Fibrin‐based haemostatic agents for reducing blood loss in adult liver resection

Abstract

Background

Liver resection is the optimal treatment for selected benign and malignant liver tumours, but it can be associated with significant blood loss. Numerous anaesthetic and surgical techniques have been developed to reduce blood loss and improve perioperative outcomes. One such technique is the application of topical fibrin‐based haemostatic agents (FBHAs) to the resection surface. There is no standard practice for FBHA use, and a variety of commercial agents and devices are available, as well as non‐FBHAs (e.g. collagen‐based agents). The literature is inconclusive on the effectiveness of these methods and on the clinical benefits of their routine use.

Objectives

To evaluate the benefits and harms of fibrin‐based haemostatic agents in reducing intraoperative blood loss in adults undergoing liver resection.

Search methods

We searched the Cochrane Hepato‐Biliary Group (CHBG) Controlled Trials Register, CENTRAL, MEDLINE, Embase, LILACS, Science Citation Index Expanded, and Conference Proceedings Citation Index‐Science up to 20 January 2023. We also searched online trial registries, checked the reference lists of all primary studies, and contacted the authors of included trials for additional published or unpublished trials.

Selection criteria

We considered for inclusion all randomised clinical trials evaluating FBHAs versus no topical intervention or non‐FBHAs, irrespective of publication type, publication status, language of publication, and outcomes reported. Eligible participants could have any liver pathology and be undergoing major or minor liver resections through open or laparoscopic surgery.

Data collection and analysis

Two review authors independently screened the results of the literature search and used data extraction forms to collate the results. We expressed dichotomous outcome results as risk ratios (RRs) and continuous outcome results as mean differences (MDs), each with their corresponding 95% confidence interval (CI). We used a random‐effects model for the main analyses. Our primary outcomes were perioperative mortality, serious adverse events, haemostatic efficacy, and health‐related quality of life. Our secondary outcomes were efficacy as sealant, adverse events considered non‐serious, operating time, and length of hospital stay. We assessed the certainty of the evidence with GRADE and presented results in two summary of findings tables.

Main results

We included 22 trials (2945 participants) evaluating FBHAs versus no intervention or non‐FBHAs; 19 trials with 2642 participants provided data for the meta‐analyses. Twelve trials reported commercial funding, one trial reported no financial support, and nine trials provided no information on funding. Below we present the most clinically relevant outcome results, also displayed in our summary of findings table.

Fibrin‐based haemostatic agents versus no intervention

Six trials (1001 participants) compared FBHAs with no intervention. One trial was at low risk of bias in all five domains, and all other trials were at high or unclear risk of bias in at least one domain. Two trials were at high risk of bias related to blinding. It is unclear if FBHAs compared with no intervention have an effect on perioperative mortality (RR 2.58, 95% CI 0.89 to 7.44; 4 trials, 782 participants), serious adverse events (RR 0.96, 95% CI 0.88 to 1.05; 4 trials, 782 participants), postoperative transfusion (RR 1.04, 95% CI 0.77 to 1.40; 5 trials, 864 participants), reoperation (RR 2.92, 95% CI 0.58 to 14.61; 2 trials, 612 participants), or postoperative bile leak (RR 1.00, 95% CI 0.67 to 1.48; 4 trials, 782 participants), as the certainty of evidence was very low for all these outcomes.

Fibrin‐based haemostatic agents versus non‐fibrin‐based haemostatic agents

Sixteen trials (1944 participants) compared FBHAs with non‐FBHAs. All trials had at least one domain at high or unclear risk of bias. Twelve trials were at high risk of bias related to blinding. It is unclear if FBHAs compared with non‐FBHAs have an effect on perioperative mortality (RR 1.03, 95% CI 0.62 to 1.72; 11 trials, 1436 participants), postoperative transfusion (RR 0.92, 95% CI 0.68 to 1.25; 7 trials, 599 participants), reoperation (RR 0.48, 95% CI 0.25 to 0.90; 3 trials, 358 participants), or postoperative bile leak (RR 1.15, 95% CI 0.60 to 2.21; 9 trials, 1115 participants), as the certainty of evidence was very low for all these outcomes. FBHAs compared with non‐FBHAs may have little or no effect on the risk of serious adverse events (RR 0.99, 95% CI 0.95 to 1.03; 9 trials, 1176 participants; low‐certainty evidence).

Authors' conclusions

The evidence for the outcomes in both comparisons (FBHAs versus no intervention and FBHAs versus non‐FBHAs) was of very low certainty (or low certainty in one instance) and cannot justify the routine use of FBHAs to reduce blood loss in adult liver resection. While the meta‐analysis showed a reduced risk of reoperation with FBHAs compared with non‐FBHAs, the analysis was confounded by the small number of trials reporting the event and the risk of bias in all these trials.

Future trials should focus on the use of FBHAs in people undergoing liver resection who are at particularly high risk of bleeding. Investigators should evaluate clinically meaningful and patient‐important outcomes and follow the SPIRIT and CONSORT statements.

Plain language summary

Use of fibrin‐based agents to reduce blood loss in adults undergoing liver surgery

Key messages

• The studies investigating fibrin‐based agents for reducing blood loss in adult liver surgery have many design flaws, so their results are very uncertain.
• Based on the available evidence, we cannot recommend or advise against the routine use of fibrin‐based agents.

What is topic of this review?

People who undergo liver surgery for tumours are at risk of significant blood loss, complications after surgery, and even death. Several techniques have been developed to reduce blood loss in liver surgery; one such method is the application of a fibrin‐based agent (fibrin is the end 'scab' protein formed during blood clotting) to the cut liver surface. Non‐fibrin‐based agents are also available. It is unclear whether fibrin‐based agents are effective for reducing blood loss and improving other outcomes compared with no treatment/placebo (dummy treatment) or non‐fibrin‐based agents/devices.

What did we want to find out?

We wanted to find out if fibrin‐based agents are more effective than no treatment/placebo or non‐fibrin base agents/devices for reducing:

• risk of death during and after surgery;
• risk of serious complications;
• time needed to stop bleeding;
• risk of needing a blood transfusion after surgery;
• risk of needing another surgery to stop the bleeding; and
• risk of the liver leaking bile after surgery.

What did we do?

We performed a rigorous search for randomised studies that compared fibrin‐based agents with no intervention/placebo or non‐fibrin‐based agents/devices. Randomised studies normally provide the most robust evidence because they allocate participants to one or another treatment at random, so that the different treatment groups are comparable in terms of characteristics that could influence results. We compared and summarised the results of the studies and rated our confidence in the evidence based on factors such as study methods and sizes.

What did we find?

We found 22 randomised studies that included a total of 2945 adults undergoing liver surgery. Six studies in 1001 adults compared fibrin‐based agents with no intervention, and 16 studies in 1944 adults compared fibrin‐based agents with non‐fibrin‐based agents/devices.

It is unclear if fibrin‐based agents compared with no treatment have an effect on risk of death within 30 days of surgery, risk of complications, blood transfusion, need for further surgery, or risk of bile leaking from the liver after surgery.

Similarly, it is unclear if fibrin‐based agents compared with non‐fibrin‐based agents have an effect on risk of death within 30 days of surgery, risk of needing a blood transfusion after surgery, or risk of bile leaking after surgery. Fibrin‐based agents may reduce the risk of needing another surgery, but we are very uncertain about the results. Fibrin‐based agents may have little or no effect on the risk of serious complications compared to non‐fibrin‐based agents.

What are the limitations of the evidence?

We have very little confidence in the evidence because many studies had design flaws, many studies were small, and most results indicated that the treatment could benefit or harm the person receiving it, or have no effect. Based on current evidence, we cannot recommend or advise against routine use of fibrin‐based agents to reduce blood loss in liver surgery. Future research should focus on use of fibrin‐based agents in people who are at particularly high risk of bleeding.

Funding
Twelve trials reported commercial funding, one trial reported no financial support, and nine trials provided no information on funding.

How up to date is this evidence

The evidence is current to 20 January 2023.

Summary of findings

Summary of findings 1. Fibrin‐based haemostatic agents versus no intervention or placebo.

Fibrin‐based haemostatic agents compared with no intervention or placebo for reducing intraoperative blood loss and improving outcomes in adult liver surgery
Population: adults undergoing liver resection for cancer or benign disease Setting: liver resection unit Intervention: FBHAs applied to resection surface Comparison: haemostasis achieved through measures not including application of a topical agent (fibrin‐based or otherwise)
Outcome	Anticipated risk difference* (95% CI)		Relative effect (95% CI)	Number of participants (RCTs)	Certainty of evidence	Comments
	Risk with no intervention	Risk with FBHAs
Perioperative mortality Median follow‐up 1.5 months after liver resection (range 30 days to 6 months)	13 per 1000	34 per 1000 (12 to 89)	RR 2.58 (0.89 to 7.44)	782 (4)	⊕⊝⊝⊝ Very lowa,b	Defined as death, regardless of cause, occurring within 30 days of surgery in or outside hospital
Serious adverse events Median follow‐up 1.5 months after liver resection (range 30 days to 6 months)	435 per 1000	418 per 1000 (383 to 457)	RR 0.96 (0.88 to 1.05)	782 (4)	⊕⊝⊝⊝ Very lowa,b	Only 2/4 trials reported averse events according to the definition provided in ICH‐GCP 2016.
Postoperative transfusion Median follow‐up 1.5 months after liver resection (range 30 days to 6 months)	168 per 1000	175 per 1000 (129 to 235)	RR 1.04 (95% CI 0.77 to 1.40)	864 (5)	⊕⊝⊝⊝ Very lowa,b	—
Reoperation Median follow‐up 1.5 months after liver resection (range 30 days to 6 months)	7 per 1000	20 per 1000 (4 to 102)	RR 2.92 (0.58 to 14.61)	612 (2)	⊕⊝⊝⊝ Very lowa,b	—
Postoperative bile leak Median follow‐up 1.5 months after liver resection (range 30 days to 6 months)	113 per 1000	113 per 1000 (76 to 167)	RR 1.00 (0.67 to 1.48)	782 (4)	⊕⊝⊝⊝ Very lowa,b	—
*The risk in the intervention group (and its 95% CI) is based on the assumedrisk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; FBHA: fibrin‐based haemostatic agent; RCT: randomised clinical trial; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^a Downgraded one level for study limitations (overall high risk of bias). 
^b Downgraded two levels due to serious imprecision (the optimal information size was not met (i.e. sample size < 1000), wide CIs in the result, few events, and the 95% CI includes both benefits and harms).

Summary of findings 2. Fibrin‐based haemostatic agents versus non‐fibrin‐based haemostatic agents.

Fibrin‐based haemostatic agents compared with non‐fibrin‐based haemostatic agents for reducing intraoperative blood loss and improving outcomes in adult liver surgery
Population: adults undergoing liver resection for cancer or benign disease Setting: clinical setting in liver resection unit Intervention: FBHAs applied to resection surface Comparison: non‐FBHAs applied to resection surface
Outcome	Anticipated risk difference (95 % CI)		Relative effect (95% CI)	Number of participants (RCTs)	Certainty of evidence	Comments
	Risk with non‐FBHAs	Risk with FBHAs
Perioperative mortality Median follow‐up 1.25 months (range 1 to 3 months)	47 per 1000	48 per 1000 (29 to 80)	RR 1.03 (0.62 to 1.72)	1436 (11)	⊕⊝⊝⊝ Very lowa,b	Defined as death, regardless of cause, occurring within 30 days of surgery in or outside hospital
Serious adverse events Median follow‐up 1 month (range 1 to 3 months)	736 per 1000	729 per 1000 (699 to 758)	RR 0.99 (0.95 to 1.03)	1176 (9)	⊕⊕⊝⊝ Lowa,c	4 trials defined adverse events according to the Medical Dictionary for Regulatory Activities.
Postoperative transfusion Median follow‐up 1 month (range 1 to 3 months)	270 per 1000	248 per 1000 (184 to 338)	RR 0.92 (0.68 to 1.25)	599 (7)	⊕⊝⊝⊝ Very lowa,b	—
Reoperation Median follow‐up 1.5 months (range 1 to 3 months)	163 per 1000	78 per 1000 (41 to 147)	RR 0.48 (0.25 to 0.90)	358 (3)	⊕⊝⊝⊝ Very lowa,b	1 trial reported an exceptionally high rate of reoperation in both groups; this is likely to have skewed the meta‐analysis.
Postoperative bile leak Median follow‐up 1.5 months (range 1 to 3 months)	31 per 1000	36 per 1000 (19 to 69)	RR 1.15 (0.60 to 2.21)	1115 (9)	⊕⊝⊝⊝ Very lowa,b	—
*The risk in the intervention group (and its 95% CI) is based on the assumedrisk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; FBHA: fibrin‐based haemostatic agent; RCT: randomised clinical trial; RR: risk ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^a Downgraded one level for study limitations (overall high risk of bias).
^b Downgraded two levels for imprecision (few events and wide 95% CI including both benefit and harm).
^c Downgraded one level for imprecision (95% CI including both benefit and harm).

Background

Description of the condition

Liver resection is the treatment of choice for many benign and malignant liver tumours (Jarnagin 2002). Despite the considerable improvements in surgical technique since the late nineteenth century (Langenbuch 1888; Warvi 1945), and the improved understanding of segmental liver anatomy (Couinaud 1999), liver resection still carries a risk of significant intraoperative bleeding (Berrevoet 2007). This is because the liver has an abundant venous vascular network and cannot vasoconstrict in response to injury (Clark 1970). In addition, bleeding can be complicated by coagulation disorders related to the underlying disease.

Estimated mortality following liver resection ranges from 3% in people without liver cirrhosis to 25% in those with cirrhosis (Jarnagin 2002: Mullin 2005). Intraoperative bleeding is the main cause of morbidity and mortality in liver resection (Ibrahim 2006; Jarnagin 2002; Olthof 2019). Allogeneic blood transfusion also increases morbidity and mortality through transfusion reactions and transmission of blood‐borne infective agents, and possibly through immunosuppressive effects of donor‐derived leukocytes (Kimura 2017; Shinozuka 2000).

Description of the intervention

Various surgical, anaesthetic, and pharmacological techniques have been applied to minimise blood loss and achieve haemostasis in liver surgery (Appendix 1). Cochrane Reviews have examined the evidence behind the effectiveness of various vascular occlusion techniques, parenchymal transection techniques, cardiopulmonary manoeuvres, and antifibrinolytics for reducing blood loss in liver resections (Gurusamy 2009a; Gurusamy 2009b; Gurusamy 2009c; Gurusamy 2012; Moggia 2016).

Fibrin‐based haemostatic agents (FBHAs) are bioabsorbable topical haemostatic agents used in liver resection to enhance haemostasis and reduce blood loss. These products contain clotting factors and act by mimicking the final stages of the coagulation cascade ending in the formation of fibrin, thus promoting the process of coagulation and wound healing. The main components of commercially available FBHAs are thrombin and fibrinogen, derived from human plasma and typically purified from cryoprecipitate.

How the intervention might work

Thrombin and fibrinogen are usually delivered to the site of bleeding via a dual‐syringe or aerosol applicator (Bektas 2014; Bjelović 2018; de Boer 2012; Figueras 2007). In some products, the clotting factors are coated on a collagen sheet and applied as a surgical patch (Bochicchio 2015; Fischer 2011; Frilling 2005; Genyk 2016; Kakaei 2013; Koea 2013; Koea 2016). Non‐commercial preparations of FBHAs utilise autologous or allogenic blood products as a source of fibrinogen, which is then combined with topical bovine thrombin (Mintz 2001). The relative concentrations of fibrinogen and thrombin within these preparations affect the haemostatic process. Higher concentrations of thrombin tend to accelerate clot formation and increase clot adhesion, whereas higher concentrations of fibrinogen produce stronger clots. Fibrinogen concentrations are usually higher and more consistent in commercial FBHAs than non‐commercial preparations (Albala 2003). Other additives, including factor XIII and antifibrinolytic agents (e.g. aprotinin or tranexamic acid; Karanicolas 2016), can also be included in these preparations to improve clot stability (Table 3).

