A retrospective study was conducted using electronic medical records of hospitalized patients from 2010 to 2017 at the Institute of Liver and Biliary Sciences (ILBS), New Delhi, India. The final diagnosis of LC was identified by International Classification of Diseases code K74.9 and further clinical and biochemical parameters as complications including sepsis, ascites, acute variceal bleeding, SBP, HE, HRS, HCC were extracted from discharge summary or taken from hospital information system. Each in-patient was allotted a date or year of admission defined as the first record of LC related admission. The subsequent admissions, if required by the patient in the same year, were considered as readmissions. If a patient required repeated admissions in a given time span, the patient was included in the study only once at index admission and only the baseline clinical profile was considered for reference evaluations. Patients with a minimum 1 year follow-up or mortality/liver transplant within 1 year were considered for inclusion into the current study. The demographic information, etiology, complications, Child-Pugh (CTP) and model for end-stage liver disease (MELD) scoring system were extracted. Out of the 22,222 admissions related to liver diseases from 2010 to 2017, 21,794 were related to LC cases, 11,067 were follow-up admissions/readmissions and 4,453 had less than 1 year of follow-up at ILBS and 1,136 cases had acute-on-chronic liver failure (ACLF), as defined by Asian Pacific Association for the Study of the Liver, were not considered in this study. Therefore, the final study cohort consisted of 5,138 admitted cirrhotic patients with observed follow-up of at least 1 year or death/liver transplantation within the first year of index presentation (Fig. 1).

As alcoholic hepatitis or any other acute liver insults presenting as ACLF, are known to have a very high disease severity and high short term mortality rates, it would be difficult to compare the disease severity and outcomes in comparison to the non-ACLF group (alcoholic cirrhosis, cirrhosis with HBV/HCV) requiring admission for reasons like acute kidney injury (AKI) or SBP or HE. Hence it was decided from the inception of the study design to avoid the confounding effects of the acute insult and hence ACLF was excluded.

The diagnosis of cirrhosis was made by clinical, biochemical, radiological, and/or histological parameters. All cirrhotics aged >18 years, with or without decompensation (ascites, variceal bleeding, HE, jaundice) requiring hospitalization due to any cause were evaluated for inclusion in the present study.

Alcohol related cirrhosis (ALC) was defined as LC occurring in the background of years of cirrhogenic alcohol consumption doses (>60 g/day for men and 40 g/day for women) in the absence of other causes of liver injury (HBV, HCV, etc.) as documented in the discharge summaries and hospital records. Non-ALC included cirrhosis due to chronic hepatitis B or C viral infection, non-alcoholic fatty liver disease, auto-immune liver disease and cholestatic liver disease. None of the patients in the non-ALC group had significant alcohol consumption (>20 g/day).

Standard of care as per the treating physician’s discretion was extended to every patient for the management of LC according to the etiology and clinical presentation. All the patients with SBP received secondary prophylaxis with norfloxacin, patients with low albumin and ascites received regular albumin infusions to maintain serum albumin levels. Large volume paracentesis was considered whenever clinically indicated. HCC screening with alfafeto-protein levels and ultrasound/triple phase computed tomography examinations were done in follow-up at every 6–12 months intervals. HCC once diagnosed, was managed by liver transplantation, trans-arterial chemo-embolization, radiofrequency ablations or palliative care, as per standard protocols. Variceal bleeding was managed with endotherapy and repeat sessions were considered every 3 weeks till eradication and further every 3 months for prevention of rebleeds. HE was managed with anti-coma measures, lactulose enemas and infections were managed according to the standard institutional protocols. Liver transplantation, mainly living donor liver transplantation, was considered in patients fulfilling the criteria and was done whenever feasible.

All the results were expressed as mean±standard deviation or percentage. Comparison of continuous variables was calculated by Student’s t-test and categorical variables were compared by Pearson’s chi-square test or Fisher’s exact test. Analysis of variance was used for comparing more than two variables for parametric test and Kruskal Wallis for non-parametric variables. Poisson regression with robust standard errors was used to determine incidence risk ratio (IRR) of in-hospital mortality with respect to year of admission (follow-up 1 year), age gender, etiology, and LC complications. Logistic regression analysis was used to measure odds ratio [OR] for alcohol etiology. Propensity risk score analysis was done using MELD and CTP score. The nearest available matching (1:1) was taken with a calliper width of 0.001 of the logit of the propensity score. All tests were 2-tailed and p value of less than 0.05 was considered as significant. Data was analysed using IBM SPSS Statistics for Windows (version 23.0; IBM Corp., Armonk, NY, USA).

The median time to hospitalization from index presentation/diagnosis of cirrhosis in the cohort of 5,138 LC patients was 2 years (range, 0.2–10). There were 4,085 males (79.5%; M:F, 3.88:1). The mean age at presentation was 48.8±14.6 years. Majority (77.3%) of the patients were ≤60 years old. At the index admission, patients had ascites (72.3%), HE (31.3%), AVB (27.0%), and acute kidney injury (24.1%). Overall, 4,355 (84.8%) of the total cirrhotic patients (5,138) had decompensated liver disease at index admission. Only 465 cases (9.1%) were receiving beta-blockers at the time of inclusion. Diabetes Mellitus (DM; 37.4%), obesity (11.6%) were the most common co-morbidities in the cohort. A total of 163 patients (3.2%) underwent liver transplant.

