable disaccharides provides additional benefit in situations where hepatic encephalopathy is precipitated by a treatable event. The RCTs included in our review do not provide detailed information on possible precipitating events, on the effects of interventions designed to ameliorate them, or on the effects, if any, of the addition of a non‐absorbable disaccharide. However, in two of the included RCTs, non‐absorbable disaccharides, used together with measures to manage upper gastrointestinal haemorrhage, prevented the development of hepatic encephalopathy (Sharma 2012; Wen 2013).

Non‐adherence to non‐absorbable disaccharides is generally ascribed to adverse gastrointestinal effects such as unpredictable diarrhoea, bloating, flatulence, and abdominal pain (Bajaj 2010c; Volk 2012). Although we did find that treatment with lactulose or lactitol was associated with a higher risk of these non‐serious adverse events, none of the RCTs included in our review evaluated compliance in a manner that allowed us to assess the potential influence of these gastrointestinal effects. Other factors may, however, be important in determining compliance with treatment both on the part of the person receiving treatment and the physician prescribing it. Thus, people with hepatic encephalopathy may be unaware of the need for long‐term treatment, may be unable to effectively titrate the dosage, and may find the side effects inconvenient especially when away from home. The physician may fail to explain the multiple ways in which non‐absorbable disaccharides produce their beneficial effects and by placing undue focus on the need for them to pass two semi‐soft stools/day may foster the belief that as long as this is achieved, there is no real need to take the medication. They may also erroneously assume that people will comply with treatment and hence fail to check adherence.

Hepatic encephalopathy imposes a significant burden on healthcare systems and the resource utilisation associated with the management of people with hepatic encephalopathy is increasing (Poordad 2007). The increased costs do not seem to reflect the duration of hospitalisation, which has decreased, but a combination of direct and indirect factors such as the costs of treatment and rehabilitation after hospitalisation (Neff 2010). None of the RCTs included in the present review assessed the costs associated with hospitalisation, but we found a clear beneficial effect of non‐absorbable disaccharides in preventing the development and recurrence of hepatic encephalopathy that would generally require hospitalisation. Use of non‐absorbable disaccharides is also associated with a reduction in the occurrence of serious liver‐related complications. This will also result in reduced hospitalisations and lengths of hospital stay.

Quality of the evidence

The previous version of this review identified several potential biases in included RCTs (Als‐Nielsen 2004). In this updated review, we identified a larger number of RCTs and additional information on essential aspects of bias control. As recommended, we combined the individual bias domains in an overall assessment (Gluud 2015). We also included an assessment of individual domains, focusing on RCTs with a low risk of selection bias (Higgins 2011a; Higgins 2011b; Savovic 2012). Based on previous evidence (Savovic 2012), we defined mortality, but not serious adverse events, as an outcome that is robust to performance and detection bias. This decision can be questioned as lack of blinding is not likely to influence the assessment of events such as variceal bleeding, hepatorenal syndrome, and liver failure. We included 14 double‐blind RCTs and cannot exclude the possibility that our analyses overestimate the effect of non‐absorbable disaccharides on hepatic encephalopathy due to lack of blinding. In contrast to the previous version of this review, we included any type of for‐profit funding as a bias domain (Gluud 2015). The decision to include this domain is debatable (Higgins 2011a; Higgins 2011b). The fact that we included gratuitous supply of interventions or placebo was the main reason why we did not identify RCTs comparing lactulose versus lactitol with a low risk of bias in the overall assessment. Based on the revised assessment of bias control combined with the assessment of the directness of evidence, heterogeneity, precision of effect estimate, and risk of publication bias we classified the quality of the evidence as moderate for the assessment of our primary outcomes mortality, hepatic encephalopathy, and serious adverse events.

The included RCTs were conducted world‐wide. The country/continent of origin included India/Pakistan (Dhiman 2000; Raza 2004; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013), the USA (Elkington 1969; Simmons 1970; Brown 1971; Rodgers 1973; McClain 1984), the Far‐East (Pai 1995; Shi 1997; Li 1999; Xing 2003; Zeng 2003; Wen 2013), Europe (Germain 1973; Corazza 1982; Heredia 1987; Morgan 1987a; Morgan 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Jankovic 1996; Horsmans 1997; Quero 1997; Riggio 2005), Mexico (Uribe 1987a; Uribe 1987b), and Egypt (Ziada 2013). A single centre in India conducted eight of the RCTs (Dhiman 2000; Prasad 2007; Sharma 2009; Mittal 2011; Sharma 2011; Agrawal 2012; Sharma 2012; Jain 2013). Four of these RCTs involved participants with minimal hepatic encephalopathy (Dhiman 2000; Prasad 2007; Mittal 2011; Jain 2013), and four evaluated primary and secondary prophylaxis (Sharma 2009; Sharma 2011; Agrawal 2012; Sharma 2012). The results of the RCTs evaluating minimal hepatic encephalopathy did not differ substantially from those in the similar RCTs undertaken in centres outside of India. We found no comparable prevention studies undertaken outside of India. Two prevention RCTs conducted in Italy looked at the effects of non‐absorbable disaccharides following transjugular intrahepatic portosystemic shunt insertion (Riggio 1989; Riggio 2005). The RCTs found no benefit on mortality, hepatic encephalopathy, or serious adverse events. However, this is a notoriously difficult situation to manage and one that depends more on careful pre‐selection of candidates than on post‐hoc exhibition of pharmacotherapy. One RCT conducted in China looked at the effect of lactulose in the prevention of hepatic encephalopathy following an acute upper gastrointestinal bleed and observed significant benefit (Wen 2013). We observed clinical variation in participant demographics between the prevention RCTs conducted in India and those conducted elsewhere, but variables such as age, gender, and the aetiology of the cirrhosis did not confound the results. RCTs evaluating the effects of non‐absorbable disaccharides for primary and secondary prevention conducted in countries outside of India would strengthen the external validity of our findings.

