A Study of Impact of Fixed-Dose Albumin Infusion on Outcome in Patients With Cirrhosis and Infection: A Randomized Open-label Clinical Trial

Abstract

Background and aim

Antibiotics and albumin infusion constitute the standard of treatment in patients with decompensated cirrhosis who have spontaneous bacterial peritonitis (SBP). Recent studies have also shown that the use of albumin in patients with advanced liver disease who have infections other than SBP leads to the resolution of acute and chronic liver failure and prevents the development of nosocomial infections. The recommended dose of albumin for these patients is out of reach for many in resource-limited settings like India. The evidence for this recommendation is also scarce. This study aimed to assess the efficacy of a lower dose of albumin infusion in addition to antibiotics on short-term mortality and morbidity in patients with cirrhosis and infections.

Patients and methods

A prospective, open-label, randomized control study was performed. Consecutive patients with cirrhosis and infections were randomized in a 2:1 ratio into two groups: group A (116) and group B (58) patients. In addition to antibiotics and standard medical therapy, group A was given albumin in a dose of 20 g/day for five days, and group B was given the recommended dose (1.5 g/kg/body weight and 1 g/kg body weight on days one and three, respectively). The primary outcome was in-hospital mortality. Secondary outcomes were improvements in clinical and laboratory parameters.

Results

Except for etiology, all the baseline clinical and laboratory variables in both groups were comparable. The in-hospital mortality in groups A and B was (11 [10.67%] vs. 6 [10.09%], (P = 0.965). The duration of hospitalization, 30-day mortality, improvement in shock and sensorium, and absolute improvements in serum creatinine, international normalized ratio (INR), and serum bilirubin were also comparable in both groups.

Conclusion

Low-dose albumin infusion in patients with cirrhosis and infections can have the same results as standard-dose albumin and can be used in resource-limited situations.

Clinical trial registration number

CTRI/2020/03/023794.

Graphical abstract

Cirrhosis and its complications constitute a common cause of liver-related mortality globally and in India. Various infections, including urinary tract infection (UTI), spontaneous bacterial peritonitis (SBP) and pneumonia, are common in these immunocompromised individuals. The incidence of these infections is about 4–5 times higher as compared to the general population. These infections commonly lead to an increase in morbidity and mortality due to sepsis and multi-organ dysfunction syndrome (MODS). The use of albumin has already been shown to reduce the risk of developing renal impairment and overall mortality in patients with SBP. Similarly, a multi-centric open label trial conducted recently showed improvement in acute on chronic liver failure (ACLF) and a reduction in the incidence of nosocomial infections in patients with cirrhosis and infections at admission.¹ Moreover, hypoalbuminemia in patients with decompensated cirrhosis leads to an increase in the bioavailability of prostaglandin E2, which is a potent immunosuppressant. Thus, albumin infusion may mitigate the immunosuppressive effect of prostaglandin.² The usual recommended dose of albumin in patients with SBP is 1 g/kg body weight on day - 1 and 1.5 mg/kg body weight on day - 3.³ In the recent study by Fernadez J. et al., a very high dose of albumin infusion (100 ml/h of 20 ml albumin) was used in patients with decompensated cirrhosis and sepsis.¹ Similarly, another prospective study by Fernadez J. et al., showed that use of high-dose albumin in patients with cirrhosis and infection was associated with downregulation of C-reactive protein, interleukin - 6 and total leukocyte count.⁴Moreover, a meta-analysis by Wong, Y. J et al., which included 3 randomized study (including the above study by Fernadez J. et al.) showed that the use of albumin may improve recovery from ACLF though it may increase the risk of developing pulmonary edema.⁵ Most of these studies used albumin at the dose commonly prescribed for patients with SBP (1.5 g/kg/day on day 1 and 1 g/kg/day on day 3).

However, in a resource-constrained country like India, the majority of these patients cannot afford albumin infusion according to the above recommendations. The aim of this study was to compare the impact of low-dose intravenous albumin infusions on outcomes in patients with cirrhosis and infection.