1. Examples of fibrin‐based haemostatic agents and their main components.

Product name	Fibrinogen concentration (mg/mL)	Thrombin concentration (IU/mL)	Factor XIII concentration (U/mL)	Antifibrinolytic agent
Beriplast P (CSL Behring, Germany)	90	500	60	Bovine aprotinin: 1000 KIU/mL
Biocol (LFB, France)	127	558	11	Bovine aprotinin: 3000 KIU/mL
Bolheal (Kaketsuken Pharmaceutical, Japan)	80	250	75	Bovine aprotinin: 1000 KIU/mL
Evicel – formerly Crosseal® (Omrix Biopharmaceuticals, Israel)	70	1000	None	None
Quixil (Omrix Biopharmaceuticals, Israel)	50	1000	None	Tranexamic acid: 85 to 105 mg/mL
Tachosil (Nycomed, Switzerland)	5.5a	2.0b	None	None
Tisseel (Baxter, Austria)	90	500	30	Bovine aprotinin: 3000 KIU/mL

^amg/cm².
^bIU/cm².
KIU: kallikrein inhibitor units.

In addition to their effect on haemostasis, FBHAs are thought to have a sealant effect, preventing the leakage of bile; as such, they may be effective in reducing the incidence of other complications of liver resection such as intra‐abdominal fluid collection (Berrevoet 2007). A single‐centre retrospective cohort study of 610 people investigating risk of bile leak following liver resection found that use of fibrin glue was associated with reduced postoperative bile leak on multivariable analysis (Capussotti 2006). However, the study lacked blinding and a control group, and was not adequately powered to investigate this association. One randomised controlled trial reported a significantly lower incidence of postoperative bile leakage with FBHAs (0%) versus no topical intervention (5.45%) in people undergoing primary closure after laparoscopic common bile duct exploration (Zhang 2019); however, this conclusion cannot be extrapolated to people undergoing liver resection. Preclinical experiments in landrace pigs suggest that FBHAs may improve liver regeneration after hepatectomy, but no published trials have studied this effect in humans (Fonouni 2019).

There are some safety concerns associated with FBHA use. FBHAs that contain plasma‐derived or bovine products carry the risk of transmitting viral and prion infections (Spotnitz 2005). One study suggested that up to 20% of people who receive plasma fibrin may experience viral transmission (Kawamura 2002). Repeated exposure to bovine products can also lead to immunologically induced coagulopathies (Banninger 1993; Rapaport 1992) and anaphylactic reactions (Mitsuhata 1994).

Why it is important to do this review

Various topical haemostatic agents are available to reduce resection surface bleeding during liver resection. Although many experts advocate FBHAs as the most favourable sealants (Berrevoet 2007), multiple meta‐analyses have found no beneficial effect of these products on clinical outcomes in liver surgery (Brustia 2016; Ding 2013; Sanjay 2013; Wells 2020). Nevertheless, FBHAs are frequently used in liver surgery with the aim of reducing blood loss and biliary leakage (Boonstra 2009; Nakajima 2002). The precise reason why previous trials have found no benefit is unclear. Analyses may be affected by confounding; for example, by including people with cirrhosis. Therefore, it is important to evaluate the literature critically and systematically for the benefits and risks of using FBHAs compared with no intervention or non‐FBHAs in these procedures. Previous reviews have grouped different controls (e.g. no intervention, collagen‐based haemostatic agent, argon plasma coagulation) into a single comparator cohort, resulting in a comparison between FBHAs and a heterogenous control group. The limitation of this approach is that heterogeneity is likely to confound the analyses. Wells 2020 performed a network analysis to compare multiple interventions, but did not include a sensitivity analysis or subgroup analysis to address the effect of excluding people with cirrhosis. Similarly, Brustia 2016, Ding 2013, and Sanjay 2013 did not address this confounder.

This review aimed to critically appraise and summate the evidence from randomised clinical trials, using Cochrane methods, to analyse the potential benefits of routine use of FBHAs in liver surgery for benign and malignant tumours (Higgins 2022a).

Objectives

To evaluate the benefits and harms of fibrin‐based haemostatic agents in reducing intraoperative blood loss in adults undergoing liver resection.

Methods

Criteria for considering studies for this review

Types of studies

We included all randomised clinical trials that compared an FBHA with no intervention/placebo or another haemostatic agent (non‐FBHA), irrespective of publication type, publication status, language of publication, and reported outcomes.

We were also interested in collecting information on adverse effects from any quasi‐randomised or observational studies identified through the searches for randomised clinical trials.

Types of participants

Adults (aged 18 years or older) undergoing liver resection, regardless of site of resection, extent of resection, and underlying liver pathology.

Types of interventions

Eligible experimental interventions were any commercial or non‐commercial FBHAs, regardless of additive (antifibrinolytic agents or factor XIII). Eligible controls were no intervention, placebo, or non‐FBHAs.

We allowed co‐interventions if administered equally to the experimental and control groups.

Types of outcome measures

Unless otherwise stated, we used the published data from the longest follow‐up for our main analyses.

Primary outcomes

• Perioperative mortality: defined as death,regardless of cause, occurring within 30 days after surgeryin or outside hospital (Jacobs 2006)

• Serious adverse events, such as infection, coagulopathy, or anaphylaxis, defined according to the International Conference on Harmonization Guidelines for Good Clinical Practice (ICH‐GCP 2016). The formal definition of an adverse event was "any untoward medical occurrence in a patient or clinical investigation subject administered a pharmaceutical product and which does not necessarily have to have a causal relationship with this treatment".

• Haemostatic efficacy as measured in the individual trial

  • Time to haemostasis (continuous)

  • Haemostasis within a predefined time period (dichotomous)

  • Volume of intraoperative blood loss: volume of blood lost during liver resection (continuous)

  • Postoperative transfusion: need for transfusion of packed red blood cells, platelets, or fresh frozen plasma during or after surgery, up to the time of discharge (dichotomous)

  • Reoperation (dichotomous)

• Health‐related quality of life, assessed with any validated scale or questionnaire

Secondary outcomes

• Efficacy as sealant

  • Postoperative bile leak (dichotomous): defined according to the International Study Group of Liver Surgery as bilirubin concentration in the drain fluid at least three times the serum bilirubin concentration on or after the third postoperative day, or as the need for radiological or operative intervention resulting from biliary collections or bile peritonitis (Koch 2011)

  • Postoperative intra‐abdominal collections (dichotomous): identified radiologically or intraoperatively

  • Abdominal drain output (continuous): daily drain volume

  • Time to removal of abdominal drains (continuous)

• Adverse events considered non‐serious or not included in the definition of serious adverse events

• Operating time

• Length of hospital stay

Search methods for identification of studies

Electronic searches

We searched the following electronic databases.

• Cochrane Hepato‐Biliary Group Controlled Trials Register (searched internally by the CHBG Information Specialist via the Cochrane Register of Studies Web; 20 January 2023)

• Cochrane Central Register of Controlled Trials (CENTRAL; 2023, Issue 1) in the Cochrane Library (searched 20 January 2023)

• MEDLINE Ovid (1946 to 20 January 2023)

• Embase Ovid (1974 to 20 January 2023)

• LILACS (BIREME; 1982 to 20 January 2023)

• Science Citation Index Expanded (Web of Science; 1900 to 20 January 2023)

• Conference Proceedings Citation Index‐Science (Web of Science; 1990 to 20 January 2023)

Appendix 2 presents the search strategies with the date range of the searches.

Searching other resources

We searched pharmaceutical company sources and the following online trial registries for ongoing or unpublished trials.

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov/; searched 20 January 2023)

• European Medicines Agency (EMA; www.ema.europa.eu/ema/; searched 20 January 2023)

• World Health Organization (WHO) International Clinical Trial Registry Platform (ICTRP; www.who.int/ictrp; searched 20 January 2023)

• Food and Drug Administration (FDA; www.fda.gov; searched 20 January 2023)

• EU Clinical Trials Register (www.clinicaltrialsregister.eu/; searched 20 January 2023)

We also searched for grey literature in the System for Information on Grey Literature in Europe 'OpenGrey' (www.opengrey.eu/; searched 20 January 2023).

Lastly, we checked the reference lists of all primary studies and relevant review articles for additional references, and we contacted authors of identified trials for additional published or unpublished trials.

During searches for relevant trials, whenever we identified observational studies (i.e. quasi‐randomised trials) related to FBHAs, we planned to record data on adverse events or harms.

Data collection and analysis

Selection of studies

Two review authors (AKM and AOA) independently screened the titles and abstracts of all records returned by the searches and discarded those that were clearly irrelevant. The same two review authors then read the full‐text articles of all potentially relevant studies to decide whether they met our inclusion criteria.

We checked for postpublication amendments and examined any relevant retraction statements and errata for information, as errata can reveal important limitations or even fatal flaws in included studies (Lefebvre 2022).

We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Page 2021a; Page 2021b), and we presented reasons for excluding studies in the Characteristics of excluded studies table.

Data extraction and management

Two review authors (AKM and AOA) independently extracted data from the included trials using standard data extraction forms. Where multiple publications from the same study existed, we included the report with the most complete data. However, we also extracted and included relevant data from other reports of the same study as appropriate.

We extracted the following study characteristics.

• Methods: study design, study period, number of study centres and location, study setting, withdrawals/dropouts, date of the study

• Participants: mean age, age range, sex, inclusion criteria, exclusion criteria, background liver parenchyma (presence of cirrhosis, steatosis, or both), type of liver resection (major or minor)

• Interventions: intervention, comparison, concomitant interventions

• Outcomes: planned outcomes in the trial protocol, if available, for later comparison during the risk of bias assessment; reported outcomes; time points of outcome data

• Notes: funding for studies, notable conflicts of interest of trial authors

Assessment of risk of bias in included studies

Two review authors (AKM and AOA) independently assessed the risk of bias in the included trials, following the recommendations outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We assessed the following risk of bias domains in each trial (Kjaergard 2001; Lundh 2017; Moher 1998; Savović 2012a; Savović 2012b; Savović 2018; Schulz 1995; Wood 2008).

• Random sequence generation

  • Low risk of bias: sequence generation was achieved using computer random number generation or a random number table. Drawing lots, tossing a coin, shuffling cards, and throwing dice would have been considered adequate if performed by an independent investigator 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 (e.g. 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, or the assessment of outcomes was not likely to be influenced by lack of blinding.

  • Unclear risk of bias: there was insufficient information to assess whether blinding was likely to induce bias on 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. Sufficient methods, such as multiple imputation, were employed to handle missing data.

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

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

• Selective outcome reporting

  • Low risk of bias: all outcomes were predefined and reported, or all clinically relevant and reasonably expected outcomes were reported. These include mortality, adverse events, and haemostatic efficacy related to the use of FBHAs. Trials had to be registered either on the ClinicalTrials.gov website (www.clinicaltrials.gov) or in a similar register, or be associated with a published protocol (e.g. published in a paper journal). For trials that were conducted and published in the years before trial registration was necessary (before July 2005), we scrutinised all trial publications to identify the trial objectives and outcomes. If the results sections of the reports included usable data on all outcomes specified in the trial objectives, we considered the trial at low risk of reporting bias.

  • Unclear risk of bias: unclear if all predefined and clinically relevant and reasonably expected outcomes were reported.

  • High risk of bias: one or more clinically relevant and reasonably expected outcomes were not reported, and data on these outcomes were likely recorded.

• Other bias

  • Low risk of bias: the trial appeared to be free of any other components that could put it at risk of bias (e.g. no commercial conflicts of interest declared in the publication).

  • Unclear risk of bias: the trial may or may not have been free of any other components that could put it at risk of bias (e.g. unclear whether the study was commercially sponsored).

  • High risk of bias: there were other factors in the trial that could put it at risk of bias (e.g. clear statement of commercial sponsorship).

We summarised the risk of bias for each included trial. We classified trials at overall high risk of bias if they had an unclear or high risk judgement in any of the above domains. We considered trials at overall low risk of bias only if all domains were at low risk.

Assessment of bias in conducting the systematic review

We conducted the review according to the published protocol and reported any deviations from it in the Differences between protocol and review section of the review (Amer 2013).

Measures of treatment effect

We meta‐analysed outcome data from different trials according to the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2022), using Review Manager 5.4 (Review Manager 2020).

For dichotomous outcomes, we calculated the treatment effect using the risk ratio (RR) with a 95% confidence interval (CI) if more than one trial reported the outcome. If the outcome of interest was reported in only one trial, we calculated the Fisher's exact test P value to identify any significant differences between the treatment groups.

For continuous outcomes, we calculated the treatment effect using the mean difference (MD) or standardised mean difference (SMD) with 95% CI. For quality of life measures, we planned to calculate standardised difference in terms of the minimal important difference (MID).

Unit of analysis issues

In trials with a parallel‐group design, the unit of analysis was the trial participant as randomised within the trial. For trials with more than two intervention arms, we used the arms relevant to our review, though we described all trial arms in the Characteristics of included studies table.

For dichotomous outcomes, we used participants rather than events as the unit of analysis. We recorded whether the trial measured adverse events in relation to frequency of participants with an adverse event or multiple adverse events in the same participant. We also planned to record where trials had incorrectly analysed multiple events in a participant as independent events. Where the number of observations appeared to be equal to the number of participants, we treated the events as the unit of analysis (Higgins 2022b).

We did not anticipate or encounter any challenges with non‐standard designs such as cross‐over trials due to the nature of the intervention investigated.

Dealing with missing data

We performed analyses on an intention‐to‐treat (ITT) basis whenever possible.

We did not exclude trials on the basis of missing data. First, we attempted to contact trial authors to obtain the missing information, and if we were unsuccessful, we attempted to impute the data with different methods according to data type.

For missing dichotomous outcomes, we included participants with incomplete or missing data in sensitivity analyses by imputing the data according to the following scenarios.

• Extreme‐case analysis favouring the experimental intervention ('best‐worst' case scenario): no dropouts/participants lost from the experimental arm and all dropouts/participants lost from the control arm experienced the outcome, with all randomised participants in the denominator.

• Extreme‐case analysis favouring the control ('worst‐best' case scenario): all dropouts/participants lost from the experimental arm and no dropouts/participants lost from the control arm experienced the outcome, with all randomised participants in the denominator.

For continuous outcomes, we calculated missing standard deviations using reported P values or CIs (Higgins 2022b). If this was not possible, we used the highest standard deviation reported in the other trials for the corresponding treatment group and outcome.

We addressed the potential effects of missing data on the results in the Discussion section, and we reported dropout rates in the Characteristics of included studies table.

Assessment of heterogeneity

We assessed heterogeneity using the Chi² test with significance set at a P value of 0.10. We quantified the degree of heterogeneity using the I² statistic and interpreted the observed values using the following thresholds (Deeks 2022).

• 0% to 40%: might not be important

• 30% to 60%: may represent moderate heterogeneity

• 50% to 90%: may represent substantial heterogeneity

• 75% to 100%: considerable heterogeneity

If we identified substantial heterogeneity (I² > 50%), we planned to explore the possible causes by prespecified subgroup analyses.

Assessment of reporting biases

We used funnel plots to visually assess whether estimates were associated with study size in meta‐analyses that included at least 10 trials. We used two tests to assess funnel plot asymmetry: the adjusted rank correlation test (Begg 1994), and the regression asymmetry test (Egger 1997).

Data synthesis

Where trials were clinically and methodologically comparable, we attempted to combine the outcomes from individual trials in a meta‐analysis to provide a pooled effect estimate for each outcome. We used the random‐effects model for meta‐analyses of outcomes reported by more than one trial (DerSimonian 1986). We performed separate analyses for trials evaluating FBHAs versus no intervention/placebo and trials evaluating FBHAs versus non‐FBHAs to reduce confounding through potentially differing effects of the control intervention.