The commonest etiology was alcohol related liver disease (39.5%), followed by non-alcoholic fatty liver disease (18.2%) and hepatitis B viral infection (10.8%). Alcohol (48.5%) was the most common cause of liver disease in men, followed by nonalcoholic fatty liver disease (NAFLD; 16.3%), HBV (11.3%), and HCV (7.6%). In females, NAFLD (25.8%) was the most common etiology followed by HCV (22.9%). Most common etiology for LC was alcohol in young, with 45% of <40 years of age and 53.9% of 41–50 years of age. However, NAFLD was the commonest cause in elderly (31.8% in patients with >60 years of age). ALC presented at younger age, at the index hospitalization compared to other etiologies (alcohol, HBV, HCV, NASH, cryptogenic were 47.9±10.8, 48.1±18.6, 53.9±10.8, 58.1±9.1, 54.8±13.6 years, respectively). The co-morbidities like hypothyroidism (7.0% vs. 24.9%, P<0.001), obesity (11% vs. 13.8%, P=0.02), and diabetes (36% vs. 40.8%, P=0.01) were more prevalent in the females.

We divided etiologies of chronic liver diseases into two groups (Table 1); ALC (n=2,027; 39.5%) and non-ALC (n=3,111; 60.5%), the later included HBV, HCV, NASH, and cryptogenic etiology. The ALC group had male preponderance (97.7% vs. 67.7%, P<0.001) with greater proportion of patients between 40–50 years (36.2% vs. 20.2%, P<0.001) compared to non-ALC group. Additionally, ALC had higher liver related complications at baseline than non-ALC group (sepsis: 20.3% vs. 14.9%, P<0.001; ascites: 82.2% vs. 65.9%, P<0.001; HRS: 12.8% vs. 7.6%, P<0.001; SBP: 21.8% vs. 15.7%, P<0.001; HE: 41.0% vs. 25.0%, P<0.001; AVB: 32.0% vs. 23.7%, P<0.001) and acute kidney injury (30.5% vs. 19.6%, P<0.001) along with greater disease severity CTP (10.6±2.0 vs. 9.0±2.3, P<0.001) and higher MELD (21.49±8.47 vs. 16.85±7.79, P<0.001). Higher proportion of ALC had decompensated liver disease at index admission (1,886 [93%] of the 2,027) compared to the non-ALC (2,469 [79.4%] of the 3,111; P<0.001). Additionally, the one year mortality was higher in ALC (42.3% vs. 27.3%, P<0.001) compared to non-ALC.

Patients in the ALC group had higher odds of being male (OR, 20.13; 95% confidence interval [CI], 14.9–27.2; P<0.001), belonging to younger age group 40–50 (OR, 4.04; 95%, CI, 3.39–4.80; P<0.001), more liver related complications at index admission; ascites (OR, 2.39; 95% CI, 2.08–2.74; P<0.001), AKI (OR, 2.09; 95% CI, 1.85–2.35; P<0.001), HE (OR, 2.39; 95% CI, 2.08–2.74; P<0.001). We found that prevalence of obesity (OR, 0.803; 95% CI, 0.67–0.96; P=0.02) or DM (OR, 0.58; 95% CI, 0.52–0.66; P=0.001) to be lower in ALC (Table 1).

The MELD score of more than 19.0 at admission predicted one year mortality with a sensitivity of 72.5% and specificity 72.0%, and area under receiver operating curve (AUROC) of 0.79 (P<0.001). A CTP score of 10.5 or more had a sensitivity 67.4% and specificity of 77.0% in predicting one year mortality with an AUROC of 0.78 (P<0.001). Nearly 40.0% patients had a MELD score more than 19.0, and 36.9% had CTP score more than 10.5. Higher proportion of patients in the ALC had MELD >19 compared to non-ALC (53.2% vs. 29.5%, P<0.001). Similarly, CTP score of more than 10.5 was present in 50.6% of ALC patients compared to 26.8% of non-ALC patients (P<0.001).

By Poisson regression, IRR of each passing year was found to be associated with mortality (P<0.001) (Table 2). It was observed that the risk of mortality in 2017 had increased by 2.58 times (IRR, 2.58; 2.02–3.28; P<0.001) compared to 2010. This increase in the IRR of mortality from 2010 to 2017, was due to an increase in the number of patients with more advanced liver disease over time with higher CTP (8.6±2.2 in 2010 to 9.5±2.3 in 2017, P<0.001) and MELD (15.6±6.0 in 2010 to 18.4±8.3 in 2017, P<0.001) with more number of liver related complications (greater prevalence of sepsis, HRS, HE, etc.) in the subsequent years.