Potential biases in the review process

A recent methodological review drew attention to outcome reporting bias in systematic reviews (Page 2014). Changes between the outcomes in protocols and published systematic reviews include the statistical significance of the results for those outcomes. We updated this review to incorporate current recommendations (Higgins 2011a; Higgins 2011b; Gluud 2015). The methods used in this update differ from those in the previous version (Als‐Nielsen 2004a; Als‐Nielsen 2004b; Als‐Nielsen 2005). As part of the update, we changed the definition of our primary outcomes to provide information on benefits as well as harms. Accordingly, we now include serious adverse events as a primary rather than a secondary outcome measure.

The selective publication of RCTs with a positive result increases the risk of outcome reporting bias (Dwan 2008). The RCTs included in the present review were all published as full paper articles and this might be interpreted as a potential publication bias. However, we combined our electronic searches with extensive manual searches of reference lists and conference proceedings. We identified a large number of abstracts, but all were published subsequently as full papers. We found no evidence of publication bias or other small study effects and very few RCTs showed evidence of outcome reporting bias. Of the 29 RCTs on non‐absorbable disaccharides versus placebo or no intervention, we were unable to include data for primary outcomes from four RCTs with 64 participants (Elkington 1969; Brown 1971; Rodgers 1973; Shi 1997). The RCTs are small and the narrative information in the published reports suggested that the intervention had a beneficial effect on hepatic encephalopathy. Exclusion of these four RCTs is unlikely to change our conclusions.

Agreements and disagreements with other studies or reviews

The previous version of this review assessed the effect of non‐absorbable disaccharides versus placebo/no intervention and lactulose versus lactitol based on a total of 19 RCTs (Als‐Nielsen 2004). Eleven RCTs compared lactulose or lactitol versus placebo/no intervention (Elkington 1969; Simmons 1970; Germain 1973; Rodgers 1973; Corazza 1982; Uribe 1987a; Uribe 1987b; Shi 1997; Watanabe 1997; Li 1999; Dhiman 2000), and eight RCTs compared lactulose versus lactitol (Heredia 1987; Morgan 1987a; Morgan 1987b; Heredia 1988; Morgan 1989; Riggio 1989; Grandi 1991; Pai 1995). Based on a meta‐analyses including four RCTs with 85 participants, the review found no effect of non‐absorbable disaccharides on mortality compared with placebo/no intervention (Simmons 1970; Germain 1973; Uribe 1987a; Dhiman 2000). A meta‐analysis including six RCTs with 207 participants showed a beneficial effect on hepatic encephalopathy (Simmons 1970; Germain 1973; Uribe 1987a; Watanabe 1997; Li 1999; Dhiman 2000), but the effect was not confirmed in an analysis that only included RCTs with a low risk of bias. We included 38 RCTs (1828 participants) in our qualitative evaluation and 34 RCTs in our qualitative analyses. Our analyses include several different groups of participants from several countries. In spite of the clinical differences, our analyses showed negligible or moderate statistical heterogeneity. Our findings disagree with previous evidence, mainly because previous reviews included fewer RCTs.

The joint guidelines from the European and American Associations for the Study of the Liver made four recommendations of relevance to this review (EASL and AASLD guideline 2014a; EASL and AASLD guideline 2014b). First, that lactulose should be the first‐choice treatment for an acute episode of overt hepatic encephalopathy in people with cirrhosis. Second, that lactulose should be used for prevention of recurrent episodes of hepatic encephalopathy after the initial episode. Third, that minimal hepatic encephalopathy should not be treated routinely. Fourth, that primary prophylaxis for prevention of the development of hepatic encephalopathy is not required in people with cirrhosis except if they are known to be at high risk.

In agreement with the guideline recommendations, we found a beneficial effect of non‐absorbable disaccharides on clinical outcomes in RCTs evaluating secondary prevention and treatment. The guidelines do not recommend routine treatment of minimal hepatic encephalopathy or primary prevention of hepatic encephalopathy. Our analyses provide a large body of evidence showing that people with minimal hepatic encephalopathy benefit from non‐absorbable disaccharides in relation to cognitive functioning and probably quality of life, and some evidence that non‐absorbable disaccharides may be considered in primary prevention.