Subgroup analysis and investigation of heterogeneity

We planned to perform the following subgroup analyses if at least 10 trials reported the outcome of interest (Deeks 2022).

• Trials at low risk of vested interests versus trials at unclear or high risk of vested interests

  • We anticipated that differences between commercial and generic preparations could impact treatment effect size.

• Trials at overall low risk of bias versus trials at overall high risk of bias.

  • Trials at high risk of bias may overestimate beneficial intervention effects or underestimate harmful intervention effects.

• Commercial FBHAs manufactured with antifibrinolytic agents versus commercial FBHAs manufactured without antifibrinolytic agents

  • We anticipated that the addition of antifibrinolytic agents to FBHAs could alter the treatment effect size, as fibrinolysis leads to clot breakdown.

• Commercial FBHAs manufactured with factor XIII versus commercial FBHAs manufactured without factor XIII

  • We anticipated that the addition of factor XIII to FBHAs could alter the treatment effect size.

• Trials including people with cirrhosis versus trials excluding people with cirrhosis (or not specifying background liver status)

  • The inclusion of participants with cirrhosis could influence the treatment effect of FBHAs due to increased bleeding diathesis secondary to associated coagulopathy.

Sensitivity analysis

We planned to perform the following sensitivity analyses to assess the robustness of our results to arbitrary decisions and assumptions made when defining the eligibility of trials.

• Extreme‐case analysis favouring the experimental intervention ('best‐worst' case scenario)

• Extreme‐case analysis favouring the control ('worst‐best' case scenario)

• Exclusion of trials that only enroled people with normal hepatic parenchyma (that excluded people with cirrhosis, steatohepatitis, etc.)

• Repeat analysis with the fixed‐effect meta‐analysis model

• Exclusion of any 'outlier' trials

• Assessment of imprecision with Trial Sequential Analysis (TSA)

Trial Sequential Analysis

Cumulative meta‐analyses are at risk of producing random errors as a result of sparse data and repetitive testing of the accumulating data (Brok 2008; Brok 2009; Thorlund 2009; Thorlund 2010; Wetterslev 2008; Wetterslev 2009). For this reason, we applied TSA (TSA 2017; Thorlund 2017). To minimise random errors, we calculated the required information size (the number of participants needed in a meta‐analysis to detect or reject a certain intervention effect), which took into account the diversity present in the meta‐analysis (Brok 2008; Brok 2009; Thorlund 2009; Thorlund 2010; Wetterslev 2008; Wetterslev 2009). In our meta‐analysis, we based the required information size on the proportion of participants with the event in the control group; an alpha (risk of type I error) of 2% for both the primary and secondary outcomes; a beta (risk of type II error) of 20%; and the assumed diversity of the meta‐analysis (Wetterslev 2009). We assumed a relative risk reduction of 20%. We used the software 0.9.5.10 Beta to perform the TSA using the random‐effects model. The fundamental assumption of TSA is that testing for significance may be performed each time a new trial is added to the meta‐analysis. We added the trials according to the year of publication; where two or more trials were published in the same year, we added them alphabetically according to the last name of the first author. We constructed the trial sequential monitoring boundaries on the basis of the required information size (Thorlund 2017; Wetterslev 2008; Wetterslev 2017). These boundaries determine the statistical inference one may draw regarding the cumulative meta‐analysis that does not reach the required information size; if the trial sequential monitoring boundary for benefit or harm is crossed before the required information size is reached, firm evidence might be established, and further trials might be superfluous. Nonetheless, if the boundaries are not exceeded, more trials are needed to detect or reject a certain intervention effect. That is determined by assessing if the cumulative Z‐curve crosses the trial sequential monitoring boundary for futility.

In TSA, we downgrade assessment of imprecision by two levels when the accrued number of participants is below 50% of the diversity‐adjusted required information size (DARIS), and by one level when 50% or greater but below 100% of DARIS. We do not downgrade when futility or DARIS is reached (Jakobsen 2014).

Summary of findings and assessment of the certainty of the evidence

We assessed the quality of evidence at the outcome level across trials using GRADEpro (GRADEpro GDT), following recommendations provided in Chapter 8, Chapter 14, and Chapter 15 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022c; Schünemann 2022a; Schünemann 2022b), and in the GRADE Handbook (Schünemann 2013). Summary of findings tables present key information regarding the certainty of the evidence, the magnitude of the effects of the interventions examined, and the sum of the available data for the outcomes. We prepared a separate summary of findings table for each comparison (FBHAs versus no intervention/placebo and FBHAs versus non‐FBHAs) with the following outcomes.

• Perioperative mortality

• Serious adverse events

• Postoperative transfusion

• Reoperation

• Postoperative bile leak

The GRADE approach considers five factors that reduce the certainty of evidence in randomised clinical trials (risk of bias, inconsistency of results, indirectness of evidence, imprecision, and publication bias) and classifies the certainty of evidence into four levels, which are interpreted as follows.

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

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

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

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

Two review authors (AKM and AOA) independently assessed the certainty of the evidence for each outcome, resolving any discrepancies through discussion with a third review author (CW). We used footnotes to justify all decisions to downgrade the certainty of evidence.

Results

Description of studies

The Characteristics of included studies table provides a detailed description of each included trial. We presented reasons for excluding studies during the full‐text review stage in the Characteristics of excluded studies table.

Results of the search

Figure 1 shows the study selection process in a PRISMA flow diagram. The searches performed on 20 January 2023 identified 2736 records. Following removal of duplicates, we screened 2277 records, of which we excluded 2251 based on information in the titles and abstracts. We retrieved 31 full‐text articles for further assessment. We excluded four trials (four reports) and included 22 trials (27 reports).

1.

PRISMA diagram. Date of search 20 January 2023.

Included studies

We included 22 randomised clinical trials of 2945 participants undergoing liver resection (Bektas 2014; Bjelović 2018; Bochicchio 2015; Chu 2008; de Boer 2012; Figueras 2007; Fischer 2011; Franceschi 2006; Frilling 2005; Genyk 2016; Gugenheim 2011; Kakaei 2013; Koea 2013; Koea 2016; Kohno 1992; Moench 2014; Noun 1996; Öllinger 2013; Scheurer 2016; Schwartz 2004; Troisi 2012; Verhoef 2014). One trial had three arms (Kakaei 2013), and the remaining 21 trials had two arms. Eight trials were single‐centre trials (Bektas 2014; Chu 2008; Figueras 2007; Gugenheim 2011; Kakaei 2013; Kohno 1992; Noun 1996; Scheurer 2016), and the remaining 14 were multicentre trials. Participants were enroled in Australia, Austria, Belgium, Denmark, France, Germany, Iran, Japan, the Netherlands, Serbia, Spain, Taiwan, the UK, and the USA.

We did not specifically search for observational studies, but considered quasi‐randomised and observational studies retrieved by the searches for randomised trials only for the reported adverse events related to the use of FBHAs in liver resection. We acknowledge that by adopting this approach, we may have missed or overlooked uncommon and late adverse events reported in observational studies (Storebø 2018).

We meta‐analysed data from 19 trials involving 2555 participants (Bektas 2014; Bjelović 2018; Bochicchio 2015; Chu 2008; de Boer 2012; Figueras 2007; Fischer 2011; Frilling 2005; Genyk 2016; Gugenheim 2011; Kakaei 2013; Koea 2013; Koea 2016; Kohno 1992; Moench 2014; Noun 1996; Öllinger 2013; Schwartz 2004; Verhoef 2014). Three trials published as abstracts provided insufficient information for inclusion in the meta‐analyses, and we were unable to obtain further information from the trial authors (Franceschi 2006; Scheurer 2016; Troisi 2012). We presented the findings of these trials separately.

Trial participants

The trials included 2955 participants undergoing liver resection, randomised to receive either an FBHA or control (no intervention or a non‐FBHA). Participant age ranged from 18 to 85 years. Five trials specifically included people with cirrhosis and provided a breakdown of the number of participants with cirrhosis in each group (Bektas 2014; Bjelović 2018; Figueras 2007; Koea 2013; Koea 2016). Three trials excluded participants with cirrhosis (Fischer 2011; Moench 2014; Öllinger 2013). The remaining 14 trials provided insufficient detail on background liver status or did not specify whether they had excluded people with cirrhosis, and we were unable to perform the associated subgroup analysis. However, we did perform sensitivity analyses by removing trials that excluded people with cirrhosis.

Trial comparisons

Experimental intervention

The FBHAs investigated were unspecified non‐commercial fibrin sealants (Bektas 2014; de Boer 2012; Figueras 2007; Gugenheim 2011; Koea 2013; Noun 1996), Fibrin Sealant Grifols (Bjelović 2018), Fibrocaps (Bochicchio 2015; Verhoef 2014), Tisseel (Chu 2008), TachoSil (Fischer 2011; Frilling 2005; Genyk 2016; Kakaei 2013; Moench 2014; Öllinger 2013; Scheurer 2016; Troisi 2012), CryoSeal (Franceschi 2006), EVARREST fibrin sealant patch (Koea 2016), Beriplast P (Kohno 1992), and Crosseal (Schwartz 2004).

Control interventions

No intervention or placebo

Six trials involving 1001 participants compared FBHAs with no (topical haemostatic) intervention (Bektas 2014; Chu 2008; de Boer 2012; Figueras 2007; Koea 2016; Noun 1996). No trials used placebos as controls.

Other topical haemostatic agents

Sixteen trials involving 1944 participants compared FBHAs with other topical haemostatic agents (Bjelović 2018; Bochicchio 2015; Fischer 2011; Franceschi 2006; Frilling 2005; Genyk 2016; Gugenheim 2011; Kakaei 2013; Koea 2013; Koea 2016; Kohno 1992; Moench 2014; Öllinger 2013; Scheurer 2016; Schwartz 2004; Troisi 2012; Verhoef 2014). Non‐FBHAs used as comparators in trials included Surgicel (Bjelović 2018; Genyk 2016; Koea 2013; Koea 2016), gelatin sponge alone (Bochicchio 2015; Verhoef 2014), argon beam coagulator (Fischer 2011; Frilling 2005), collagen‐based haemostatic agent (Franceschi 2006), PlasmaJet (Gugenheim 2011), Avitene (Kohno 1992), SanguStop (Moench 2014), Veriset (Öllinger 2013; Troisi 2012), and Hemopatch (Scheurer 2016). One trial compared an FBHA with two non‐FBHAs (Surgicel and Glubaran; Kakaei 2013).

Co‐interventions

Four trials investigated FBHAs with the addition of factor XIII and an antifibrinolytic agent (Bektas 2014; Figueras 2007; Kohno 1992; Noun 1996). Two trials investigated FBHAs with the addition of an antifibrinolytic only (de Boer 2012; Gugenheim 2011).

Follow‐up

Five trials did not specify the follow‐up duration (Chu 2008; Fischer 2011; Gugenheim 2011; Kohno 1992; Scheurer 2016). Two trials limited follow‐up to the inpatient stay (Kakaei 2013; Noun 1996). In the remaining 15 trials, follow‐up ranged from 29 days to six months after surgery.

Dropouts and intention‐to‐treat analysis

Seven trials reported dropouts, with 61 dropouts in total (Bjelović 2018; Bochicchio 2015; Fischer 2011; Frilling 2005; Genyk 2016; Koea 2016; Verhoef 2014). Twelve trials used an ITT analysis (Bjelović 2018; de Boer 2012; Figueras 2007; Fischer 2011; Frilling 2005; Genyk 2016; Koea 2013; Koea 2016; Moench 2014; Öllinger 2013; Schwartz 2004; Verhoef 2014).

No trials were stopped early due to benefit or harm.

Funding

Funding entities included ProFibrix Inc (Bochicchio 2015; Verhoef 2014), Ethicon (Koea 2013; Koea 2016), Baxter (Bektas 2014), Grifols (Bjelović 2018), Johnson & Johnson (de Boer 2012), Nycomed (Fischer 2011), Thermogenesis corp (Franceschi 2006), Takeda pharma (Genyk 2016), Aesculap (Moench 2014), and Covidien (Öllinger 2013). One trial was funded through academic grants without financial support from a commercial entity (Figueras 2007). Nine trials provided no information on funding (Chu 2008; Frilling 2005; Gugenheim 2011; Kakaei 2013; Kohno 1992; Noun 1996; Scheurer 2016; Schwartz 2004; Troisi 2012).

Excluded studies

We excluded four studies. In Uetsuji 1994, the FBHA was not applied directly to the liver surface following resection, but rather to the undersurface of the diaphragm and around the insertion of the liver ligaments. The trial investigated whether FBHAs used in this manner were effective at preventing pleural effusions. Two trials compared one FBHA with another FBHA (Kawasaki 2017; Kobayashi 2016). Liu 1993 was a quasi‐randomised trial, with participants allocated to treatment groups based on researcher's judgement; this trial did not report adverse events related to the intervention.

Risk of bias in included studies

The following section contains an overview of some common biases present in the included trials. Figure 2 presents review authors' judgements about each risk of bias item for each included study, and Figure 3 presents review authors' judgements about each risk of bias item as percentages across all included studies.

2.

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

3.

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

Allocation

Random sequence generation

We judged nine trials at low risk of bias for this domain because they described appropriate methods of randomisation (Bektas 2014; Bjelović 2018; de Boer 2012; Fischer 2011; Genyk 2016; Kakaei 2013; Koea 2013; Moench 2014; Schwartz 2004). Thirteen trials provided insufficient information on the method of randomisation (Bochicchio 2015; Chu 2008; Figueras 2007; Franceschi 2006; Frilling 2005; Gugenheim 2011; Koea 2016; Kohno 1992; Noun 1996; Öllinger 2013; Scheurer 2016; Troisi 2012; Verhoef 2014), leading to an unclear risk of bias judgement.

Allocation concealment

Eleven trials described appropriate methods of allocation concealment so were at low risk of bias related to allocation concealment (Bektas 2014; Bjelović 2018; de Boer 2012; Fischer 2011; Genyk 2016; Gugenheim 2011; Koea 2013; Kohno 1992; Moench 2014; Öllinger 2013; Schwartz 2004). Eleven trials provided insufficient information regarding allocation concealment (Bochicchio 2015; Chu 2008; Figueras 2007; Franceschi 2006; Frilling 2005; Kakaei 2013; Koea 2016; Noun 1996; Scheurer 2016; Troisi 2012; Verhoef 2014), leading to an unclear risk of bias judgement.

Blinding

One trial was at low risk of bias in both blinding domains (de Boer 2012). We judged 13 trials at high risk of performance and detection bias (Bektas 2014; Bjelović 2018; Bochicchio 2015; Fischer 2011; Frilling 2005; Genyk 2016; Kakaei 2013; Koea 2013; Koea 2016; Moench 2014; Noun 1996; Öllinger 2013; Verhoef 2014). One trial was at unclear risk of performance bias and high risk of detection bias (Scheurer 2016). Six trials provided insufficient information to make a judgement (Chu 2008; Figueras 2007; Franceschi 2006; Gugenheim 2011; Kohno 1992; Troisi 2012). Schwartz 2004 blinded participants, staff, and data analysts (low risk of performance bias), but not the operating surgeon; as haemostatic efficacy was based on operator judgement, we rated the trial at high risk of detection bias.

Incomplete outcome data

We judged 16 trials at low risk of attrition bias because they described the follow‐up protocols (Bektas 2014; Bjelović 2018; Bochicchio 2015; de Boer 2012; Figueras 2007; Frilling 2005; Genyk 2016; Kakaei 2013; Koea 2013; Koea 2016; Kohno 1992; Moench 2014; Noun 1996; Öllinger 2013; Schwartz 2004; Verhoef 2014). The remaining six trials provided insufficient information on the follow‐up protocol (unclear risk of bias).

