Systematic Review

The Role of Frailty in Advanced Chronic Liver Disease

Sarah P. Bowers; Paul N. Brennan; John F. Dillon


Aliment Pharmacol Ther. 2023;57(3):280-289. 

In This Article

Abstract and Introduction


Background: Frailty is a known predictor of outcome and mortality in patients undergoing liver transplantation. However, most patients remain unsuitable transplant candidates. It is not yet known if the assessment of frailty in non-transplant candidates can aid prognostication.

Aim: To collate and interrogate the various frailty tools presently used to predict mortality in the non-transplant cirrhosis setting.

Methods: A comprehensive review of MEDLINE and EMBASE databases for articles published from inception to March 2022 was undertaken, excluding those where patients underwent transplantation or had hepatocellular carcinoma.

Results: We identified 12 observational cohort studies, featuring 9 frailty indices. These were from various global healthcare settings and of fair or good quality. Most were objective tools utilising clinician-based assessments. All frailty scores predicted prognosis, with variability in the method of application, and utilisation in long- or short-term mortality. Three studies directly compared different indices in the same population. There was some evidence that simple tools could perform as well, if not better, than more complex, time-consuming scores.

Conclusions: Various frailty tools can reproducibly evaluate mortality in patients with cirrhosis who are ineligible for transplant. However, further prospective head-to-head comparative studies are needed. In addition to determining model utility, studies should focus on important relative considerations which may limit widespread implementation including, ease of use and limited resources, given the global disparity of liver care provision. These tools may positively identify specific patient cohorts at risk of impending deterioration, thereby stratifying those patients likely to benefit from early integration with palliative care.


Liver disease accounts for over 2 million deaths per year worldwide, with half of these being attributed to cirrhosis, which is the 11th most common cause of death.[1] In contrast to most other malignant and non-malignant conditions, deaths from liver disease have risen in the United Kingdom, particularly in those under 65.[2] Morbidity and mortality from liver disease disproportionately affect those of working age, with liver disease accounting for an estimated 62,000 years of working life lost in the UK alone.[3] Furthermore, cirrhosis impairs health-related quality of life including mental health and physical factors and reduces the ability to perform activities of daily living (ADL).[4–6]

Presently, there are no approved pharmacological therapies for reversing fibrosis which results in cirrhosis, with liver transplantation remaining the only curative option for those with end-stage cirrhosis or hepatocellular carcinoma (HCC). Unfortunately, many patients will be too sick or have advanced co-morbidities that would preclude transplant referral, while a significant proportion of those referred for transplant assessment are ineligible. Among those patients listed for transplantation, around 7% die annually while on the waiting list in the UK.[7] There is an inherent mismatch in those patients necessitating liver transplantation, compared to the available organs donated, in spite of the use of split grafting and living donor alternatives to traditional orthotopic liver transplant using cadaveric organs.

The current most commonly employed prognostication tools for stratification of liver disease severity are Child-Pugh and MELD; both initially designed to predict mortality in those undergoing surgical or radiological interventions for portal hypertension.[8–10] These tools were adapted into clinical practice for both identifying eligibility, and waitlist prioritisation for transplantation, as well as being used for general prognostication. However, there have been criticisms of these tools. Child-Pugh includes subjective markers (ascites and encephalopathy) which can lead to inter-rater variability.[11,12] Both tools utilise INR, which is poorly representative of true coagulopathy, and is subject to inter-laboratory variation within laboratory systems.[13,14] Whilst they both have been shown to have the predictive capability in prognosis, there is a large variance between them in prognosticating complex clinical circumstances, e.g. mortality in ICU and in those undergoing surgery.[11,15] Newer advances in the assessment of clinical biomarkers and their utility in prognosis remain unvalidated, while often necessitating, invasive testing methods and are limited to resource-rich settings.[16,17] There has been a number of studies assessing the prognostic ability of these tools, highlighting the uncertainty and limitations of those currently applied.[18]

This leads to a need to develop alternative approaches for which frailty has been considered a key composite. The concept of frailty was originally derived in a geriatric population, where it was explored as an age-related biologic syndrome of decreased reserve and resistance to stressors. Resultantly, frailty is characterised by cumulative declines across multiple physiological systems and increased vulnerability to adverse outcomes.[19,20] A recent systematic review revealed that there were no frailty indices specifically developed or validated for younger populations, however, there was evidence that some existing markers of frailty had predictive ability in this group.[21]

The role of frailty in liver disease has been established for decades, with the original Child score using 'nutritional status' before it was replaced with prothrombin time by Pugh.[8,9] The pathophysiology of frailty in liver disease remains complex and is increasingly appreciated as a multisystem sequela. Presently, there are a number of putative theories existing around decreasing synthetic liver function causing resultant multisystem dysfunction with relative immunocompromise, as well as endocrine, gut microbiome, skeletal and neuromuscular maladaptation.[22,23] An estimated 17%–43% of those with cirrhosis and 25% on liver transplant waiting lists are identified as frail.[22,24] Frailty has also repeatedly been demonstrated to be an independent predictor of hepatic decompensation, hospitalisation, transplant delisting and post-transplant complications.[22,25,26]

Indeed, much of the research around frailty assessment in cirrhosis has been undertaken in the context of transplantation. However, only a minority of patients are eligible for a liver transplant, owing primarily to ongoing alcohol or substance misuse, poor performance status or co-morbidities—all of which can contribute to frailty. In 2016–2017, 14,696 deaths related to liver disease were recorded in the United Kingdom;[3] during the same time interval, only 980 liver transplants were undertaken.[7]

Therefore, the vast majority of patients with end-stage liver disease are managed conservatively. Existing tools for predicting mortality in liver disease, namely, Child-Pugh and MELD, are aimed to identify patients for aggressive attempts at curative transplantation and fail to holistically assess the patient as a whole. They also do not necessarily relate to patient needs or quality of life thus leaving a gap and problem in identifying patients who could be treated with more palliative intent earlier in the process.[27,28] To date, there have been no identified reviews assessing the ability of existing measurements of frailty to predict prognosis in patients with end-stage liver disease, who are unsuitable for transplantation. This review aims to understand the role that existing frailty tools may have in prognosticating in such persons.