Methods

Study design

This prospective, non-inferiority, openlabel randomized control study was conducted at a teaching tertiary care center in eastern India. The study was approved by the institute's ethical committee and registered at the Clinical Trials Registry in India (identifier CTRI/2020/03/023794). This study was conducted in accordance with the World Medical Association Declaration of Helsinki. Written informed consent was obtained from all patients. The duration of the study was from March 2020 to December 2021. The study was approved by the institute's ethical committee (KIIT/KIMS/IEC/209/2019). All the participants had given full consent for the study.

Sample Size

The sample size was calculated based on a previous study, which showed 10% mortality in the study population⁶ with albumin infusion (power1 = 0.10) and 29% mortality without albumin infusion (power2 = 0.29). With 5% level of significance (α = 0.05) and 80% power (1-β), the calculated total sample size was 174 (2:1 randomization) with 116 patients in the study group and 58 patients in the control group.

Study Population

All consecutively hospitalized patients (>18 years of age) with cirrhosis and proven systemic infections (both complicated and uncomplicated) were included in the study. Patients with ACLF as per the Asia Pacific Association for Study of Liver (APASL) were also included in the study.^7,8 Patients were excluded from the study if they had a mean arterial pressure (MAP) of < 65 mm Hg, congestive heart failure with an ejection fraction of <50%, hepatocellular carcinoma, chronic kidney disease with a serum creatinine of >2 mg/dL, mechanical ventilation at inclusion time, or patients who refused to participate.

Study Protocol

Consecutive patients with cirrhosis with proven infection were randomly assigned via a computer-generated randomization table in a 2:1 ratio to groups A and B, which were concealed in a sealed envelope until randomization. The random allocation sequence was generated by a statistician, while enrollment and assignment of participants to the intervention were done by the principal investigator. After randomization, both groups received similar antibiotics and standard medical therapy. Group A patients were given albumin in a fixed dose of 20 g/day for five days, and group B patients were given the recommended doses of 1.5 g/kg body weight and 1 g/kg body weight on days one and three, respectively.⁹ Standard medical treatment (SMT) consisted of salt restriction, oral lactulose and rifaximin (for hepatic encephalopathy (HE)), and terlipressin (for hepatorenal syndrome) if required. The antibiotics selected were: A) Intravenous 3rd generation cephalosporin for spontaneous bacterial peritonitis.⁹ B) Intravenous 3rd-generation cephalosporin plus linezolid or clindamycin for skin infections¹⁰ C) Intravenous macrolide for community-acquired pneumonia.¹¹ D) Intravenous piperacillin-tazobactam plus vancomycin or linezolid for hospital-acquired pneumonia.¹² E) Injection 3rd-generation cephalosporin or amoxicillin-clavulanic acid for urinary tract infections.¹³ F) Inj 3rd-generation cephalosporin plus clindamycin for aspiration pneumonia.¹⁴ Antibiotics were changed based on the in vitro susceptibility of isolated bacteria or empirically if there was a lack of clinical response in patients with negative cultures. Monitoring was done on days 1 and 5, and the following parameters were recorded: heart rate, respiratory rate, MAP, total leukocyte count, liver function test, Child-Turcotte-Pugh score (CTP), model for end-stage liver disease score (MELD-Na), temperature, serum creatinine, prothrombin time, INR, and sensorium). Patients were followed until discharge or death and followed up for 30 days after discharge (either by follow-up visit or telephonically).