Selective reporting

Eighteen trials had published trial protocols and reported the prespecified outcomes, so were at low risk of reporting bias (Bektas 2014; Bjelović 2018; de Boer 2012; Fischer 2011; Frilling 2005; Genyk 2016; Gugenheim 2011; Kakaei 2013; Koea 2013; Koea 2016; Kohno 1992; Moench 2014; Noun 1996; Öllinger 2013; Scheurer 2016; Schwartz 2004; Troisi 2012; Verhoef 2014). The remaining four trials provided insufficient information to make a judgement on reporting bias (unclear risk).

Other potential sources of bias

Nine trials provided no information on funding sources (Chu 2008; Frilling 2005; Gugenheim 2011; Kakaei 2013; Kohno 1992; Noun 1996; Scheurer 2016; Schwartz 2004; Troisi 2012), so we could not judge risk of bias from commercially vested interests (unclear risk of bias). Commercial entities funded 12 trials, introducing potential risk of bias from commercially vested interests (Bektas 2014; Bjelović 2018; Bochicchio 2015; de Boer 2012; Fischer 2011; Franceschi 2006; Genyk 2016; Koea 2013; Koea 2016; Moench 2014; Öllinger 2013; Verhoef 2014). However, none of those trials reported significantly outlying results, and 11 trials were at low risk of reporting bias, suggesting that the impact of commercial funding on bias was likely minimal (low risk of other bias). One trial was not funded by a commercial entity and was at low risk of bias from commercially vested interest (Figueras 2007). Four trials were published in abstract form only (Chu 2008; Franceschi 2006; Scheurer 2016; Troisi 2012), with insufficient information provided to fully ascertain if there were other sources of bias present in the studies, resulting in an unclear risk of other bias.

Effects of interventions

See: Table 1; Table 2

Fibrin‐based haemostatic agents versus no intervention or placebo

No trials used placebo as the control intervention. Five trials compared FBHAs with no intervention in 1001 participants (Bektas 2014; de Boer 2012; Figueras 2007; Koea 2016; Noun 1996). As there were fewer than 10 trials in this comparison, we could not perform any of the prespecified subgroup analyses. No trials excluded people with cirrhosis, so we were unable to perform the associated sensitivity analysis. Although Koea 2016 reported two dropouts (one participant in each group), outcome data for dichotomous outcomes were available, so sensitivity analyses for the 'best‐worse' case scenario and 'worst‐best' case scenario were unnecessary.

Table 1 presents the main findings of our meta‐analyses of trials that compared FBHAs with no intervention.

Primary outcomes

Perioperative mortality

Four trials reported perioperative mortality (Bektas 2014; de Boer 2012; Figueras 2007; Koea 2016). It is unclear if FBHAs compared with no intervention have an effect on the risk of perioperative mortality (RR 2.58, 95% CI 0.89 to 7.44; P = 0.08, I² = 0%; 782 participants; very low‐certainty evidence; Analysis 1.1). To test the robustness of the model, we also used a fixed‐effect model for the main analysis; the result favoured no intervention (RR 2.80, 95% CI 1.02 to 7.73; P = 0.05, I² = 0%; analysis not shown).

1.1. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 1: Perioperative mortality

Trial Sequential Analysis for imprecision

We downgraded the certainty of the evidence by two levels for imprecision as the accrued number of participants was below 50% of the DARIS. The TSA showed that the DARIS was 68,668 participants, calculated based on the proportion of death of 1.3% in the control group, a relative risk reduction of 20%, an alpha (type I error) of 2%, a beta (type II error) of 20% (power of 80%), and a diversity of 0%. TSA could not be constructed because there was too little information.

Serious adverse events

Four trials reported serious adverse events (Bektas 2014; de Boer 2012; Figueras 2007; Koea 2016). It is unclear if FBHAs compared with no intervention have an effect on risk of serious adverse events (RR 0.96, 95% CI 0.88 to 1.05; P = 0.38, I² = 0%; 782 participants: very low‐certainty evidence; Analysis 1.2). Bektas 2014 and Koea 2016 defined and reported serious adverse events based on the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use guidance on clinical safety data management; whereas de Boer 2012 and Figueras 2007 reported complications but did not formally define serious adverse events in their methods.

1.2. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 2: Serious adverse events

Haemostatic efficacy: time to haemostasis

Only Koea 2016 reported time to haemostasis, which was shorter in the FBHA group than in the no intervention group (4 minutes versus 4.7 minutes, P < 0.001, very low certainty of evidence).

Haemostatic efficacy: haemostasis within a predefined time period

Two trials reported the likelihood of achieving haemostasis within four minutes of FBHA application (Bektas 2014; Koea 2016). The likelihood of achieving haemostasis within four minutes was higher in the FBHA group compared to no intervention, with no heterogeneity on meta‐analysis (RR 2.12, 95% CI 1.65 to 2.73; P < 0.001, I² = 0%; 172 participants; very low‐certainty evidence; Analysis 1.3); however, the evidence is very uncertain due to the low number of trials and risk of bias.

1.3. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 3: Haemostasis within a predefined time period

Haemostatic efficacy: volume of intraoperative blood loss

Three trials reported volume of intraoperative blood loss (de Boer 2012; Figueras 2007; Koea 2016). Intraoperative blood loss was higher with FBHAs than with no intervention (MD 101.26 mL, 95% CI 15.81 mL to 186.71 mL; P = 0.02, I² = 0%; 712 participants; very low‐certainty evidence; Analysis 1.4); however, the evidence is very uncertain due to the low number of trials and high risk of bias.

1.4. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 4: Volume of intraoperative blood loss (mL)

Haemostatic efficacy: postoperative transfusion

Five trials reported risk of postoperative transfusion (Bektas 2014; de Boer 2012; Figueras 2007; Koea 2016; Noun 1996). It is unclear if FBHAs compared with no intervention have an effect on risk of postoperative transfusion (RR 1.04, 95% CI 0.77 to 1.40; P = 0.79, I² = 0%; 864 participants; very low‐certainty evidence; Analysis 1.5).

1.5. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 5: Postoperative transfusion

Haemostatic efficacy: reoperation

Two trials reported risk of reoperation (de Boer 2012; Figueras 2007). It is unclear if FBHAs compared with no intervention have an effect on risk of reoperation (RR 2.92, 95% CI 0.58 to 14.61; P = 0.19, I² = 0%; very low‐certainty evidence; 612 participants; Analysis 1.6) due to the low number of trials reporting the outcome.

1.6. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 6: Reoperation

Health‐related quality of life

No trials reported health‐related quality of life.

Secondary outcomes

Efficacy as sealant: postoperative bile leak

Four trials reported postoperative bile leak (Bektas 2014; de Boer 2012; Figueras 2007; Koea 2016). It is unclear if FBHAs compared with no intervention have an effect on risk of bile leak (RR 1.00, 95% CI 0.67 to 1.48; P = 0.99, I² = 0%; 782 participants; very low‐certainty evidence; Analysis 1.7).

1.7. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 7: Postoperative bile leak

Efficacy as sealant: postoperative intra‐abdominal collections

Six trials reported postoperative intra‐abdominal collections (Bektas 2014; Chu 2008; de Boer 2012; Figueras 2007; Koea 2016; Noun 1996). It is unclear if FBHAs compared with no intervention have an effect on risk of intra‐abdominal collections (RR 1.01, 95% CI 0.81 to 1.28; P = 0.91, I² = 0%; 906 participants; very low‐certainty evidence; Analysis 1.8).

1.8. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 8: Postoperative intra‐abdominal collections

Sensitivity analysis

Chu 2008 reported outcomes in two separate abstracts published in 2006 and 2008. According to the 2006 abstract, 47 participants were randomised (28 to the FBHA group and 19 to the no intervention group), whereas the 2008 abstract stated that 142 participants were randomised. The 2006 abstract reported data on the incidence of intra‐abdominal fluid collections and the 2008 abstract reported the total length of drainage. We performed a sensitivity analysis by removing this trial which was at risk of bias due to imbalanced group sizes. This sensitivity analysis showed no difference between treatment groups in the risk of intra‐abdominal collections (RR 1.04, 95% CI 0.83 to 1.32; P = 0.72, I² = 0%; 3 trials, 859 participants; Analysis 1.9).

1.9. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 9: Postoperative intra‐abdominal collections: sensitivity analysis

Efficacy as sealant: abdominal drain output

Four trials reported abdominal drain output in 588 participants (Bektas 2014; Chu 2008; Figueras 2007; Noun 1996); however; Chu 2008 did not provide sufficient information for inclusion in the meta‐analysis. It is unclear if FBHAs compared with no intervention have an effect on abdominal drain output (MD −71.20mL, 95% CI −423.54 mL to 281.14 mL; P = 0.69, I² = 46%; 3 trials, 447 participants; very low‐certainty evidence; Analysis 1.10).

1.10. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 10: Abdominal drain output (mL)

Efficacy as sealant: time to removal of abdominal drains

Four trials reported time to removal of abdominal drains (Bektas 2014; Chu 2008; de Boer 2012; Figueras 2007). It is unclear if FBHAs compared with no intervention have an effect on time to removal of abdominal drains (MD −0.53 days, 95% CI −2.03 days to 0.97 days; P = 0.49, I² = 78%; 822 participants; very low‐certainty evidence; Analysis 1.11).

1.11. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 11: Time to removal of abdominal drains (days)

Adverse events considered non‐serious or not included in the definition of serious adverse events

No trials reported adverse events considered non‐serious.

Operating time

Three trials reported operating time (Figueras 2007; Koea 2016; Noun 1996). It is unclear if FBHAs compared with no intervention have an effect on operating time (MD 14.86 minutes, 95% CI −0.34 minutes to 30.06 minutes; P = 0.06, I² = 0%; 479 participants; very low‐certainty evidence; Analysis 1.12).

1.12. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 12: Operating time (minutes)

Length of hospital stay

Three trials reported length of hospital stay (Figueras 2007; Koea 2016; Noun 1996). It is unclear if FBHAs compared with no intervention have an effect on length of hospital stay (MD −0.15 days, 95% CI −1.63 days to 1.33 days; P = 0.85, I² = 0%; 479 participants; very low‐certainty evidence; Analysis 1.13).

1.13. Analysis.

Comparison 1: Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo, Outcome 13: Length of hospital stay (days)

Fibrin‐based haemostatic agents versus non‐fibrin‐based haemostatic agents

Thirteen trials compared FBHAs with non‐FBHAs in 1641 participants (Bjelović 2018; Bochicchio 2015; Fischer 2011; Frilling 2005; Genyk 2016; Gugenheim 2011; Kakaei 2013; Koea 2013; Kohno 1992; Moench 2014; Öllinger 2013; Schwartz 2004; Verhoef 2014). Three trials excluded participants with cirrhosis (Fischer 2011; Moench 2014; Öllinger 2013). We were unable to perform subgroup analyses because fewer than 10 trials reported the outcome of interest, or due to small subgroup sizes (fewer than two trials reporting at least one event). We performed sensitivity analyses where appropriate. Although six trials reported a total of 63 dropouts (38 in the FBHA group and 25 in the non‐FBHA group), outcome data for dichotomous outcomes were available, so sensitivity analyses for the 'best‐worst' case scenario and 'worst‐best' case scenario were unnecessary (Bjelović 2018; Bochicchio 2015; Fischer 2011; Frilling 2005; Genyk 2016; Verhoef 2014).

Table 2 presents the main findings of our meta‐analyses of trials that compared FBHAs with non‐FBHAs.

Primary outcomes

Perioperative mortality

Thirteen trials (1641 participants) reported perioperative mortality (Bjelović 2018; Bochicchio 2015; Fischer 2011; Frilling 2005; Genyk 2016; Gugenheim 2011; Kakaei 2013; Koea 2013; Kohno 1992; Moench 2014; Öllinger 2013; Schwartz 2004; Verhoef 2014); there were no events reported in Kakaei 2013 and Koea 2013. It is unclear if FBHAs compared with non‐FBHAs have an effect on the risk of perioperative mortality (RR 1.03, 95% CI 0.62 to 1.72; P = 0.91, I² = 0%; 11 trials, 1436 participants; very low‐certainty evidence; Analysis 2.1). Figure 4 shows a funnel plot of this analysis; it is symmetrical and does not suggest publication bias.

2.1. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 1: Perioperative mortality

4.

To ensure the robustness of the model, we repeated the main analysis using a fixed‐effect model. There was no difference in perioperative mortality risk between the groups (RR 1.01, 95% CI 0.63 to 1.62; P = 0.97, I² = 0%; analysis not shown).

Sensitivity analysis

When we removed the trials that had excluded people with cirrhosis (Fischer 2011; Moench 2014; Öllinger 2013), we found no difference between the effect of FBHAs versus non‐FBHAs on perioperative mortality (RR 1.02, 95% CI 0.58 to 1.80; P = 0.93, I² = 0%; 8 trials, 1142 participants; Analysis 2.2). Gugenheim 2011 investigated non‐commercial FBHAs. As we were unable to perform a subgroup analysis based on risk of vested interests (less than 10 trials), we performed a sensitivity analysis by removing Gugenheim 2011; there was no difference between the groups (RR 0.98, 95% CI 0.58 to 1.68; P = 0.95, I² = 0%; 10 trials, 1378 participants; Analysis 2.3).

2.2. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 2: Perioperative mortality: sensitivity analysis removing trials that excluded participants with cirrhosis

2.3. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 3: Perioperative mortality: sensitivity analysis removing a trial that investigated a non‐commercial FBHA

Trial Sequential Analysis for imprecision

We downgraded the certainty of the evidence by two levels for imprecision as the accrued number of participants was below 50% of the DARIS of 18,406 participants (Figure 5).

5.

Fibrin‐based haemostatic agents versus non‐fibrin‐based haemostatic agents

Perioperative mortality

The Trial Sequential Analysis (TSA) showed that the diversity‐adjusted required information size (DARIS) is 18,406 participants, calculated based on the proportion of death of 4.7% in the control group; a relative risk reduction (RRR) of 20%; an alpha (type Ⅰ error) of 2%; a beta (type Ⅱ error) of 20% (power of 80%); and a diversity of 0%. The cumulative Z‐curve did not cross the conventional boundary based on the included four trials (1436 participants), nor the trial sequential monitoring boundaries of benefit and harm, suggesting that no conclusive evidence was found; hence more trials are needed. The TSA‐adjusted 95% confidence interval is 0.13 to 8.27. The x‐axis denotes the amassed versus required information size of the trial participants. The y‐axis denotes the Z values, representing the accumulating statistical information. The blue line (Z‐curve) shows the cumulative Z value. The small red lines, at the top and bottom corners, show the trial sequential boundaries for benefit or harm, representing the threshold for statistical significance in TSA. The horizontal green dotted lines show the threshold for significance in conventional meta‐analysis, at 1.96 of the Z value, corresponding to 0.05 of the P value. The red dotted lined triangular shape shows the futility boundaries and futility area in TSA.

Serious adverse events

Nine trials reported serious adverse events (Bjelović 2018; Fischer 2011; Frilling 2005; Genyk 2016; Koea 2013; Kohno 1992; Moench 2014; Öllinger 2013; Verhoef 2014). FBHAs compared with non‐FBHAs may have little or no effect on the risk of serious adverse events (RR 0.99, 95% CI 0.95 to 1.03; P = 0.63, I² = 0%; in 1176 participants; low‐certainty evidence; Analysis 2.4). Four trials defined serious adverse events according to the Medical Dictionary for Regulatory Activities (MedDRA), developed under the auspices the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (Fischer 2011; Frilling 2005; Öllinger 2013; Verhoef 2014). The remaining five trials did not formally define serious adverse events.

2.4. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 4: Serious adverse events

Sensitivity analysis

When we removed the trials that had excluded people with cirrhosis, we found no difference between the groups (RR 0.99, 95% CI 0.94 to 1.03; P = 0.62, I² = 0%; 6 trials, 882 participants; Analysis 2.5).