Figure 1

Trial flow diagram.

Figure 2

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

Figure 3

Figure 4

Figure 5

Figure 6

Analysis 1.1

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 1 Mortality.

Analysis 1.2

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 2 Mortality in trials with a low risk of bias.

Analysis 1.3

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 3 Hepatic encephalopathy.

Analysis 1.4

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 4 Serious adverse events.

Analysis 1.5

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 5 Quality of life: sickness impact profile.

Analysis 1.6

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 6 Non‐serious adverse events.

Analysis 1.7

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 7 Number connection test, end of treatment.

Analysis 1.8

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 8 Ammonia end of treatment.

Analysis 1.9

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 9 Ammonia change from baseline.

Analysis 1.10

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 10 Mortality in worst‐case scenario analyses.

Analysis 1.11

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 11 Hepatic encephalopathy worst‐case scenario analysis.

Analysis 1.12

Comparison 1 Non‐absorbable disaccharides versus placebo/no intervention, Outcome 12 Serious adverse events worst‐case scenario analysis.

Analysis 2.1

Comparison 2 Prevention trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 1 Mortality.

Analysis 2.2

Comparison 2 Prevention trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 2 Mortality and bias control.

Analysis 2.3

Comparison 2 Prevention trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 3 Hepatic encephalopathy.

Analysis 2.4

Comparison 2 Prevention trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 4 Serious adverse events.

Analysis 2.5

Comparison 2 Prevention trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 5 Non‐serious adverse events.

Analysis 3.1

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 1 Mortality.

Analysis 3.2

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 2 Mortality in trials with a low risk of bias.

Analysis 3.3

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 3 Mortality in acute or chronic hepatic encephalopathy.

Analysis 3.4

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 4 Hepatic encephalopathy.

Analysis 3.5

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 5 Acute or chronic hepatic encephalopathy.

Analysis 3.6

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 6 Serious adverse events.

Analysis 3.7

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 7 Serious adverse events in acute or chronic hepatic encephalopathy.

Analysis 3.8

Comparison 3 Treatment trials: non‐absorbable disaccharides versus placebo/no intervention, Outcome 8 Non‐serious adverse events.

Analysis 4.1

Comparison 4 Lactulose versus lactitol, Outcome 1 Mortality.

Analysis 4.2

Comparison 4 Lactulose versus lactitol, Outcome 2 Hepatic encephalopathy.

Analysis 4.3

Comparison 4 Lactulose versus lactitol, Outcome 3 Serious adverse events.

Analysis 4.4

Comparison 4 Lactulose versus lactitol, Outcome 4 Non‐serious adverse events.

Analysis 4.5

Comparison 4 Lactulose versus lactitol, Outcome 5 Number Connection Test: end of treatment.

Analysis 4.6

Comparison 4 Lactulose versus lactitol, Outcome 6 Number Connection Test: change from baseline.

Analysis 4.7

Comparison 4 Lactulose versus lactitol, Outcome 7 Venous blood ammonia: end of treatment.

Analysis 4.8

Comparison 4 Lactulose versus lactitol, Outcome 8 Venous blood ammonia: change from baseline.