Definitions

The diagnosis of cirrhosis of the liver was based on clinical features, imaging, and laboratory investigations. The definition of sepsis and calculation of the sequential organ failure assessment score (SOFA) were based on the guidelines surviving sepsis in 2016.¹⁵ Sepsis was defined as life-threatening organ dysfunction caused by the dysregulated host response to infection.¹⁵ Organ dysfunction was identified as an acute change in total SOFA score ≥ 2 points due to infection. The baseline SOFA score was assumed to be zero in patients not known to have pre-existing organ dysfunction as defined in the 2016 surviving sepsis guidelines.¹⁵ The etiology of sepsis was determined as per clinical symptoms and laboratory investigations. A) Lower respiratory tract infections (LRTI) were diagnosed by the history of cough and/expectoration and/signs of consolidation-coarse crepitation on auscultation, positive sputum culture/positive broncho-alveolar lavage/x-ray chest suggestive of new infiltrations and/consolidation findings of ground-glass opacities. B) UTI were diagnosed based on the history of burning micturition and/or the number of pus cells with normal values of up to 5 pus cells/high power field (HPF) in males and up to 8 pus cells/HPF in females, respectively, and/or positive isolate from culture. C) The diagnosis of spontaneous bacterial peritonitis (SBP) was made when there was a positive ascitic fluid culture and/or an elevated ascitic fluid absolute polymorphonuclear count (≥250/μl) without evidence of an intra-abdominal surgically treatable source of infection. D) Skin and soft tissue infections were diagnosed clinically by the presence of pain and inflammatory signs (red color of the skin together with increased temperature with or without the presence of fluid or the formation of bullae).

Primary Objectives

To assess the impact of a lower dose of albumin infusion compared to the recommended dose (as used in patients with SBP) of albumin infusion on in-hospital mortality.

Secondary Objectives

To assess the impact of a lower dose of albumin infusion on improvement in clinical and laboratory parameters like 1) duration of hospitalization, 2) 30 days’ mortality, 3) hypotension (MAP improvement ≥10 mm of hg), 4) sensorium (improvement in HE by at least one grade), 5) serum creatinine, 6) INR, and 7) serum bilirubin.

Statistical Analysis

The data were collected and entered into the Microsoft Excel 2019 version. SPSS version 27 (IBM Corporation) was used to analyze the data further. The association between categorical variables and continuous variables was determined using the chi-square test or Fischer exact test and the independent t-test or a Mann–Whitney U test, respectively. The median was used along with the Mann–Whitney U test to calculate the duration of the average hospitalization. Univariate and multivariate analyses were done for the per-protocol analysis (total of 158 patients, Group A = 103 and Group B = 55). A univariate analysis was done to identify factors predicting the development of in-hospital mortality. Variables that reached statistical significance (P ≤ 0.05) in univariate analysis were subsequently included in multivariate analysis and analyzed by binary logistic regression for categorical variables and multiple linear regression for continuous variables to identify independent predictors of in-hospital mortality. All reported 'P' values were two-tailed, and a P value ≤ 0.05 was considered statistically significant.

Results

A total of 232 patients were screened, among whom 174 patients included in the study. Among the 174 included patients, 2:1 randomization was done, with 116 in group A and 58 in group B (Figure 1). All the baseline clinical variables, including CTP, model for end-stage liver disease score (MELD-Na), SOFA, and laboratory variables in both groups, were comparable (Table 1, Table 2). Except for alcohol, other causes of cirrhosis are comparable in both groups. The most common source of infection was UTI (58.6% vs. 56.9% in groups A and B respectively), followed by LRTI (26.7% vs. 27.6% in groups A and B, respectively). SBP was present in (21.7% vs. 21.6% in groups A and B, respectively). The most common organism isolated (Table 4) was Staphylococcus aureus (6.8% vs. 14% in groups A and B, respectively) (Table 3).

Figure 1.

Consort diagram of the study.

Table 1.

Baseline Characteristics of Clinical Data in the Study Cohort.