2.5. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 5: Serious adverse events: sensitivity analysis removing trials that excluded participants with cirrhosis

Haemostatic efficacy: time to haemostasis

Eight trials reported time to haemostasis (Bjelović 2018; Bochicchio 2015; Fischer 2011; Frilling 2005; Moench 2014; Öllinger 2013; Schwartz 2004; Verhoef 2014). Time to haemostasis was shorter in the FBHA group compared with the non‐FBHA group on meta‐analysis, but the evidence is of very low certainty (MD −1.58 minutes, 95% CI −2.68 minutes to −0.49 minutes; P = 0.004, I² = 93%; 1093 participants; very low‐certainty evidence; Analysis 2.6). Moench 2014 was an outlier for this outcome, with results favouring non‐FBHA (MD 1.18 minutes, 95% CI 0.73 minutes to 1.63 minutes; 126 participants). This was a single‐blind non‐inferiority trial comparing Sangustop (collagen‐based haemostatic fleece) with Tachosil (carrier‐bound fibrin sealant), and the primary outcome was the number of participants achieving complete haemostasis by three minutes postapplication. We found no specific reason in the trial methodology or results to explain the outlying result; however, as time to haemostasis following application was not a primary outcome, the trial may not have been sufficiently powered to form firm conclusions in this regard.

2.6. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 6: Time to haemostasis (minutes)

Sensitivity analysis

When we removed the trials that had excluded people with cirrhosis, we found reduced time to haemostasis associated with FBHAs (MD −1.77 minutes, 95% CI −2.54 minutes to −1.01 minutes; P < 0.001, I² = 71%; 5 trials, 799 participants; Analysis 2.7). Removing Moench 2014, which reported an outlying result, also showed reduced time to haemostasis with FBHAs versus non‐FBHAs (−1.86 minutes, 95% CI −2.52 to −1.21 minutes; P < 0.001, I² = 70%; 7 trials, 967 participants; Analysis 2.8).

2.7. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 7: Time to haemostasis (minutes): sensitivity analysis removing trials that excluded participants with cirrhosis

2.8. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 8: Time to haemostasis (minutes): sensitivity analysis removing trials with extreme outliers

Haemostatic efficacy: haemostasis within a predefined time period

Seven trials reported the number of participants achieving haemostasis within a predefined time following application: the predefined time was four minutes in three trials (Bjelović 2018; Koea 2013; Öllinger 2013) and five minutes in four trials (Bochicchio 2015; Genyk 2016; Koea 2013; Moench 2014; Verhoef 2014). Meta‐analysis showed that use of FBHAs compared with non‐FBHAs increased the likelihood of participants achieving haemostasis within the predefined time period; however, the evidence is of very low certainty owing to considerable heterogeneity between trials and high risk of bias (RR 1.30, 95% CI 1.04 to 1.62; P = 0.02, I² = 90%; 789 participants; very low‐certainty evidence; Analysis 2.9). Two trials were outliers for this outcome and both favoured FBHAs: Bjelović 2018 (RR 2.23, 95% CI 1.41 to 3.52, 70 participants) and Koea 2013 (RR 2.79, 95% CI 1.73 to 4.50, 84 participants). Bjelović 2018 was an open‐label multicentre phase 3 randomised controlled trial comparing an FBHA (Fibrin Sealant Grifols) with a cellulose‐based haemostatic agent (Surgicel); it reported superior haemostatic efficacy with the FBHA. Eligibility in this trial was determined through a qualitative assessment of bleeding by the operating surgeon. If bleeding was considered 'moderate' following primary haemostatic measures, the person was eligible and randomised to an intervention. This may have introduced heterogeneity due to potential subjectivity in assessment of bleeding. However, as randomisation occurred after the assessment, the likelihood of selection bias impacting the results is low. Koea 2013 was an open‐label multicentre randomised clinical trial evaluating use of a fibrin pad with manual compression versus use of an approved topical haemostatic agent (containing oxidised regenerative cellulose). The trial reported superior haemostatic efficacy with the fibrin pad. Randomisation occurred intraoperatively following identification of a target bleeding site where conventional measures failed to achieve haemostasis. This was a subjective assessment by the operating surgeon; however, as randomisation occurred following assessment of the target bleeding site, this subjectivity is unlikely to have led to selection bias that could confound the results.

2.9. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 9: Haemostasis within a predefined time period

Sensitivity analysis

When we removed the trials that had excluded people with cirrhosis, we found greater likelihood of achieving haemostasis at the predefined time in the FBHA group compared with the non‐FBHA group (RR 1.54, 95% CI 1.22 to 1.94; P < 0.001, I² = 80%; 5 trials, 614 participants; Analysis 2.10). The sensitivity analysis without Bjelović 2018 and Koea 2013, which reported outlying results, showed no difference between the groups (RR 1.11, 95% CI 0.94 to 1.31; P = 0.20, I² = 82%; 5 trials, 635 participants; Analysis 2.11).

2.10. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 10: Haemostasis within a predefined time period: sensitivity analysis removing trials that excluded participants with cirrhosis

2.11. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 11: Haemostasis within a predefined time period: sensitivity analysis removing trials with extreme outliers

Haemostatic efficacy: volume of intraoperative blood loss

Three trials reported volume of intraoperative blood loss (Kakaei 2013; Koea 2013; Kohno 1992). It is unclear if FBHAs compared with non‐FBHAs have an effect on intraoperative blood loss (MD 111.24 mL, 95% CI −102.96 mL to 325.45 mL; P = 0.31, I² = 0%; 176 participants; very low‐certainty evidence; Analysis 2.12).

2.12. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 12: Volume of intraoperative blood loss (mL)

Haemostatic efficacy: postoperative transfusion

Seven trials reported postoperative transfusion (Bochicchio 2015; Gugenheim 2011; Kakaei 2013; Koea 2013; Moench 2014; Öllinger 2013; Verhoef 2014). It is unclear if FBHAs compared with non‐FBHAs have an effect on risk of postoperative transfusion (RR 0.92, 95% CI 0.68 to 1.25; P = 0.60, I² = 18%; 599 participants; very low‐certainty evidence; Analysis 2.13).

2.13. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 13: Postoperative transfusion

Sensitivity analysis

When we removed the trials that had excluded people with cirrhosis, we found no difference between the groups (RR 0.79, 95% CI 0.56 to 1.12; P = 0.18, I² = 0%; 5 trials, 423 participants; Analysis 2.14).

2.14. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 14: Postoperative transfusion: sensitivity analysis removing trials that excluded participants with cirrhosis

Reoperation

Six trials (867 participants) reported reoperation (Bjelović 2018; Bochicchio 2015; Gugenheim 2011; Kohno 1992; Moench 2014; Schwartz 2004). Participants from only three trials required reoperation (Bochicchio 2015; Kohno 1992; Schwartz 2004). FBHAs were associated with a lower risk of reoperation relative to non‐FBHAs, but the evidence is very uncertain (RR 0.48, 95% CI 0.25 to 0.90; P = 0.02, I² = 0%; 3 trials, 358 participants; very low‐certainty evidence; Analysis 2.15).

2.15. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 15: Reoperation

Sensitivity analysis

Schwartz 2004 reported unexpectedly high reoperation rates (16% in the FBHA group and 36% in the non‐FBHA group), which likely skewed the meta‐analysis result. When we removed this trial, we found no difference between the groups (RR 0.63, 95% CI 0.12 to 3.22; P = 0.58, I² = 0%; 2 trials, 242 participants; Analysis 2.16). The trial authors provided no comment on the unexpectedly high rate of reoperation, and there were no distinguishing features in the cohort that could explain this anomaly.

2.16. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 16: Reoperation: sensitivity analysis removing trials with extreme outliers

Health‐related quality of life

No trials reported health‐related quality of life.

Secondary outcomes

Postoperative bile leak

Nine trials reported postoperative bile leak (Bjelović 2018; Frilling 2005; Genyk 2016; Kakaei 2013; Koea 2013; Kohno 1992; Moench 2014; Öllinger 2013; Verhoef 2014). FBHAs compared with non‐FBHAs may have little or no effect on risk of bile leak (RR 1.15, 95% CI 0.60 to 2.21; P = 0.67, I² = 0%; 1115 participants; low‐certainty evidence; Analysis 2.17).

2.17. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 17: Postoperative bile leak

Sensitivity analysis

When we removed the trials that had excluded people with cirrhosis, we found no difference between the groups (RR 1.19, 95% CI 0.58 to 2.44; P = 0.64, I² = 0%; 7 trials, 939 participants; Analysis 2.18).

2.18. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 18: Postoperative bile leak: sensitivity analysis removing trials that excluded participants with cirrhosis

Postoperative intra‐abdominal collections

Six trials reported postoperative intra‐abdominal collections (Genyk 2016; Kakaei 2013; Koea 2013; Moench 2014; Öllinger 2013; Schwartz 2004). It is unclear if FBHAs compared with non‐FBHAs have an effect on risk of intra‐abdominal collections (RR 0.72, 95% CI 0.24 to 2.21; P = 0.57, I² = 42%; 664 participants; very low‐certainty evidence; Analysis 2.19).

2.19. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 19: Postoperative intra‐abdominal collections

Abdominal drain output

Three trials reported abdominal drain output (Fischer 2011; Frilling 2005; Verhoef 2014). It is unclear if FBHAs compared with non‐FBHAs have an effect on abdominal drain output (MD −212.49 mL, 95% CI −939.53 mL to 514.55 mL; P = 0.57, I² = 99%; 296 participants; very low‐certainty evidence; Analysis 2.20). Fischer 2011 was an outlier for this outcome, reporting reduced drain output with FBHAs (MD − 916 mL, 95% CI −1005.03 mL to −826.97 mL; 119 participants). We found no specific reasons in the trial design or results that could explain this finding.

2.20. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 20: Abdominal drain output (mL)

Sensitivity analysis

A sensitivity analysis without the outlier demonstrated increased drain output with FBHAs versus non‐FBHAs (MD 138.53 mL, 95% CI 66.51 mL to 210.55 mL; P < 0.001, I² = 0%; 2 trials, 177 participants; Analysis 2.21).

2.21. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 21: Abdominal drain output (mL): sensitivity analysis removing trials with extreme outliers

Time to removal of abdominal drains

Four trials reported time to drain removal (Fischer 2011; Frilling 2005; Kakaei 2013; Kohno 1992). It is unclear if FBHAs compared with non‐FBHAs have an effect on time to drain removal (MD 0.07 days, 95% CI −1.39 days to 1.53 days; P = 0.93, I² = 81%; 332 participants; very low‐certainty evidence; Analysis 2.22).

2.22. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 22: Time to removal of abdominal drains (days)

Adverse events considered non‐serious or not included in the definition of serious adverse events

No trials reported adverse events considered non‐serious.

Operating time

Four trials reported operating time (Genyk 2016; Koea 2013; Kohno 1992; Öllinger 2013). It is unclear if FBHAs compared with non‐FBHAs have an effect on operating time (MD −1.27 minutes, 95% CI −22.73 minutes to 20.19 minutes; P = 0.91, I² = 0%; 419 participants; very low‐certainty evidence; Analysis 2.23).

2.23. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 23: Operating time (minutes)

Length of hospital stay

Five trials reported length of hospital stay (Genyk 2016; Kakaei 2013; Koea 2013; Öllinger 2013; Schwartz 2004). It is unclear is FBHAs compared with non‐FBHAs have an effect on length of hospital stay (MD −0.20 days, 95% CI −1.36 days to 0.96 days; P = 0.74, I² = 8%; 503 participants; very low‐certainty evidence; Analysis 2.24).

2.24. Analysis.

Comparison 2: Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs, Outcome 24: Length of hospital stay (days)

Qualitative analysis of studies not included in the meta‐analyses

Three trials provided insufficient information for inclusion in the meta‐analysis (Franceschi 2006; Scheurer 2016; Troisi 2012). None of these trials reported on perioperative mortality.

Franceschi 2006 compared an FBHA with a collagen‐based haemostatic agent in 153 participants undergoing liver resection, randomised at 10 sites. Time to haemostasis was significantly shorter in the FBHA group (3.43 minutes) compared with the collagen‐based haemostatic agent group (8.65 minutes), and a greater proportion of the FBHA group achieved haemostasis within 10 minutes of application (94.0% versus 60.4%). We were unable to include Franceschi 2006 in the meta‐analysis because it did not report the number of participants allocated to each group.

Scheurer 2016 compared FBHA with a collagen‐based haemostatic agent, based on five unspecified qualitative outcomes, in 100 participants undergoing liver resection. The trial reported no difference between groups regarding benefits secondary to the intervention. We were unable to include Scheurer 2016 in the meta‐analysis because it provided insufficient details regarding the qualitative indicators.

Troisi 2012 compared an FBHA with a non‐FBHA in 50 participants undergoing liver resection at six centres, with time to haemostasis as the primary outcome. Time to haemostasis was shorter in the FBHA group (1.0 minutes versus 3.0 minutes, P < 0.001), and safety profiles were similar in both groups. We were unable to include Troisi 2012 in the meta‐analysis because it did not specify the number of participants randomised to each group or report the number of participants achieving haemostasis at a prespecified time.

Out of the four excluded studies, there was one quasi‐randomised study of relevance to our review, but it did not report data on harm.

Discussion

Summary of main results

Twenty‐two trials involving 2945 participants were eligible for this review, and 19 trials involving 2642 participants contributed data to meta‐analyses. Six trials compared FBHAs with no topical intervention, and 16 trials compared FBHAs with non‐FBHAs. No trials compared FBHAs with placebo. There was very low‐certainty evidence that FBHAs reduced time to haemostasis and increased the likelihood of participants achieving haemostasis at a predefined time compared with no intervention and non‐FBHAs. These findings did not translate to beneficial clinical outcomes. We observed no difference in perioperative mortality, adverse events, or postoperative transfusion requirements. Volume of intraoperative blood loss was higher with FBHAs versus no intervention, but there was no difference between FBHAs and non‐FBHAs, and the evidence for this outcome in both comparisons was of very low certainty. The shorter time to haemostasis in the FBHA group compared with the no intervention group could plausibly be associated with reduced intraoperative blood loss, but this was not observed in our meta‐analysis. It is unclear why FBHAs showed superior haemostatic efficacy results, but it may be related to a type 1 statistical error where a difference (in time to haemostasis or intraoperative blood loss) was observed by chance only.

For risk of reoperation, there was no difference between FBHAs and no intervention, but FBHAs were superior to non‐FBHAs. The results of the analysis in the second comparison were likely skewed by a single trial that reported an unexpectedly high reoperation rate (Schwartz 2004). The trial authors provided no comment on this observation, and there were no distinguishing features in the cohort that could explain the anomaly. A sensitivity analysis without this trial showed no difference in the reoperation rate between FBHAs and non‐FBHAs. Moreover, the evidence for this outcome in both comparisons is of very low certainty. Risk of reoperation was only reported in three trials. This outcome may be affected by publication bias, as that there is a for‐profit interest in FBHAs. The evidence on risk of reoperation was very uncertain. No trials reported health‐related quality of life.

We observed no difference in the effect of FBHAs and either control (no intervention or non‐FBHAs) on risk of bile leak, risk of intra‐abdominal collections, abdominal drain output, time to removal of abdominal drains, operating time, and length of hospital stay. Additionally, there was inconsistency between trials in the reporting of adverse events, with variable use of a standardised reporting tool; this may have confounded the analysis of adverse events.

Overall completeness and applicability of evidence

The 22 trials included in this review are from a range of countries (Australia, Austria, Belgium, Denmark, France, Germany, Iran, Japan, the Netherlands, Serbia, Spain, Taiwan, the UK, and the USA), and several were multicentre trials. Trials reported both short‐term outcomes (e.g. duration of inpatient stay) and longer‐term outcomes, up to six months after surgery. Eleven of 22 trials were published after 2012. For all these reasons, our findings are applicable across multiple settings in the modern era of liver surgery.