Summary of findings for the main comparison. Non‐absorbable disaccharides versus placebo/no intervention for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Non‐absorbable disaccharides versus placebo/no intervention for the prevention and treatment of hepatic encephalopathy in people with cirrhosis
Population: prevention and treatment of hepatic encephalopathy in people with cirrhosis Intervention: non‐absorbable disaccharides (lactulose and lactitol) Control: placebo/no intervention Setting: in‐hospital (overt hepatic encephalopathy) and outpatient (minimal hepatic encephalopathy and prevention trials) Duration of follow‐up: the duration depended on the type of encephalopathy with 5 days for acute, 74 days for chronic, 70 days for minimal, and 207 days for prevention of hepatic encephalopathy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Non‐absorbable disaccharides versus placebo/no intervention
Mortality	Study population		RR 0.59 (0.40 to 0.87) when including all RCTs; RR 0.63 (0.41 to 0.97) when including RCTs with a low risk of bias	1487 (24 studies)	⊕⊕⊕⊝ moderate¹	Trial Sequential Analysis: The Trial Sequential Analysis found a beneficial effect of the intervention including all RCTs, but when the analysis only included RCTs with a low risk of bias. Assessment method: Assessed based on the total number of participants who died.
	88 per 1000	49 per 1000 (32 to 75)
	Moderate
	20 per 1000	11 per 1000 (7 to 17)
Hepatic encephalopathy	Study population		RR 0.58 (0.5 to 0.69)	1415 (22 studies)	⊕⊕⊕⊝ moderate²	Trial Sequential Analysis: The Trial Sequential Analysis found a beneficial effect of the intervention including all RCTs, but when the analysis only included RCTs with a low risk of bias. Assessment method: Assessed based on the definitions in included RCTs (number of participants without a clinically relevant improvement of hepatic encephalopathy).
	469 per 1000	272 per 1000 (234 to 323)
	Moderate
	423 per 1000	245 per 1000 (211 to 292)
Serious adverse events	Study population		RR 0.47 (0.36 to 0.6)	1487 (24 studies)	⊕⊕⊕⊝ moderate³	Trial Sequential Analysis: The Trial Sequential Analysis found a beneficial effect of the intervention including all RCTs, but when the analysis only included RCTs with a low risk of bias. Assessment method: Assessed and defined as any untoward medical occurrence that led to death, was life threatening, or required hospitalisation or prolongation of hospitalisation (ICH‐GCP 2007).
	207 per 1000	97 per 1000 (75 to 124)
	Moderate
	142 per 1000	67 per 1000 (51 to 85)
Quality of life (secondary outcome)		No overall estimate available			⊕⊝⊝⊝ very low⁴	We were unable to combine the data into an overall analysis due to unacceptably high heterogeneity. Assessment method: Based on the quality of life questionnaires.
Non‐serious adverse events (secondary outcome)	Study population		RR 2.47 (1.24 to 4.93)	739 (9 studies)	⊕⊝⊝⊝ very low⁵	Assessment method: The outcome includes all adverse events that do not fulfil the criteria for 'serious' (ICH‐GCP 2007).
	106 per 1000	261 per 1000 (131 to 521)
	Moderate
	63 per 1000	156 per 1000 (78 to 311)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; RCT: randomised clinical trial; RR: risk ratio
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹Mortality is downgraded one level to 'moderate quality evidence' because the Trial Sequential Analysis found insufficient evidence when we limited the analysis to include only RCTs with a low risk of bias. ²Hepatic encephalopathy is downgraded one level to 'moderate quality evidence' because none of the RCTs had a low risk of bias in the overall assessment. ³Serious adverse events is downgraded one level to 'moderate quality evidence' because none of the RCTs had a low risk of bias in the overall assessment. ⁴Quality of life is downgraded three levels to 'very low quality evidence' because i) none of the included RCTs had a low risk of bias, ii) the heterogeneity was considerable, and iii) we were unable to combine the data in an overall analysis. ⁵Non‐serious adverse events is downgraded three levels to 'very low quality evidence' because i) none of the included RCTs had a low risk of bias, ii) the confidence intervals were wide (uncertainty), and iii) we were only able to include data from nine RCTs in our meta‐analysis.

Summary of findings for the main comparison. Non‐absorbable disaccharides versus placebo/no intervention for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Summary of findings 2. Lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis
Population: prevention and treatment of hepatic encephalopathy in people with cirrhosis Intervention: lactulose Control: lactitol Setting: in‐hospital (overt hepatic encephalopathy) and outpatient (minimal hepatic encephalopathy and prevention trials) Duration of follow‐up: the duration depended on the type of encephalopathy with 5 days for acute, 74 days for chronic, 70 days for minimal, and 207 days for prevention of hepatic encephalopathy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Lactulose versus lactitol
Mortality	Study population		RR 1.3 (0.59 to 2.85)	225 (8 studies)	⊕⊝⊝⊝ very low¹	Trial Sequential Analysis: The Trial Sequential Analysis found no evidence to support or refute a difference between the 2 interventions being compared. Assessment method: Assessed based on the total number of participants who died.
	71 per 1000	92 per 1000 (42 to 202)
	Moderate
	0 per 1000	0 per 1000 (0 to 0)
Hepatic encephalopathy	Study population		RR 1 (0.84 to 1.19)	194 (7 studies)	⊕⊝⊝⊝ very low¹	Trial Sequential Analysis: The Trial Sequential Analysis found no evidence to support or refute a difference between the 2 interventions being compared. Assessment method: Assessed based on the definitions in included RCTs (number of participants without a clinically relevant improvement of hepatic encephalopathy).
	286 per 1000	286 per 1000 (240 to 340)
	Moderate
	200 per 1000	200 per 1000 (168 to 238)
Serious adverse events	Study population		RR 1.56 (0.84 to 2.88)	245 (9 studies)	⊕⊝⊝⊝ very low¹	Trial Sequential Analysis: The Trial Sequential Analysis found no evidence to support or refute a difference between the 2 interventions being compared. Assessment method: Assessed based on the definitions in included RCTs (number of participants without a clinically relevant improvement of hepatic encephalopathy.
	106 per 1000	165 per 1000 (89 to 304)
	Moderate
	77 per 1000	120 per 1000 (65 to 222)
Quality of life (secondary outcome)	—	No data were available for this outcome	—	—	—	None of the included RCTs assessed quality of life.
Non‐serious adverse events (secondary outcome)	Study population		RR 1.55 (0.88 to 2.74)	169 (6 studies)	⊕⊝⊝⊝ very low²	Assessment method: The outcome includes all adverse events that do not fulfil the criteria for 'serious' (ICH‐GCP 2007).
	247 per 1000	383 per 1000 (217 to 677)
	Moderate
	246 per 1000	381 per 1000 (216 to 674)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; RCT: randomised clinical trial; RR: risk ratio
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹Mortality, hepatic encephalopathy, and serious adverse events are downgraded three levels to 'very low quality evidence' because i) the Trial Sequential Analysis found insufficient evidence to support or refute a difference between the intervention and control group, ii) the confidence intervals were wide, and ii) none of the included RCTs had a low risk of bias in the overall assessment of bias control. ²Non‐serious adverse events is downgraded three levels to 'very low quality evidence' because i) none of the included RCTs had a low risk of bias in the overall assessment of bias control, ii) only six RCTs reported the outcome, and iii) the confidence intervals were wide (uncertainty).