Variable	Group A N = 116	Group B N = 58	P value
Age (years) (mean ± SD)	54.34 ± 12.84	52.52 ± 13.54	0.388a
Sex (male) n (%)	93 (79.48)	51(89.4)	0.202b
Ascites n (%)	77 (66.4%)	38 (65.5%)	0.910b
Jaundice n (%)	52 (44.8%)	22 (37.9%)	0.386b
Altered sensorium n (%)	34 (29.3%)	23 (39.7%)	0.170b
UGI bleed n (%)	19 (16.4%)	13 (22.4%)	0.333b
Cough n (%)	18 (15.5%)	7 (12.1%)	0.541b
Diminished urine output n (%)	13 (11.2%)	8 (13.8%)	0.622b
Fever n (%)	10 (8.6%)	6 (10.3%)	0.711b
Pain abdomen n (%)	5 (5.2%)	6 (10.3%)	0.204b
Alcohol n (%)	69 (59.5%)	44 (75.9%)	0.033b
NASH n (%)	25 (21.6%)	10 (17.2%)	0.504b
Cryptogenic n (%)	14 (12.1%)	3 (5.2%)	0.149b
Others n (%)	8 (6.9%)	1 (1.7%)	0.146b
Urinary tract infection	68 (58.6%)	33 (56.9%)	0.828b
Lower respiratory tract infection	31 (26.7%)	6 (27.6%)	0.904b
Spontaneous bacterial peritonitis	18 (21.7%)	8 (21.6%)	0.994b
Skin and soft tissue infection	11 (9.5%)	9 (15.5%)	0.239b
≥2 site infections	9 (7.8%)	9 (15.5%)	0.113b
CTP score (median) (IQR)	10 4	10 (8–12)	0.767c
MELDNa score (mean ± SD)	21.90 ± 6.35	24.04 ± 7.60	0.560a
SOFA score (mean ± SD)	3.60 ± 2.02	3.71 ± 1.94	0.748a
MAP (mm of Hg) (median) (IQR)	82 (80–84)	82 (82–84)	0.505c
Heart rate (beats/minute) (median) (IQR)	90 (84–90)	90 (82.75–90)	0.376c
Temperature F (median) (IQR)	98 (98–98.4)	98 (98.0–98.2)	0.341c
Respiratory rate breaths/minute (median) (IQR)	18 (16–19)	18 (16–18)	0.550c

CTP = Child-Turcotte-Pugh; MAP = Mean arterial pressure; MELDNa = Model for end stage liver disease sodium; NASH = Non-alcoholic steatohepatitis; SD = Standard deviation; SOFA = Sequential organ failure assessment score; UGI = Upper gastrointestinal.

^aIndependent t test.

^bChi-square test.

^cMann whitney U test.

Table 2.

Baseline Characteristics of Laboratory Data in the Study Cohort.

Variable	Group A N = 116	Group B N = 58	P value
Serum albumin(mg/dl) (median) (IQR)	2.60 (2.40–2.90)	2.50 (2.17–2.82)	0.148b
Serum bilirubin(mg/dl) (median) (IQR)	2.37(1.21–7.16)	2.75(1.60–9.08)	0.169b
Serum sodium(meq/dl) (mean ± SD)	130.10 ± 13.67	129.10 ± 7.17	0.600a
INR (median) (IQR)	1.59 (1.40–1.90)	1.63(2.17–2.82)	0.384b
Platelet count (lakh/cu mm) (median) (IQR)	130 (73–152)	140 (87.50–160.25)	0.322b
Serum creatinine(mg/dl) (median) (IQR)	0.84 (0.62–1.21)	0.96 (0.69–1.45)	0.343b
Glomerular filtration rate (GFR) (ml/min/1.73m2) (median) (IQR)	76.50 (48.25–106.75)	76.50 (44.75–117.25)	0.929b
Total leucocyte count (x 103/micro liter) (median) (IQR)	7.17 (5.01–10.09)	8.90 (5.26–13.27)	0.009b
Hemoglobin(gm/dl) (mean ± SD)	9.27 ± 4.48	9.37 ± 2.77	0.880a

INR = international normalized ratio; IQR = Inter quartile range; SD = Standard deviation.

^aIndependent t-test.

^bMann–Whitney U test.

Table 4.

Organisms Isolated From any Site in Group A and Group B.