The trials included in this review varied in terms of haemostatic efficacy outcomes, with some trials measuring time to haemostasis and others measuring haemostasis within a predefined time period. No trials defined (or formally assessed) achievement of haemostasis according to the guidelines developed by the Scientific and Standardisation Subcommittee on Control of Anticoagulation of the International Society on Thrombosis and Haemostasis, which includes an assessment of haemoglobin, need for transfusion, and invasive intervention for haemostasis for 'non‐visible' bleeding (Khorsand 2016). This may be reflected in our results, as the I² value for the haemostatic efficacy outcomes in the analyses comparing FBHAs with non‐FBHAs exceeded 30%, demonstrating significant heterogeneity between the trials.

Readmission rates following surgery is an important surrogate of surgical complications, but no trials reported this outcome, which further limits the completeness and wider applicability of the evidence. Whether routine use of FBHAs is justified also depends on the financial cost of the intervention, but no trials compared total healthcare‐associated costs. Such an analysis would be difficult to perform due to difficulties in data capture (e.g. emergency attendances to healthcare facilities outside the hospital where the surgery was performed) and differences in costs between countries in multinational trials.

There are multiple confounding factors in the assessment of FBHA efficacy for reducing blood loss in adult liver resection. Subgroup analyses can clarify the potential impact of these factors on treatment effect. However, we were unable to perform most prespecified subgroup analyses because too few trials reported the outcome or the subgroups were too small. Additionally, any subgroup analyses are likely to be grossly underpowered. Not all trials reported the number of participants with cirrhosis. Exclusion or inclusion of such patients, and the proportion of the total study population they comprise, could significantly affect trial results.

Although clinical equipoise between interventions and comparators was present in all the included trials, the results may be confounded by additional rescue interventions (anaesthetic or surgical) that likely occurred when the FBHA or control intervention did not achieve haemostasis. Indeed, it would be unethical not to allow other rescue measures in these cases due to the significant risk of morbidity and mortality associated with persistent bleeding. Reporting the need for additional rescue interventions would provide an additional marker of haemostatic efficacy, and future trials should consider incorporating this outcome.

Quality of the evidence

The Risk of bias in included studies section provides a summary of the identified biases. Generally, most studies were at high risk of performance and detection bias due to lack of blinding. This is unsurprising, as the outcome assessor (usually the operating surgeon) administered the intervention. The trials demonstrated significant heterogeneity in reporting of primary outcomes, likely owing to the subjectivity of assessing haemostatic efficacy. Only two trials did not report perioperative mortality (Chu 2008; Noun 1996); therefore, we have very little confidence in the effect estimate and the true effect is likely to be substantially different from the estimated effect. No trials reported all our outcomes of interest, and there was variation across trials in the methods of reporting some outcomes, meaning not all trials could be included in the corresponding meta‐analyses. Additionally, inconsistent reporting of some important factors that may affect bleeding (e.g. cirrhosis, neoadjuvant chemotherapy, method of parenchymal transection, pringle time) may have confounded our results.

Potential biases in the review process

We attempted to limit biases at every stage of our review. We performed a systematic search for trials, and two review authors independently screened trials against our inclusion criteria. Two review authors independently extracted data from the included trials using a standardised data extraction form, resolving any discrepancies by consulting a third review author. We only aimed to perform the subgroup analyses prespecified in our protocol to limit bias from multiple comparisons. However, it is possible that we missed some relevant trials.

Agreements and disagreements with other studies or reviews

The most recent meta‐analysis investigating FBHA use for reducing blood loss in adult liver resection is Wells 2020. It reported that fibrin patch and fibrin glue were significantly more effective in achieving haemostasis at 10 minutes (the primary outcome) compared with no topical intervention and collagen‐based haemostatic agents. However, as in this review, Wells 2020 found that improved haemostasis efficacy did not translate to improved postoperative outcomes, as there was no difference between groups in need for transfusion, interventions for postoperative bleeding, or adverse events.

Authors' conclusions

Implications for practice.

We found that fibrin‐based haemostatic agents (FBHAs) compared with no intervention or non‐FBHAs may have little or no effect on clinical outcomes in adults undergoing liver resection, despite any improvement in time to haemostasis or likelihood of achieving haemostasis at four or five minutes after application. The evidence for almost all outcomes was of very low certainty. Our sensitivity analyses removing trials that excluded people with cirrhosis showed no benefit associated with FBHAs. These analyses were likely underpowered and confounded given the inconsistency in reporting background liver status and variable number of participants with cirrhosis. The financial implication of reduced time to haemostasis remains unclear, as no trials reported a cost‐benefit analysis. Given the lack of evidence for clinical benefit, we cannot justify routine use of FBHAs to reduce blood loss in adult liver resection.

Implications for research.

It is unclear whether routine use of FBHAs provides any clinical benefits for adults undergoing liver surgery. Future research should focus on use of FBHAs in people with a particularly high risk of bleeding (those with cirrhosis, previous neoadjuvant chemotherapy, intolerance of low central venous pressure anaesthesia, etc.). Given that published surveys demonstrate a preference amongst liver surgeons for routine use of FBHAs, pharmaceutical corporations are likely to continue to market new agents and back future trials (Lochan 2013; Nakajima 2002). These trials should collect data on background liver status, transection technique, and neoadjuvant chemotherapy use, and provide stricter definitions of achievement of haemostasis based on international standardised guidelines (Khorsand 2016).

History

Protocol first published: Issue 12, 2013

Appendices

Appendix 1. Surgical and anaesthetic techniques used to minimise intraoperative blood loss and maintain haemostasis in liver surgery*

Anaesthetic techniques

1) Lowering venous pressure

2) Autologous transfusion

3) Antifibrinolytic agents

Surgical techniques

1) Anatomical dissection

2) Hepatic inflow occlusion

3) Ischaemic preconditioning

4) Total vascular exclusion

5) In situ hypothermic perfusion

Surgical instruments

1) Cavitron ultrasonic surgical aspirator

2) Radiofrequency ablation needle dissection

3) Monopolar Dissecting Sealer (saline‐linked radiofrequency sealer; TissueLink Medical Inc.)

4) Water jet dissection (laminar liquid jet dissection)

5) Argon beam coagulation

6) Harmonic scalpel (vibration‐powered coagulation and cutting device; Ethicon Endo‐Surgery).

*Adapted from Berrevoet 2007.

Appendix 2. Search strategies

Database	Time span	Search strategy
The Cochrane Hepato‐Biliary Group Controlled Trials Register (via the Cochrane Register of Studies Web)	1946 to 20 January 2023	(((topical and (hemostatic or haemostatic)) or fibrin* or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard)) AND (((liver or hepatic) and (surgery or surgical or resect* or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy))
Cochrane Central Register of Controlled Trials in the Cochrane Library	2023, Issue 1	#1 MeSH descriptor: [Fibrinogen] explode all trees #2 MeSH descriptor: [Hemostatics] explode all trees #3 ((topical and (hemostatic or haemostatic)) or fibrin* or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard) #4 #1 or #2 or #3 #5 MeSH descriptor: [Hepatectomy] explode all trees #6 (((liver or hepatic) and (surgery or surgical or resect* or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy)) #7 #5 or #6 #8 #4 and #7
MEDLINE Ovid	1946 to 20 January 2023	1. exp Fibrinogen/ 2. exp Hemostatics/ 3. ((topical and (hemostatic or haemostatic)) or fibrin* or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard).mp. [mp=title, book title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 4. 1 or 2 or 3 5. HEPATECTOMY/ 6. (((liver or hepatic) and (surgery or surgical or resect* or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy)).mp. [mp=title, book title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 7. 5 or 6 8. 4 and 7 9. (randomized controlled trial or controlled clinical trial or retracted publication or retraction of publication).pt. 10. clinical trials as topic.sh. 11. (random* or placebo*).ab. or trial.ti. 12. 9 or 10 or 11 13. exp animals/ not humans.sh. 14. 12 not 13 15. 8 and 14
Embase Ovid	1946 to 20 January 2023	1. exp hemostatic agent/ 2. ((topical and (hemostatic or haemostatic)) or fibrin* or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 3. 1 or 2 4. exp liver resection/ 5. (((liver or hepatic) and (surgery or surgical or resect* or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy)).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 6. 4 or 5 7. 3 and 6 8. Randomized controlled trial/ or Controlled clinical study/ or randomization/ or intermethod comparison/ or double blind procedure/ or human experiment/ or retracted article/ 9. (random$ or placebo or parallel group$1 or crossover or cross over or assigned or allocated or volunteer or volunteers).ti,ab. 10. (compare or compared or comparison or trial).ti. 11. ((evaluated or evaluate or evaluating or assessed or assess) and (compare or compared or comparing or comparison)).ab. 12. (open adj label).ti,ab. 13. ((double or single or doubly or singly) adj (blind or blinded or blindly)).ti,ab. 14. ((assign$ or match or matched or allocation) adj5 (alternate or group$1 or intervention$1 or patient$1 or subject$1 or participant$1)).ti,ab. 15. (controlled adj7 (study or design or trial)).ti,ab. 16. (erratum or tombstone).pt. or yes.ne. 17. or/8‐16 18. (random$ adj sampl$ adj7 ('cross section$' or questionnaire$ or survey$ or database$1)).ti,ab. not (comparative study/ or controlled study/ or randomi?ed controlled.ti,ab. or randomly assigned.ti,ab.) 19. Cross‐sectional study/ not (randomized controlled trial/ or controlled clinical study/ or controlled study/ or randomi?ed controlled.ti,ab. or control group$1.ti,ab.) 20. (((case adj control$) and random$) not randomi?ed controlled).ti,ab. 21. (Systematic review not (trial or study)).ti. 22. (nonrandom$ not random$).ti,ab. 23. 'Random field$'.ti,ab. 24. (random cluster adj3 sampl$).ti,ab. 25. (review.ab. and review.pt.) not trial.ti. 26. 'we searched'.ab. and (review.ti. or review.pt.) 27. 'update review'.ab. 28. (databases adj4 searched).ab. 29. (rat or rats or mouse or mice or swine or porcine or murine or sheep or lambs or pigs or piglets or rabbit or rabbits or cat or cats or dog or dogs or cattle or bovine or monkey or monkeys or trout or marmoset$1).ti. and animal experiment/ 30. Animal experiment/ not (human experiment/ or human/) 31. or/18‐30 32. 17 not 31 33. 7 and 32
LILACS (Bireme)	1982 to 20 January 2023	((topical and (hemostatic or haemostatic)) or fibrin$ or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard) [Words] and (((liver or hepatic) and (surgery or surgical or resect$ or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy)) [Words]
Science Citation Index Expanded (Web of Science)	1900 to 20 January 2023	#5 #4 AND #3 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=(((liver or hepatic) and (surgery or surgical or resect* or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy)) #1 TS=((topical and (hemostatic or haemostatic)) or fibrin* or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard)
Conference Proceedings Citation Index – Science (Web of Science)	1990 to 20 January 2023	#5 #4 AND #3 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=(((liver or hepatic) and (surgery or surgical or resect* or segmentectomy or sectionectomy or trisectionectomy or lobectomy)) or (hepatectomy or hepatectomies or hemihepatectomy or hemi‐hepatectomy)) #1 TS=((topical and (hemostatic or haemostatic)) or fibrin* or TachoSil or TachoComb or Tisseel or Beriplast or Biocol or EviCel or Crosseal or Hemaseel or Quixil or Bolheal or ViGuard)

Data and analyses

Comparison 1. Fibrin‐based haemostatic agents (FBHAs) versus no intervention or placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Perioperative mortality	4	782	Risk Ratio (M‐H, Random, 95% CI)	2.58 [0.89, 7.44]
1.2 Serious adverse events	4	782	Risk Ratio (M‐H, Random, 95% CI)	0.96 [0.88, 1.05]
1.3 Haemostasis within a predefined time period	2	172	Risk Ratio (M‐H, Random, 95% CI)	2.12 [1.65, 2.73]
1.4 Volume of intraoperative blood loss (mL)	3	712	Mean Difference (IV, Random, 95% CI)	101.26 [15.81, 186.71]
1.5 Postoperative transfusion	5	864	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.77, 1.40]
1.6 Reoperation	2	612	Risk Ratio (M‐H, Random, 95% CI)	2.92 [0.58, 14.61]
1.7 Postoperative bile leak	4	782	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.67, 1.48]
1.8 Postoperative intra‐abdominal collections	6	906	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.81, 1.28]
1.9 Postoperative intra‐abdominal collections: sensitivity analysis	5	859	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.83, 1.32]
1.10 Abdominal drain output (mL)	3	447	Mean Difference (IV, Random, 95% CI)	‐71.20 [‐423.54, 281.14]
1.11 Time to removal of abdominal drains (days)	4	822	Mean Difference (IV, Random, 95% CI)	‐0.53 [‐2.03, 0.97]
1.12 Operating time (minutes)	3	479	Mean Difference (IV, Random, 95% CI)	14.86 [‐0.34, 30.06]
1.13 Length of hospital stay (days)	3	479	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐1.63, 1.33]