Summary of findings 2. Lactulose versus lactitol for the prevention and treatment of hepatic encephalopathy in people with cirrhosis

Table 1. Definitions and assessment of overt hepatic encephalopathy with corresponding recommended definitions in the EASL/AASLD guidelines

Trial	Definition in trial publication	Definition based on classification in EASL/AASLD guidelines	Assessment of hepatic encephalopathy
Elkington 1969	Chronic persistent hepatic encephalopathy	Persistent	Mental status assessed using Parsons‐Smith criteria Arterial blood ammonia concentrations Electroencephalogram
Simmons 1970	Acute, acute remittent, and chronic remittent hepatic encephalopathy	Episodic (81%) Recurrent (19%)	Mental status assessed on a scale similar to but more extensive than the West Haven Criteria Venous blood ammonia concentrations
Brown 1971	Chronic persistent hepatic encephalopathy	Persistent	Mental status Blood ammonia concentrations Electroencephalogram*
Germain 1973	Chronic persistent hepatic encephalopathy	Persistent	Mental status assessed using Parson‐Smith criteria Psychometric tests Venous blood ammonia concentrations Electroencephalogram
Rodgers 1973	Chronic persistent hepatic encephalopathy	Persistent	Clinical assessment of mental status Blood ammonia concentrations Electroencephalogram*
Corazza 1982	Chronic persistent hepatic encephalopathy	Persistent	Encephalopathy Intensity Score Plasma ammonia concentrations
Heredia 1987	Acute hepatic encephalopathy	Episodic/recurrent	Conn score Number Connection Test Blood ammonia concentrations Electroencephalogram
Morgan 1987a	Acute hepatic encephalopathy	Episodic	Portal Systemic Encephalopathy Sum and Index
Morgan 1987b	Chronic persistent hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index
Uribe 1987a	Acute hepatic encephalopathy	Episodic	Portal Systemic Encephalopathy Sum and Index
Uribe 1987b	Chronic persistent hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index
Heredia 1988	Chronic recurrent hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index*
Grandi 1991	Chronic hepatic encephalopathy	Persistent	Portal Systemic Encephalopathy Sum and Index modified by omitting the electroencephalogram
Pai 1995	Acute hepatic encephalopathy	Episodic	Portal Systemic Encephalopathy Sum and Index
Jankovic 1996	Acute hepatic encephalopathy	Episodic	Mental status using West Haven criteria Number connection Test A Electroencephalogram*
Raza 2004	Acute hepatic encephalopathy	Episodic	Clinical scoring Modified Portal Systemic Encephalopathy Sum and Index with electroencephalogram omitted and Digit Symbol test replacing Number Connection Test A
*The trial is not included in the analysis of hepatic encephalopathy, because we were unable to extract data on the number of participants with (or without) an overall improvement.

Table 1. Definitions and assessment of overt hepatic encephalopathy with corresponding recommended definitions in the EASL/AASLD guidelines

Table 2. Liver‐related serious adverse events

Event	Non‐absorbable disaccharides	Placebo/no intervention
Variceal bleeding	19/438 (4%)	17/336 (5%)
Hepatorenal syndrome	10/196 (5%)	7/153 (5%)
Spontaneous bacterial peritonitis	10/140 (7%)	16/138 (12%)
Liver failure	9/189 (5%)	7/117 (6%)
The overall risk of serious adverse events is analysed as one of the primary outcomes.