Organism isolated	Group A N = 116	Group B N = 58	Total N = 174
Staphylococcus aureus, n (%)	8(6.8)	8(14)	16
Klebsiella pneumonia, n (%)	8(6.8)	6(10.3)	14
Escherichia coli	9(7.6)	3(5.1)	12
Enterococcus faecalis, n (%)	8(6.8)	2(3.4)	10
Candida, n (%)	6(5.1)	0	6
Staphylococcus haemolyticus, n(%)	3(2.5)	2(3.4)	5
Burkholederiacepacia, n (%)	2(1.7)	2(3.4)	4
Pseudomonas, n (%)	1(0.8)	2(3.4)	3
Acinetobacter baumanni, n (%)	1(0.8)	1(1.7)	2
Citrobacter freundi, n (%)	1(0.8)	0	1
Enterobacter, n (%)	1(0.8)	0	1
Patients with ≥ 2 organisms isolated, n (%)	4(3.4)	1(1.7)	5

Table 3.

Baseline Characteristics of Laboratory Values on Day 1, 3, and 5.

Variable	N = 174	N = 174	N = 174	ANOVA
	Day 1	Day 3	Day 5	P value
Serum creatinine(mg/dl) (median) (IQR)	0.90 (0.62–1.31)	0.82 (0.60–1.24)	0.75 (0.57–1.20)	0.359
Serum bilirubin(mg/dl) (median) (IQR)	2.66 (1.33–7.84)	2.48 (1.30–7.32)	2.44 (1.20–7.32)	0.889
Serum albumin(mg/dl) (median) (IQR)	2.60 (2.40–3.1)	2.60 (2.40–3.10)	2.80 (2.50–3.12)	<0.001

IQR = Inter quartile range.

The median CTP in group A on day - 1 and day - 5 were 10^{9, 10, 11} and 9^{8, 9, 10} respectively. Similarly, in group B, on day - 1 and on day - 5 the CTP score were 10^{8, 9, 10, 11, 12} and 9,^{8, 9, 10, 11} respectively. The median serum creatinine (mg/dL) in groups A and B on day one was 0.93 (0.63–1.24) and 0.78 (0.59–1.10), and on day five it was 0.95 (0.66–1.38) and 0.74 (0.57–1.26), respectively. The median serum albumin (mg/dL) in group A and group B on day one was 2.60 (2.40–2.90) and 2.80 (2.50–3.10), and on day five it was 2.50 (2.20–2.90), and 2.80 (2.40–3.30) respectively (Table 5). However, the difference between the groups was not statistically significant.

Table 5.

Comparison of Clinical and Laboratory Parameters on Day 1,3, and 5.

Variable	Day 1		P Value	Day 3		P value	Day 5		P Value
	Group A N = 104	Group B N = 54		Group A N = 104	Group B N = 54		Group A N = 104	Group B N = 54
MAP (mm of Hg) (median) (IQR)	82 (80–86)	82(80–84)	0.505b	82 (80–84)	82 (80–84)	0.829b	80 (80–82)	81(80–84)	0.667b
MELDNa score (mean ± SD)	22.19 ± 6.26	23.55 ± 7.35	0.229a	20.73 ± 6.05	21.92 ± 7.70	0.901a	20.36 ± 6.14	21.85 ± 7.93	0.197a
CTP score (median) (IQR)	10(9–11)	10(8–12)	0.508b	9 (8–11)	9 (8–11)	0.836b	9(8–10)	9(8–11)	0.893b
SOFA score (mean ± SD)	3.62 ± 2.02	3.64 ± 1.93	0.964a	3.61 ± 2.07	3.67 ± 2.29	0.682a	3.77 ± 2.57	3.75 ± 2.53	0.960a
Serum creatinine(mg/dl) (median) (IQR)	0.93 (0.63–1.24)	0.95 (0.66–1.38)	0.343b	0.83 (0.62–1.18)	0.83 (0.60–1.27)	0.852b	0.78 (0.59–1.10)	0.74 (0.57–1.26)	0.861b
INR (median) (IQR)	1.60 (1.41–1.93)	1.61 (1.36–2.14)	0.384b	1.58 (1.30–2.0)	1.65 (1.37–2.0)	0.557b	1.60 (1.29–1.86)	1.60 (1.37–2.02)	0.247b
Serum bilirubin(mg/dl) (median) (IQR)	2.57 (1.19–6.90)	2.67 (1.55–8.81)	0.169b	2.47 (1.17–6.69)	2.88 (1.44–9.08)	0.216b	2.30 (1.10–7.25)	3.15 (1.33–8.00)	0.929b
Serum albumin(mg/dl) (median) (IQR)	2.60 (2.40–2.90)	2.50 (2.20–2.90)	0.169b	2.60 (2.4–3.0)	2.60 (2.30–3.10)	0.899b	2.80 (2.50–3.10)	2.80 (2.40–3.30)	0.663b