Comparison 2. Fibrin‐based haemostatic agents (FBHAs) versus non‐FBHAs.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Perioperative mortality	11	1436	Risk Ratio (M‐H, Random, 95% CI)	1.03 [0.62, 1.72]
2.2 Perioperative mortality: sensitivity analysis removing trials that excluded participants with cirrhosis	8	1142	Risk Ratio (M‐H, Random, 95% CI)	1.03 [0.58, 1.80]
2.3 Perioperative mortality: sensitivity analysis removing a trial that investigated a non‐commercial FBHA	10	1378	Risk Ratio (M‐H, Random, 95% CI)	0.99 [0.58, 1.70]
2.4 Serious adverse events	9	1176	Risk Ratio (M‐H, Random, 95% CI)	0.99 [0.95, 1.03]
2.5 Serious adverse events: sensitivity analysis removing trials that excluded participants with cirrhosis	6	882	Risk Ratio (M‐H, Random, 95% CI)	0.99 [0.94, 1.03]
2.6 Time to haemostasis (minutes)	8	1093	Mean Difference (IV, Random, 95% CI)	‐1.58 [‐2.68, ‐0.49]
2.7 Time to haemostasis (minutes): sensitivity analysis removing trials that excluded participants with cirrhosis	5	799	Mean Difference (IV, Random, 95% CI)	‐1.77 [‐2.54, ‐1.01]
2.8 Time to haemostasis (minutes): sensitivity analysis removing trials with extreme outliers	7	967	Mean Difference (IV, Random, 95% CI)	‐1.86 [‐2.52, ‐1.21]
2.9 Haemostasis within a predefined time period	7	789	Risk Ratio (M‐H, Random, 95% CI)	1.30 [1.04, 1.62]
2.10 Haemostasis within a predefined time period: sensitivity analysis removing trials that excluded participants with cirrhosis	5	614	Risk Ratio (M‐H, Random, 95% CI)	1.54 [1.22, 1.94]
2.11 Haemostasis within a predefined time period: sensitivity analysis removing trials with extreme outliers	5	635	Risk Ratio (M‐H, Random, 95% CI)	1.11 [0.94, 1.31]
2.12 Volume of intraoperative blood loss (mL)	3	176	Mean Difference (IV, Random, 95% CI)	111.24 [‐102.96, 325.45]
2.13 Postoperative transfusion	7	599	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.68, 1.25]
2.14 Postoperative transfusion: sensitivity analysis removing trials that excluded participants with cirrhosis	5	423	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.56, 1.12]
2.15 Reoperation	3	358	Risk Ratio (M‐H, Random, 95% CI)	0.48 [0.25, 0.90]
2.16 Reoperation: sensitivity analysis removing trials with extreme outliers	2	242	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.12, 3.22]
2.17 Postoperative bile leak	9	1115	Risk Ratio (M‐H, Random, 95% CI)	1.15 [0.60, 2.21]
2.18 Postoperative bile leak: sensitivity analysis removing trials that excluded participants with cirrhosis	7	939	Risk Ratio (M‐H, Random, 95% CI)	1.19 [0.58, 2.44]
2.19 Postoperative intra‐abdominal collections	6	664	Risk Ratio (M‐H, Random, 95% CI)	0.72 [0.24, 2.21]
2.20 Abdominal drain output (mL)	3	296	Mean Difference (IV, Random, 95% CI)	‐212.49 [‐939.53, 514.55]
2.21 Abdominal drain output (mL): sensitivity analysis removing trials with extreme outliers	2	177	Mean Difference (IV, Random, 95% CI)	138.53 [66.51, 210.55]
2.22 Time to removal of abdominal drains (days)	4	332	Mean Difference (IV, Random, 95% CI)	0.07 [‐1.39, 1.53]
2.23 Operating time (minutes)	4	419	Mean Difference (IV, Random, 95% CI)	‐1.27 [‐22.73, 20.19]
2.24 Length of hospital stay (days)	5	503	Mean Difference (IV, Random, 95% CI)	‐0.20 [‐1.36, 0.96]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Bektas 2014.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 30 +/‐ 10 days from date of surgery
Participants	Setting: Germany Number randomised: 70
Interventions	Intervention: fibrin‐based sealant (n = 35) Per subject, 2 10‐mL packages of fibrin sealant were thawed (at 33 °C to 37 °C) prior to randomisation. The frozen solutions of 5 mL of fibrinogen with synthetic aprotinin and 5 mL of thrombin (500 IU/mL) came in 2 prefilled syringes mounted into a syringe clip. If not assigned, the product was discarded. Fibrin sealant was applied using routine procedures: a thin layer was sprayed from a distance of 10–15 cm onto the entire liver resection surface where oozing was present. No carrier material and no compression of the FS layer were permitted. Temporary vascular clamping was released after 3 minutes. This period allowed for application and polymerisation of the fibrin sealant. Repeated spray administration (in case of incomplete coverage) was permitted only prior to the initial haemostasis evaluation at 4 minutes. Thereafter, no further manipulation was allowed until the end of the 10‐minute observation period. Control: manual compression alone (n = 35) Manual compression was performed according to standard procedures: a dry surgical gauze swab was used to manually apply an even light pressure onto the oozing resection surface. Temporary vascular clamping was released as soon as manual compression was in place. The gauze flap was lifted to check for haemostasis at 4 minutes, 6 minutes, 8 minutes, and 10 minutes after initial positioning and removed after haemostasis had been achieved. If any bleeding was observed when lifting the swab, manual compression was reapplied until the next assessment.
Outcomes	Primary outcomes Percentage of participants with haemostasis at 4 minutes postapplication Incidence of adverse events Secondary outcomes Percentage with haemostasis at 6 minutes, 8 minutes, and 10 minutes Intraoperative bleeding Postoperative bleeding Transfusion requirement Drain volume
Notes	6 participants (3 from each group) withdrew consent after randomisation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Predetermined randomisation scheme stratified by study centre using the random number generator algorithm of Wichmann and Hill.
Allocation concealment (selection bias)	Low risk	The randomisation envelope number was obtained from an electronic data capture system. The randomisation envelope assigned was opened in the operating room after confirmation of the intraoperative eligibility criteria.
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding.
Blinding of outcome assessment (detection bias) All outcomes	High risk	No blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data reported for all participants; reasons for exclusion provided.
Selective reporting (reporting bias)	Low risk	Outcomes clearly reported.
Other bias	Low risk	No other bias identified.

Bjelović 2018.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 90 days from date of surgery
Participants	Setting: multicentre (Hungary, Serbia, Spain, USA) Number randomised: 325
Interventions	Intervention: Fibrin Sealant Grifols (n = 163) Fibrin Sealant Grifols is composed of frozen solutions of highly purified human fibrinogen (80 mg/mL) and human thrombin (500 IU/mL) with calcium chloride (5.9 mg/mL) supplied as a ready‐to‐use kit, in preloaded glass syringes assembled on a syringe holder. For the specific use on the surface of hepatic resections, the fibrin sealant was delivered using spray applicator tips (Gas Assisted Applicator Kit, 5 cm3. Micromedics, Inc. Eagan, MN, USA). Grifols fibrin sealant was applied onto the target bleeding site surface by spraying (10 cm distance with a 1–1.75 bar pressure) in short bursts (0.1–0.2 mL). The maximum total volume of Grifols fibrin sealant was 12 mL (2 kits). Control: oxidised cellulose sheets (Surgicel; n = 162)
Outcomes	Primary outcome: proportion of participants who achieved haemostasis at the target bleeding site by 4 minutes without rebleeding until completion of surgical closure. Rebleeding was defined as bleeding from the target bleeding site requiring a further haemostatic intervention after haemostasis was previously achieved. Secondary outcomes Time to haemostasis in minutes Cumulative proportion of participants achieving haemostasis at the target bleeding site after 2 minutes, 3 minutes, 4 minutes, 5 minutes, 7 minutes, 10 minutes Prevalence of treatment failures (persistent bleeding beyond 4 minutes, rebleeding before completion of surgical closure, brisk and forceful bleeding that jeopardised participant safety, or requirement of alternative haemostatic treatments of manoeuvre during the 10‐minute observation period)
Notes	9 participants discontinued the study (Fibrin Sealant Grifols n = 5, Surgicel n = 4), with reasons not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation software.
Allocation concealment (selection bias)	Low risk	Randomisation software.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded to outcome.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded to outcome.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Adequate follow‐up and outcome data provided.
Selective reporting (reporting bias)	Low risk	All relevant outcomes were reported, and reasons for exclusion stated.
Other bias	Low risk	No other bias identified.

Bochicchio 2015.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 29 days from date of surgery
Participants	Setting: international multicentre trial Number randomised: 721 (including people undergoing non‐hepatic surgery (spinal, vascular and soft tissue); 181 participants underwent hepatic resection (wedge or anatomic)
Interventions	Intervention: Fibrocaps (n = 482; 120 underwent liver resection) Sprinkled directly from the vial onto the bleeding site followed by application of a gelatin sponge; or sprinkled onto a moist gelatin sponge and applied to the bleeding site. Light manual pressure applied with sterile gauze. Gelatin sponges could be replaced if necessary. Control: gelatin sponge alone (n = 230; 61 underwent liver resection) Sponge applied to the target bleeding site. Light manual pressure applied with sterile gauze. Gelatin sponges could be replaced if necessary.
Outcomes	Primary outcome: time to haemostasis (within 5 minutes) Secondary outcomes Mean time to haemostasis Probability of haemostasis at 3 minutes Probability of haemostasis at 5 minutes Use of alternative haemostatic agent Reoperation Transfusion requirement at 29 days
Notes	5 participants discontinued the study, distribution across groups and reasons for discontinuation were not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear how participants were allocated and whether blocks were used.
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants followed up, with reasons for exclusion provided.
Selective reporting (reporting bias)	Unclear risk	All relevant outcomes reported.
Other bias	Low risk	No other bias identified.

Chu 2008.

Study characteristics
Methods	Study design: randomised trial Follow‐up: not specified
Participants	Setting: single‐centre trial in Sydney, Australia Number randomised: 142
Interventions	Intervention: Tisseel Control: no intervention
Outcomes	Primary outcomes Duration of drainage Volume of fluid collected Incidence of bile leak
Notes	Number of participants in each group not specified Published in abstract form only
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information to make a judgement.
Allocation concealment (selection bias)	Unclear risk	Insufficient information to make a judgement.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient information to make a judgement.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to make a judgement.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient information to make a judgement.
Selective reporting (reporting bias)	Unclear risk	Insufficient information to make a judgement.
Other bias	Unclear risk	Insufficient information to make a judgement.

de Boer 2012.

Study characteristics
Methods	Study design: randomised trial Follow‐up: minimum 30‐days following discharge from hospital
Participants	Setting: multicentre study in the Netherlands Number randomised: 310
Interventions	Intervention: fibrin‐based sealant (n = 156) 10 mL fibrin sealant was sprayed on the resection surface and on the bare surface of the diaphragm (after haemostasis and biliostasis at the liver resection surface by conventional techniques such as suture, clip application, or coagulation). All local investigators were instructed in the use of fibrin sealant by the company before trial commencement. Control: no intervention (n = 154) No further intervention following haemostasis and biliostasis at the liver resection surface by conventional techniques such as suture, clip application, or coagulation
Outcomes	Primary outcome: resection surface‐related complications, defined as a composite endpoint of bile leakage, bleeding, or abscess Secondary outcomes Drain fluid haemoglobin on day 3 Drain fluid bilirubin on day 3 Postoperative complications Mortality
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer random‐number generator.
Allocation concealment (selection bias)	Low risk	Sequentially numbered opaque and sealed envelope system.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants, local investigators responsible for data gathering, data analysts, and radiologists were unaware of the study group assignment.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Participants, local investigators responsible for data gathering, data analysts, and radiologists were unaware of the study group assignment
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants accounted for. Missing data were treated as missing according to a list‐wise deletion approach.
Selective reporting (reporting bias)	Low risk	Most outcomes in protocol reported (except length of stay).
Other bias	Low risk	No other bias identified.

Figueras 2007.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 6‐months after surgery
Participants	Setting: single‐centre study in Spain Number randomised: 300 participants
Interventions	Intervention: fibrin glue (n = 150) 5 mL of Tissucol (Baxter‐Inmuno, Vienna, Austria) was applied in aerosol form on the raw surface of the liver. An absorbable collagen sponge (Johnson & Johnson) was also applied with manual pressure, after spraying the fibrin glue. Control: no intervention (n = 150) Neither the fibrin sealant nor the collagen sponge was used.
Outcomes	Primary outcomes Mean total blood loss Percentage of participants with intra‐ or postoperative transfusion Mean number packed red cells transfused Percentage with intraoperative fresh frozen plasma transfusion Percentage with intraoperative platelet transfusion Secondary outcomes Mean number of days postoperative drainage Mean overall drain volume Percentage of reoperations Percentage of biliary fistula, abdominal abscess or abdominal bleeding Percentage requiring percutaneous drainage of collection Mortality
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation process not specified.
Allocation concealment (selection bias)	Unclear risk	Method of concealment, if used, not specified.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel not specified.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of participants and personnel not specified.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Randomised patients analysed. No dropouts/exclusions after randomisation reported.
Selective reporting (reporting bias)	Unclear risk	No protocol published.
Other bias	Low risk	Funding from academic source, with no commercial interest. No other bias identified.

Fischer 2011.

Study characteristics
Methods	Study design: randomised trial Follow‐up: not reported
Participants	Setting: European multicentre trial (10 centres) Number randomised: 121 (2 participants excluded postrandomisation)
Interventions	Intervention: TachoSil (n = 60) TachoSil, a sterile, ready‐to‐ use, absorbable surgical patch consisting of an equine collagen sponge coated with human fibrinogen and human thrombin measuring 9.5 x 4.8 x 0.5 cm, was applied on the resection surfaces after being moistened with physiological saline. The yellow‐coated side (active side) of the patch was held against the resection surface for 3 minutes to ensure uniform contact. The resection site(s) had to be covered at least 1 cm beyond its margin, and if > 1 patch was needed, they had to overlap. TachoSil treatment could be repeated after 5 to 8 minutes. Control: argon beam coagulator (ABC: n = 59) Management of bleeding with the ABC was done according to the routine of the centres. All hepatic resection areas were treated, and treatment was repeated if haemostasis was not achieved within 5 minutes. The duration and number of treatments were recorded. Co‐interventions: escape treatment was applied if haemostasis was not obtained within 10 minutes after the application of study treatment. No specific escape treatment was defined, but was chosen according to local protocols
Outcomes	Primary outcome: time to haemostasis Secondary outcomes Total drainage volume Total postoperative duration of drainage Drain fluid haemoglobin and bilirubin Measurement (with ultrasound) volume of blood/liquid at resection surface
Notes	2 participants were excluded postrandomisation due to incorrect randomisation. 6 participants discontinued the study due to adverse events (2 in TachoSil group, 4 in ABC group).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Centralised telephone randomisation system.
Allocation concealment (selection bias)	Low risk	Centralised telephone randomisation system.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Assessors not blinded to allocation.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	No follow‐up protocol specified.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Low risk	No other bias identified.

Franceschi 2006.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 30 days postsurgery Compared CryoSeal with a collagen‐based haemostat (CBH)
Participants	Setting: multicentre study in the USA Number randomised: 153
Interventions	Intervention: CryoSeal (n not reported) CryoSeal FS System consists of a medical device (the CS‐1 instrument) and a proprietary plasma processing disposable that work in concert to rapidly (approximately 60 minutes) prepare both components of a fibrin sealant (cryoprecipitate and thrombin), from a single unit of autologous. When the cryoprecipitate and the thrombin produced by the CryoSeal FS System are mixed with the supplied applicator, a fibrin sealant is obtained Control: topical application of a collagen based haemostat (n not reported)
Outcomes	Primary outcomes Time to haemostasis Haemostasis at 10 minutes Complications
Notes	Published in abstract form only.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not specified.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not specified.
Selective reporting (reporting bias)	Unclear risk	Not specified.
Other bias	Unclear risk	Insufficient information in abstract to make a judgement on the presence of other biases. Funded by commercial entity.

Frilling 2005.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 30 days postsurgery
Participants	Setting: multicentre study in 6 European cities Number randomised: 121
Interventions	Intervention: TachoSil (n = 59, 6 dropouts) TachoSil was applied under aseptic conditions. Before use, TachoSil was moistened with physiological saline. Application to the sites was done within 1 minute after moistening and after blood and other fluids had been cleansed from the site. The yellow‐coated side of the patch was held (either with a moistened surgical glove or a moist pad) against the resection surface for 3 minutes to assure uniform contact. It was recommended that blood and secreted fluids were removed from surgical gloves and instruments before application of TachoSil to avoid undesired adhesion to the patch. The resection site had to be covered at least 1 cm beyond its margins; if > 1 patch was needed, they had to overlap. Control: argon beam coagulation (ABC; n = 62, 3 dropouts) Management of haemorrhagic sites in ABC group was done in accordance with the routine of the centre. All hepatic resection areas were treated, and treatment was repeated in case haemostasis was not obtained after 5 minutes.
Outcomes	Primary outcome: time to haemostasis Secondary outcomes Proportion of patients with haemostasis at 10 minutes Volume of drainage at day 1 and day 2 Haemoglobin concentration of drain fluid Total duration of drain fluid
Notes	Reasons for dropouts were not specified.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All patients followed up to 30 days.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Unclear risk	Funding source not specified.

Genyk 2016.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 6 months postsurgery
Participants	Setting: multicentre study in 19 sites in the USA Number randomised: 224
Interventions	Intervention: TachoSil (n = 114, 13 excluded from analysis) Control: Surgicel (n =110, 11 excluded from analysis)
Outcomes	Primary outcome: haemostasis at 3 minutes Secondary outcomes Haemostasis at 5 minutes Time to haemostasis
Notes	101 participants randomised to Tachosil were compared with 99 patients randomised to Surgicel Reasons for exclusion from Tachosil group were as follows. Mortality (n = 4) Withdrawal of consent (n = 3) Loss to follow‐up (n = 4) Not specified (n = 2) Reasons for exclusion from Surgicel group were as follows. Mortality (n = 6) Loss to follow‐up (n = 2) Withdrawal of consent (n = 1) Not specified (n = 2)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Centralised interactive voice or web response system used for intraoperative randomisation.
Allocation concealment (selection bias)	Low risk	Centralised interactive voice or web response system used for intraoperative randomisation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	No obvious attempt to blind investigators other than operator. Achievement of haemostasis can be subjective.
Blinding of outcome assessment (detection bias) All outcomes	High risk	No obvious attempt to blind investigators other than operator. Achievement of haemostasis can be subjective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition fully accounted for.
Selective reporting (reporting bias)	Low risk	Trial protocol published. Comparable to outcomes in published report.
Other bias	Low risk	No other bias identified.