Table 2. Liver‐related serious adverse events

Table 3. Quero 1996: Sickness Impact Profile selected subscores

End of treatment	Control (n = 21)		Lactulose (n = 19)
	Mean	Standard deviation	Mean	Standard deviation
Psychological subscore	8.0	11	10.9	14
Physical subscore	2.8	4	4.8	6

Table 3. Quero 1996: Sickness Impact Profile selected subscores

Table 4. Prasad 2007: Sickness Impact Profile selected subscores

Change from baseline	Control (n = 20)		Lactulose (n = 25)
	Mean	Standard deviation	Mean	Standard deviation
Psychosocial scales
Social interactions	0.5	0.68	8.5	1.35
Alertness	‐0.75	1.13	10.43	1.73
Emotional behaviour	2.76	1.83	8.98	1.55
Communication	0.75	1.19	2.66	1.22
Total psychological subscore	0.77	0.41	8.47	0.98
Physical scales
Ambulation	‐1.89	1.12	3.67	0.80
Mobility	1.22	1.18	5.36	1.35
Body care and movements	0.72	0.42	1.62	0.55
Total physical subscore	0.01	0.52	2.99	0.56
Independent scales
Sleep and rest	2.29	1.35	9.04	1.95
Work	‐0.06	1.44	15.83	4.45
Home management	0.94	1.19	12.64	2.71
Recreation and pastimes	‐0.28	1.11	11.59	1.97
Eating	‐0.56	1.31	3.88	1.21

Table 4. Prasad 2007: Sickness Impact Profile selected subscores

Table 5. Mittal 2009: Sickness Impact Profile selected subscores

Change from baseline	Control (n = 31)		Lactulose (n = 35)
	Mean	Standard deviation	Mean	Standard deviation
Subscores
Sleep and rest	2.87	6.5	11.64	5.5
Emotional behaviour	0.40	4.1	9.84	4.8
Body care and movements	–0.38	1.9	3.20	2.4
Home management	–0.25	5.7	6.34	5.20
Mobility	0.59	5.5	4.64	4.3
Social interaction	1.63	3.2	3.88	2.8
Alertness	0.18	2.4	3.63	2.2
Ambulation	–0.18	2.9	5.10	4.2
Communication	0.80	3.3	2.07	5.1
Work	0.64	2.5	9.46	15.7
Recreation and pastime	3.06	4.4	7.74	5.7
Eating	1.12	3.1	2.48	3.1
Psychosocial	1.13	2.4	5.17	2.9
Physical	–0.05	2.0	3.59	2.1

Table 5. Mittal 2009: Sickness Impact Profile selected subscores

Table 6. Zeng 2003: WHO‐Bref selected subscores

End of treatment	Control (n = 20)		Short term lactulose (n = 20)		Long‐term lactulose (n = 20)
	Mean	Standard deviation	Mean	Standard deviation	Mean	Standard deviation
Physical health	28	19	37	18	54	19
Psychological health	42	14	44	15	58	15
Social relationships	38	16	42	15	60	17
Environment	51	18	53	15	51	13

Table 6. Zeng 2003: WHO‐Bref selected subscores

Comparison 1. Non‐absorbable disaccharides versus placebo/no intervention

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mortality Show forest plot	24	1487	Risk Ratio (IV, Random, 95% CI)	0.59 [0.40, 0.87]

2 Mortality in trials with a low risk of bias Show forest plot	8	705	Risk Ratio (IV, Random, 95% CI)	0.63 [0.41, 0.97]

3 Hepatic encephalopathy Show forest plot	22	1415	Risk Ratio (IV, Random, 95% CI)	0.58 [0.50, 0.69]

4 Serious adverse events Show forest plot	24	1487	Risk Ratio (IV, Random, 95% CI)	0.47 [0.36, 0.60]

5 Quality of life: sickness impact profile Show forest plot	3		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.1 Change from baseline	2	120	Mean Difference (IV, Random, 95% CI)	7.18 [5.28, 9.07]
5.2 End of treatment	1	40	Mean Difference (IV, Random, 95% CI)	0.90 [‐4.13, 5.93]
6 Non‐serious adverse events Show forest plot	9		Risk Ratio (IV, Random, 95% CI)	Subtotals only

6.1 Overall	9	739	Risk Ratio (IV, Random, 95% CI)	2.47 [1.24, 4.93]
6.2 Diarrhoea	7	634	Risk Ratio (IV, Random, 95% CI)	6.41 [1.84, 22.40]
6.3 Bloating	6	563	Risk Ratio (IV, Random, 95% CI)	4.50 [1.17, 17.27]
6.4 Nausea	1	60	Risk Ratio (IV, Random, 95% CI)	11.00 [0.64, 190.53]
6.5 Constipation	2	298	Risk Ratio (IV, Random, 95% CI)	0.04 [0.01, 0.29]
6.6 Hyponatraemia	1	45	Risk Ratio (IV, Random, 95% CI)	0.35 [0.01, 8.11]
6.7 Anal fissure	1	45	Risk Ratio (IV, Random, 95% CI)	0.35 [0.01, 8.11]
6.8 Hyperglycaemia	1	45	Risk Ratio (IV, Random, 95% CI)	0.35 [0.01, 8.11]
7 Number connection test, end of treatment Show forest plot	6	275	Mean Difference (IV, Random, 95% CI)	‐5.56 [‐11.59, 0.47]

8 Ammonia end of treatment Show forest plot	6	374	Mean Difference (IV, Random, 95% CI)	‐11.64 [‐21.14, ‐2.14]

8.1 Venous	5	216	Mean Difference (IV, Random, 95% CI)	‐15.66 [‐27.79, ‐3.53]
8.2 Arterial	1	158	Mean Difference (IV, Random, 95% CI)	‐2.23 [‐6.89, 2.43]
9 Ammonia change from baseline Show forest plot	3	155	Mean Difference (IV, Random, 95% CI)	18.97 [8.86, 29.09]