Elderly patients (>60 years) and males had higher risk of mortality by 28% (IRR, 1.21; 1.10–1.34, P<0.001) and 36% (IRR, 1.36; 1.20–1.54, P<0.001) respectively compared to <40 year of age and females. The highest independent risk of mortality was seen with ALC group (IRR, 1.9; 1.71–2.07, P<0.001), those with sepsis (IRR, 4.5; 4.07–4.97, P<0.001) followed by those presenting with HE (IRR, 3.8; 3.45–4.19; P<0.001), HRS (IRR, 2.58; 2.3–2.9; P<0.001), ascites (IRR, 2.4; 2.1–2.7; P<0.001), SBP (IRR, 1.92; 1.7–2.1; P<0.001) and acute variceal bleeding (IRR, 1.30; 1.2–1.5; P<0.001). However, lower relative risk (RR) of 1 year mortality (P<0.001) was observed in cirrhotics requiring admission for HCC.

The overall one year mortality in ALC cohort was 42.3% in comparison to 27.3% in the non-ALC cohort. The trend for one year mortality had an initial increase from 2010 to 2012 (25.8% to 50.9%, P<0.001) and then remained nearly uniform from 2013 to 2017 (around 42.6%) which correlates with the increase in the share of alcohol related cirrhotics from 2010–2012 (30.3% to 46.3%, P<0.001) and then only marginally (46.3% to 42.6%, P=0.59) till 2017. Additionally, it was seen over the years that the liver disease severity as measured by mean MELD remained higher in the ALC compared to non-ALC group over the years along with higher 1 year mortality (Fig. 2).

Out of 5,138 patients admitted for LC, 1,707 (33.2%) died within a year. Of them, 1,248 (24.3%) died in the first hospitalization itself. The remaining 459 patients (8.9%) died during the follow-up period of 1 year. More than half (59.5%; 2,316/3,890) of the patients surviving the first admission, required at least one readmission during the subsequent 1 year, with an additional mortality of 19.8% (459/2,316) in those requiring readmission. The remaining 1,574 (40.5%) of 3,890 patients, surviving the first admission, remained on regular follow-up visits without need for in-patient care (Table 3).

Half of the patients who died either in the first admission or in the first year of follow-up, had ALC, especially those with sepsis and other liver related complications (SBP, HE, and HRS) with higher liver disease severity (higher MELD and CTP scores), while those with HBV, HCV, and NASH related liver diseases and those with HCC had relatively better 1 year survival. Altogether, 112 (5.5%) of 2,027 alcoholic cirrhotic patients had alcohol relapse and most of them (109/112 [97.3%]) who relapsed, had decompensated liver disease and required readmission.

We determined the number of complications, at baseline and grouped them as; any one complication, any two complications, any three complication and so on as shown in Figure 3A. Out of total cases, 1,412 had only one complication, such as only ascites, sepsis, HRS, SBP, AVB, or HE. Total number of complications of more than or equal to five were merged for analysis due to smaller sample size in these groups. It was seen that as the number of complications increased, their corresponding mean MELD score also increased (Fig. 3A).

Cirrhotic patients with ascites, HE and a third complication (any of the sepsis, HRS, AVB, or SBP) at index admission (i.e., complication ≥3 vs. complication <3) are unlikely to have good 1 year survival (39.8% vs. 76.3%, P<0.001). Figure 3B depicts the 1 year follow-up survival with ALC compared to non-ALC. Kaplan-Meier curve has shown the probability of 1 year survival in ALC patients to be much lower than the non-ALC group (56.9% vs. 71.9%, P<0.05).

LC patients who required admission were less likely to survive 1 year if the liver disease was due to alcohol, the patient had a MELD of >20.9±6.9 (P<0.001), CTP >10.3±1.86 (P<0.001), if the patient had three or more liver related complications, especially with sepsis, HE or HRS at admission.

We matched MELD and CTP scores using Propensity Score matching, between ALC versus non-ALC groups to assess the etiology and disease severity independent risk on the outcomes. The nearest available matching (1:1) was used to construct the control group with a calliper width of 0.001 of the logit of the propensity score. After matching, a total of 2,282 patients, ALC (n=1,141) and non-ALC (n=1,141) were identified. In the matched cohort, there was no difference noted in the MELD and CTP scores (MELD: 19.4±7.8 vs. 19.3±7.7, P=0.70 and CTP: 9.89±1.98 vs. 9.91±2.02, P=0.75). ALC patients were more likely to be younger (OR, 3.75; 95% CI, 2.84–4.95; P<0.001) and also with more chances of liver related complications: AVB (OR, 1.34; 95% CI, 1.12–1.61; P=0.001), followed by, ascites (OR, 1.25; 95% CI, 1.01–1.55; P=0.04), and HE (OR, 1.18; 95% CI, 1.0–1.4; P=0.05) at index admission compared to the disease severity matched non-ALC group. ALC group also had higher readmission rate in the following year in comparison to non-ALC after CTP and MELD matching.