Gugenheim 2011.

Study characteristics
Methods	Study design: randomised trial Follow‐up: unclear
Participants	Setting: single‐centre study in France Number randomised: 58
Interventions	Intervention: fibrin glue (n = 29) Application of 5 mL of fibrin glue (Tissucol‐Laboratoire Baxter SA, Maurepas, France) in the aerosol form on the raw surface of the liver. A sufficient amount of fibrin glue was scattered on the cut surface of the liver until a thin fibrin layer covered the whole cut surface. Control: PlasmaJet (n = 29) The whole cut surface of the liver was evenly coagulated with the Plasma Jet (Plasma Surgical Limited, Theale, UK) set at 40% power, keeping the probe at a 5 mm distance and at an angle of 45 degrees from the tissue. If the surgeon considered haemostasis as insufficient, Plasma Jet was raised to 100% power to complete haemostasis.
Outcomes	Primary outcomes Postoperative mortality Postoperative transfusions Reoperation incidence Incidence of patients requiring percutaneous drainage of collection
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Low risk	Random assignment by opening an envelope in which the allotted treatment was hidden.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Attempts to blind participants and personnel not specified in report.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Attempts to blind participants and personnel not specified in report.
Incomplete outcome data (attrition bias) All outcomes	High risk	Follow‐up protocol not specified.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Unclear risk	Funding not specified.

Kakaei 2013.

Study characteristics
Methods	Study design: 3‐arm randomised trial Follow‐up: duration of inpatient stay
Participants	Setting: single‐centre study in Tabriz, Iran Number randomised: 45
Interventions	Intervention: Tachosil (n = 15) TachoSil, a sterile, ready‐to‐use, absorbable surgical patch consisting of an equine collagen sponge coated with human fibrinogen and human thrombin measuring 9.5 × 4.8 × 0.5 cm, was applied on the resection surfaces after being moistened with physiological saline. Control 1: Surgicel (n = 15) 10 × 10 cm Control 2: Glubran 2 (n = 15) Glubran 2 in the aerosol form was used (1 package of 1 mL Glubran 2 for each participant).
Outcomes	Primary outcomes Time to haemostasis Bleeding Bile leak Wound infection Secondary outcomes Total drain volume Total postoperative duration of drainage Total volume of transfused blood products
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Online calculator.
Allocation concealment (selection bias)	Unclear risk	Not specified in report.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants completed follow‐up.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Unclear risk	Funding not specified.

Koea 2013.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 60‐days
Participants	Setting: multicentre study with sites in UK, Europe, New Zealand, and Australia Number randomised: 84
Interventions	Intervention: fibrin pad (n = 39) Control: standard of care (n = 45)
Outcomes	Primary outcome: haemostasis at 4 minutes Secondary outcomes Haemostasis at 10 minutes Transfusion requirements Hospital stay Bile leak Blood loss
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation with sealed envelopes.
Allocation concealment (selection bias)	Low risk	Allocation concealed in sealed envelope.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants adhered to follow‐up protocol.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Low risk	No other bias identified.

Koea 2016.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 60‐days
Participants	Setting: international multicentre study (USA, UK, Europe, Australia, New Zealand) Number randomised: 102
Interventions	Intervention: fibrin sealant patch (n = 52) Control: standard of care (n = 52)
Outcomes	Primary outcome: haemostasis at 4 minutes Secondary outcomes Haemostasis at 10 minutes Bile leak Transfusion requirements Hospital stay Blood loss
Notes	5 participants discontinued the study for the following reasons Fibrin sealant patch Lost to follow‐up (n = 1) Died before study completion (n = 1) Standard of care Lost to follow‐up (n = 1) Died before study completion (n = 1) Withdrew consent (n = 1)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient detail to make a judgement.
Allocation concealment (selection bias)	Unclear risk	Insufficient detail to make a judgement.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants adhered to follow‐up protocol.
Selective reporting (reporting bias)	Low risk	All relevant outcome measures reported.
Other bias	Low risk	No other bias identified.

Kohno 1992.

Study characteristics
Methods	Study design: randomised trial Follow‐up: not reported
Participants	Setting: single‐centre study in Japan Number randomised: 62
Interventions	Intervention: fibrin glue (Beriplast P; n = 31) Beriplast P also contains factor XIII and aprotinin Control: microcrystalline collagen powder (Avitene; n = 31)
Outcomes	Primary outcomes Achieving dry cut surface Postoperative rebleeding Bile leak Mortality Complications Secondary outcomes Drain output Duration of drainage
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient detail to make a judgement.
Allocation concealment (selection bias)	Low risk	Sealed envelope used to conceal allocations.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient detail to make a judgement.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient detail to make a judgement.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Follow‐up protocol described; excluded participants accounted for with reasoning explained.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Unclear risk	Funding not specified.

Moench 2014.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 3‐months postsurgery
Participants	Setting: multicentre study with sites in Germany and Austria Number randomised: 128
Interventions	Intervention: TachoSil (n = 65) Control: Sangustop (collagen‐based haemostatic agent; n = 63)
Outcomes	Primary outcome: haemostasis within 3 minutes Secondary outcomes Haemostasis at 5 minutes Haemostasis at 10 minutes Time to haemostasis Complications
Notes	2 participants excluded postrandomisation from Sangustop group: 1 for protocol violation, and 1 because they did not receive the intervention.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation with identical sealed opaque envelopes.
Allocation concealment (selection bias)	Low risk	Randomisation with identical sealed opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of participants, but no blinding of personnel.
Blinding of outcome assessment (detection bias) All outcomes	High risk	No blinding of personnel.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants followed up, and exclusions explained.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Low risk	No other bias identified.

Noun 1996.

Study characteristics
Methods	Study design: randomised trial Follow‐up: until discharge from hospital
Participants	Setting: single‐centre study in France Number randomised: 82
Interventions	Intervention: fibrin glue (n = 38) Control: no intervention (n = 44)
Outcomes	Primary outcomes Appearance of liver margin at closure Volume of fluid drained Haemoglobin content of fluid drained Bilirubin content of fluid drained Incidence of abdominal collections
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient detail in report to make a judgement.
Allocation concealment (selection bias)	Unclear risk	Insufficient detail in report to make a judgement.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Insufficient information to make a judgement.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Operator not blinded for assessment of resection margin.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants randomised and accounted for in report.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Unclear risk	Funding not specified.

Scheurer 2016.

Study characteristics
Methods	Study design: randomised trial Follow‐up: not specified
Participants	Setting: single‐centre study in Denmark Number randomised: 100
Interventions	Intervention: Tachosil (n = 50) Control: Hemopatch (n = 50)
Outcomes	Primary outcome: composite of 5 unspecified quality indicators. If the operating surgeon determined that any of the indicators was 'unsatisfactory', the haemostat was categorised as a failure.
Notes	Published in abstract form only.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient detail in abstract.
Allocation concealment (selection bias)	High risk	Participants and personnel not blinded.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient detail in abstract.
Blinding of outcome assessment (detection bias) All outcomes	High risk	No blinding of operators who assessed outcome.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient detail in abstract.
Selective reporting (reporting bias)	Low risk	Primary outcome reported, albeit with minimal detail.
Other bias	Unclear risk	Insufficient detail in abstract. Funding not specified.

Schwartz 2004.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 6 weeks postsurgery (6 months in people with hepatitis C infection)
Participants	Setting: multicentre study (USA, UK) Number randomised: 121
Interventions	Intervention: Crosseal (n = 58, 3 excluded from final analyses due to protocol violation) Control: fibrin‐based haemostats (Actifoam, Avitene, Gelfoam, Oxycel, Surgicel, Thrombinar: n = 63, 2 excluded from final analyses due to protocol violation)
Outcomes	Primary outcomes Time to haemostasis Haemostasis at 10 minutes Secondary outcomes Volume of blood loss between the time of initial application of the haemostatic agent and closure of the abdomen Duration of postoperative bilious drainage between drain insertion and removal Occurrence of abdominal collections complications Total intraoperative blood loss Postoperative fluid loss Number of units of haemoglobin‐containing blood products (whole blood or packed red blood cells) transfused during the postoperative period Preoperative haemoglobin levels compared with haemoglobin levels 1 day postoperatively Minimum haemoglobin level in the postoperative period Total time of operative procedure Duration of drainage
Notes	Control was any FDA‐approved fibrin‐based haemostatic agent, which was used as standard of care by the operating surgeon.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation by external statistician.
Allocation concealment (selection bias)	Low risk	Sealed envelopes used to conceal allocation.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants, staff and data analysts were blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Operating surgeon assessed haemostatic efficacy and was not blinded to allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All patients followed up as per protocol.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Unclear risk	Funding not specified.

Troisi 2012.

Study characteristics
Methods	Study design: randomised trials Follow‐up: 30 days postsurgery
Participants	Setting: multicentre European study (Belgium, Germany, Austria) Number randomised: 50
Interventions	Intervention: Tachosil (n not specified) Control: Veriset (n not specified)
Outcomes	Primary outcome: time to haemostasis
Notes	Published in abstract form only.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient detail in abstract to form judgement.
Allocation concealment (selection bias)	Unclear risk	Insufficient detail in abstract to form judgement.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient detail in abstract to form judgement.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient detail in abstract to form judgement.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient detail in from abstract to form judgement.
Selective reporting (reporting bias)	Low risk	Primary outcome reported.
Other bias	Unclear risk	Insufficient detail in abstract to form judgement. Funding not specified.

Verhoef 2014.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 29 days postsurgery
Participants	Setting: multicentre study with sites in the USA and Netherlands Number randomised: 126
Interventions	Intervention: Fibrocaps (n = 86) Control: gelatine sponge (n = 40)
Outcomes	Primary outcomes Time to haemostasis Treatment emergent adverse events Secondary outcomes Haemostasis at 3 minutes Haemostasis at 5 minutes Haemostasis at 10 minutes Blood product requirements
Notes	3 participants in the fibrocaps group were lost to follow‐up.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient data in report to make a judgement.
Allocation concealment (selection bias)	Unclear risk	Insufficient data in report to make a judgement.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Participants and personnel not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants followed up as per protocol.
Selective reporting (reporting bias)	Low risk	All relevant outcomes reported.
Other bias	Low risk	No other bias identified.

Öllinger 2013.

Study characteristics
Methods	Study design: randomised trial Follow‐up: 30 days postsurgery
Participants	Setting: multicentre European study (sites in Austria, Germany, and Belgium) Number randomised: 50
Interventions	Intervention: Veriset (n = 32) Control: Tachosil (n = 18)
Outcomes	Primary outcomes Time to haemostasis Treatment emergent adverse events Secondary outcomes Haemostasis within 3 minutes Transfusion requirements Drain output volume Time to drain removal Length of hospital stay Length of ICU stay
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient detail in report to make a judgement.
Allocation concealment (selection bias)	Low risk	Sealed envelopes used.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Operator not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Operator not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants followed up as per protocol.
Selective reporting (reporting bias)	Low risk	All relevant outcome data reported.
Other bias	Low risk	No other bias identified.

ICU: intensive care unit: n: number of participants.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Kawasaki 2017	Direct comparison of 2 FBHAs.
Kobayashi 2016	Direct comparison of 2 FBHAs.
Liu 1993	Quasi‐randomised study with allocation based on researcher judgement.
Uetsuji 1994	FBHA not applied to cut liver surface.

FBHA: fibrin‐based haemostatic agent.

Differences between protocol and review

See Amer 2013 (review protocol).

Description of the condition

We added two references: one study addressing the potential morbidity and mortality risk associated with allogeneic blood transfusion (Kimura 2017), and one addressing morbidity and mortality risk associated with intraoperative bleeding in adult liver resection (Olthof 2019).

Description of the intervention

In the protocol, our reference to the previous Cochrane Review (published in 2009) exploring cardiopulmonary interventions to decrease blood loss and blood transfusion requirements was updated in 2012. This reference is Gurusamy 2012. We have also added a reference to another Cochrane Review exploring methods for reducing blood loss during liver resection (Moggia 2016).

How the intervention might work

Where we discuss the addition of an antifibrinolytic agent, we added a reference to a trial investigating the systemic administration of tranexamic acid in people undergoing liver resection (Karanicolas 2016). We also referenced a trial that reported a reduction in bile leakage following laparoscopic common bile duct exploration with topical application of a fibrin sealant (Zhang 2019).

Why is it important to do this review

We added two further references to more recent meta‐analyses that failed to show benefit with topical application of FBHAs in adults undergoing liver resection (Brustia 2016; Wells 2020).

Primary outcomes

Under haemostatic efficacy, we added 'haemostasis within a predefined time period'. This was because several trials did not report time to haemostasis in minutes, but instead reported the number of participants achieving haemostasis at a predefined time (commonly four or five minutes following application of the agent).

We moved 'health‐related quality of life' to our primary outcomes.

We reformulated some of the outcomes for greater clarity.

Subgroup analysis and investigation of heterogeneity

We only performed subgroup analyses where at least 10 trials contributed data to the outcome (Deeks 2021 [https://revman.cochrane.org/#/324711012516145145/dashboard/htmlView/1.27.10?revertEnabled=true#REF‐Deeks‐2021]).

In the protocol, we planned of subgroup analyses comparing trials at overall low risk of bias with trials at overall high risk of bias, but all trials were at high or unclear risk of bias in at least one domain.

In the protocol, we planned a subgroup analysis of trials that included people with cirrhosis versus trials that excluded people with cirrhosis. However, not all trials specified background liver status or inclusion/exclusion of people with cirrhosis, so in place of a subgroup analysis, we performed a sensitivity analysis by removing trials that excluded people with cirrhosis.

The protocol prespecified a subgroup analysis of major (three or more segments) versus minor liver resection. We were unable to do this, as all trials included people undergoing major and minor liver resection and did not report outcomes separately for each group.

We also planned to perform subgroup analyses of trials investigating a commercial FBHA versus trials investigating a non‐commercial FBHA, trials where factor XIII was added to the FBHA versus trials where FBHA was applied without factor XIII, and trials where an antifibrinolytic agent was added to the FBHA versus trials where the FBHA was applied without an antifibrinolytic agent. We were only able to perform the first subgroup analysis (commercial versus non‐commercial FBHAs) for perioperative mortality in comparison 2 (FBHAs versus non‐FBHAs). For all other outcomes, fewer than 10 trials contributed data.

Trial Sequential Analysis

In the protocol, we planned to perform Trial Sequential Analysis (TSA) for all outcomes, but in the review, we only performed TSA for our first primary outcome in both comparisons.

Contributions of authors

AKM wrote the initial review, performed the initial analyses, and assessed all studies for inclusion/exclusion.
AOA assessed all studies for inclusion/exclusion.
SJT provided advice on the analyses and writing of the review.
ERT provided advice on the analyses and writing of the review.
SAW provided advice on the analyses and writing of the review.
DMM provided advice on the analyses and writing of the review.
CW provided advice on the analyses and supervised the review.

All review authors approved the final version.

Sources of support

Internal sources

• Institute of Transplantation, The Freeman Hospital, Newcastle upon Tyne, UK

  Institute of Transplantation

External sources

• The Editorial Team Office, The Cochrane Hepato‐Biliary Group, Copenhagen, Denmark

  Help during the review development and running the editorial process.

Declarations of interest

AKM: none
AOA: none
SJT: none
ERT: none
SAW: none
DMM: none
CW: none

New