9.1 Arterial	2	134	Mean Difference (IV, Random, 95% CI)	10.45 [5.60, 15.31]
9.2 Venous	1	21	Mean Difference (IV, Random, 95% CI)	44.0 [32.34, 55.66]
10 Mortality in worst‐case scenario analyses Show forest plot	24		Risk Ratio (IV, Random, 95% CI)	Subtotals only

10.1 Worst‐case scenario	24	1487	Risk Ratio (IV, Random, 95% CI)	0.61 [0.42, 0.88]
10.2 Extreme worst‐case scenario analysis	24	1487	Risk Ratio (IV, Random, 95% CI)	0.64 [0.44, 0.94]
11 Hepatic encephalopathy worst‐case scenario analysis Show forest plot	22	2830	Risk Ratio (IV, Random, 95% CI)	0.60 [0.54, 0.66]

11.1 Worst‐case scenario	22	1415	Risk Ratio (IV, Random, 95% CI)	0.59 [0.50, 0.69]
11.2 Extreme worst‐case scenario	22	1415	Risk Ratio (IV, Random, 95% CI)	0.60 [0.51, 0.70]
12 Serious adverse events worst‐case scenario analysis Show forest plot	24	2974	Risk Ratio (IV, Random, 95% CI)	0.48 [0.41, 0.57]

12.1 Worst‐case scenario analysis	24	1487	Risk Ratio (IV, Random, 95% CI)	0.47 [0.37, 0.61]
12.2 Extreme worst‐case scenario analysis	24	1487	Risk Ratio (IV, Random, 95% CI)	0.49 [0.38, 0.62]

Comparison 1. Non‐absorbable disaccharides versus placebo/no intervention

Comparison 2. Prevention trials: non‐absorbable disaccharides versus placebo/no intervention

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mortality Show forest plot	6	668	Risk Ratio (IV, Random, 95% CI)	0.63 [0.40, 0.98]

1.1 Primary	4	370	Risk Ratio (IV, Random, 95% CI)	0.56 [0.27, 1.17]
1.2 Secondary	2	298	Risk Ratio (IV, Random, 95% CI)	0.67 [0.39, 1.16]
2 Mortality and bias control Show forest plot	6	668	Risk Ratio (IV, Random, 95% CI)	0.63 [0.40, 0.98]

2.1 Low risk of bias	5	538	Risk Ratio (IV, Random, 95% CI)	0.64 [0.41, 0.99]
2.2 High risk of bias	1	130	Risk Ratio (IV, Random, 95% CI)	0.33 [0.01, 8.03]
3 Hepatic encephalopathy Show forest plot	6	668	Risk Ratio (IV, Random, 95% CI)	0.47 [0.33, 0.68]

3.1 Primary	4	370	Risk Ratio (IV, Random, 95% CI)	0.48 [0.23, 0.98]
3.2 Secondary	2	298	Risk Ratio (IV, Random, 95% CI)	0.44 [0.31, 0.64]
4 Serious adverse events Show forest plot	6	668	Risk Ratio (IV, Random, 95% CI)	0.48 [0.33, 0.70]

4.1 Primary prevention	4	370	Risk Ratio (IV, Random, 95% CI)	0.50 [0.24, 1.03]
4.2 Secondary prevention	2	298	Risk Ratio (IV, Random, 95% CI)	0.44 [0.31, 0.64]
5 Non‐serious adverse events Show forest plot	4		Risk Ratio (IV, Random, 95% CI)	Subtotals only

Comparison 2. Prevention trials: non‐absorbable disaccharides versus placebo/no intervention

Comparison 3. Treatment trials: non‐absorbable disaccharides versus placebo/no intervention

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mortality Show forest plot	18	819	Risk Ratio (IV, Random, 95% CI)	0.49 [0.23, 1.05]

1.1 Overt	6	172	Risk Ratio (IV, Random, 95% CI)	0.36 [0.14, 0.94]
1.2 Minimal	12	647	Risk Ratio (IV, Random, 95% CI)	0.82 [0.24, 2.86]
2 Mortality in trials with a low risk of bias Show forest plot	18	819	Risk Ratio (IV, Random, 95% CI)	0.49 [0.23, 1.05]

2.1 Low risk of bias	3	167	Risk Ratio (IV, Random, 95% CI)	0.56 [0.12, 2.68]
2.2 High risk of bias	15	652	Risk Ratio (IV, Random, 95% CI)	0.47 [0.20, 1.13]
3 Mortality in acute or chronic hepatic encephalopathy Show forest plot	6	172	Risk Ratio (IV, Random, 95% CI)	0.36 [0.14, 0.94]

3.1 Acute	3	102	Risk Ratio (IV, Random, 95% CI)	0.36 [0.14, 0.94]
3.2 Chronic	3	70	Risk Ratio (IV, Random, 95% CI)	0.0 [0.0, 0.0]
4 Hepatic encephalopathy Show forest plot	16	747	Risk Ratio (IV, Random, 95% CI)	0.63 [0.53, 0.74]

4.1 Overt	5	140	Risk Ratio (IV, Random, 95% CI)	0.62 [0.39, 0.99]
4.2 Minimal	11	607	Risk Ratio (IV, Random, 95% CI)	0.63 [0.52, 0.76]
5 Acute or chronic hepatic encephalopathy Show forest plot	5	140	Risk Ratio (IV, Random, 95% CI)	0.62 [0.39, 0.99]

5.1 Acute	3	102	Risk Ratio (IV, Random, 95% CI)	0.59 [0.34, 1.00]
5.2 Chronic	2	38	Risk Ratio (IV, Random, 95% CI)	0.55 [0.07, 4.10]
6 Serious adverse events Show forest plot	18	819	Risk Ratio (IV, Random, 95% CI)	0.42 [0.26, 0.69]

6.1 Overt	6	172	Risk Ratio (IV, Random, 95% CI)	0.40 [0.16, 1.02]
6.2 Minimal	12	647	Risk Ratio (IV, Random, 95% CI)	0.43 [0.24, 0.78]
7 Serious adverse events in acute or chronic hepatic encephalopathy Show forest plot	6	172	Risk Ratio (IV, Random, 95% CI)	0.40 [0.16, 1.02]

7.1 Acute	3	102	Risk Ratio (IV, Random, 95% CI)	0.40 [0.16, 1.02]
7.2 Chronic	3	70	Risk Ratio (IV, Random, 95% CI)	0.0 [0.0, 0.0]
8 Non‐serious adverse events Show forest plot	5	191	Risk Ratio (IV, Random, 95% CI)	2.12 [0.62, 7.28]

Comparison 3. Treatment trials: non‐absorbable disaccharides versus placebo/no intervention

Comparison 4. Lactulose versus lactitol

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mortality Show forest plot	8	225	Risk Ratio (IV, Random, 95% CI)	1.30 [0.59, 2.85]

1.1 Overt hepatic encephalopathy	6	174	Risk Ratio (IV, Random, 95% CI)	1.30 [0.59, 2.85]
1.2 Minimal hepatic encephalopathy	1	20	Risk Ratio (IV, Random, 95% CI)	0.0 [0.0, 0.0]
1.3 Prevention of hepatic encephalopathy	1	31	Risk Ratio (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2 Hepatic encephalopathy Show forest plot	7	194	Risk Ratio (IV, Random, 95% CI)	1.00 [0.84, 1.19]

2.1 Overt hepatic encephalopathy	5	162	Risk Ratio (IV, Random, 95% CI)	1.08 [0.60, 1.96]
2.2 Minimal hepatic encephalopathy	1	20	Risk Ratio (IV, Random, 95% CI)	1.0 [0.83, 1.20]
2.3 Prevention hepatic encephalopathy	1	12	Risk Ratio (IV, Random, 95% CI)	0.20 [0.01, 3.46]
3 Serious adverse events Show forest plot	9	245	Risk Ratio (IV, Random, 95% CI)	1.56 [0.84, 2.88]

4 Non‐serious adverse events Show forest plot	6		Risk Ratio (IV, Random, 95% CI)	Subtotals only

4.1 Overall	6	169	Risk Ratio (IV, Random, 95% CI)	1.55 [0.88, 2.74]
4.2 Diarrhoea	3	61	Risk Ratio (IV, Random, 95% CI)	0.80 [0.39, 1.64]
4.3 Bloating and flatulence	4	128	Risk Ratio (IV, Random, 95% CI)	2.20 [1.06, 4.54]
4.4 Nausea	4	104	Risk Ratio (IV, Random, 95% CI)	3.20 [0.76, 13.43]
4.5 Hyponatraemia	1	25	Risk Ratio (IV, Random, 95% CI)	3.23 [0.14, 72.46]
4.6 Abdominal pain	3	91	Risk Ratio (IV, Random, 95% CI)	0.95 [0.47, 1.91]
4.7 Asthenia	1	31	Risk Ratio (IV, Random, 95% CI)	0.35 [0.02, 8.08]
5 Number Connection Test: end of treatment Show forest plot	4	84	Mean Difference (IV, Random, 95% CI)	‐4.22 [‐16.12, 7.68]

6 Number Connection Test: change from baseline Show forest plot	1	25	Mean Difference (IV, Random, 95% CI)	0.20 [‐0.54, 0.94]

7 Venous blood ammonia: end of treatment Show forest plot	3	72	Mean Difference (IV, Random, 95% CI)	6.47 [‐8.36, 21.29]

8 Venous blood ammonia: change from baseline Show forest plot	1	25	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐0.80, 0.40]

Comparison 4. Lactulose versus lactitol