CTP = Child-Turcotte-Pugh; INR = international normalized ratio; IQR = Inter quartile range; MAP = Mean arterial pressure; MELDNa = Model for end stage liver disease sodium; SD = Standard deviation; SOFA = Sequential organ failure assessment score.

^aIndependent t test.

^bMann–Whitney U test.

A total of 19 patients had ACLF; 13 (11.2%) and 6 (10.3%) in Group A and B, respectively, according to the Asian Pacific Association for the Study Of The Liver (APASL) and among them patients having organ failure according to the APASL Asia Research Consortium (AARC) score in group A were grade I = 2 (1.7%) patients, grade II = 9 (7.7%), and grade III = 2 (1.7%) patients, and in group B, grade I = 0 patients, grade II = 3 (5.5%), and grade III = 3 (5.5%) patients, respectively.

The mean improvement in serum albumin (mg/dL) on days 1 and 5 in groups A and B was 0.21 ± 0.52 and 0.37 ± 0.61 respectively. The mean improvement in serum creatinine (mg/dL) on days 1 and 5 in groups A and B was 0.15 ± 0.60 and 0.03 ± 0.54, respectively. The mean improvement in serum bilirubin (mg/dL) on days 1 and 5in groups A and B was 0.25 ± 2.30 and 0.17 ± 2.87, respectively. The mean improvement in MAP (mm of Hg) on days 1 and 5 in groups A and B was 0.00 ± 3.78 and −1.49 ± 9.97, respectively (Table 6). However, the difference between the groups is not statistically significant.

Table 6.

Mean Improvement of Clinical and Laboratory Parameters on Day 5.

Variable	Group A N = 104	Group B N = 54	P Value
Mean improvement in MAP (mm of Hg) (mean ± SD)	0.00 ± 3.78	−1.49 ± 9.97	0.179a
Mean improvement in MELDNa score, (mean ± SD)	1.61 ± 5.22	1.64 ± 6.18	0.979a
Mean improvement in CTP score (mean ± SD)	0.56 ± 1.23	0.60 ± 1.69	0.876a
Mean improvement in SOFA score (mean ± SD)	−0.15 ± 1.79	−0.11 ± 2.14	0.910a
Mean improvement in serum creatinine (mg/dl) (mean ± SD)	0.15 ± 0.60	0.03 ± 0.54	0.2193
Mean improvement in INR (mean ± SD)	0.41 ± 0.48	0.63 ± 0.60	0.807a
Mean improvement in serum bilirubin (mg/dl) (mean ± SD)	0.25 ± 2.30	0.17 ± 2.87	0.852b
Mean improvement in serum albumin (mg/dl) (mean ± SD)	0.21 ± 0.52	0.37 ± 0.61	0.093a

CTP = Child-Turcotte-Pugh; INR = international normalized ratio; MAP = Mean arterial pressure; MELDNa = Model for end stage liver disease sodium; SD = Standard deviation; SOFA = Sequential organ failure assessment score.

^aIndependent t test.

^bMann–Whitney U test.

A total of 11 and 6 patients died in groups A and B, respectively. The difference in in-hospital mortality rate was not statistically significant (Table 7). The most common cause of death was sepsis, with multi-organ failure in both groups. In group A, two patients died of acute respiratory distress syndrome, and one patient died of an upper gastrointestinal bleed. On subgroup analysis, the in-hospital mortality was higher in patients with SBP (20% vs. 16.6% in groups A and B, respectively), whereas in non-SBP infections, the in-hospital mortality was 9.19% vs. 10.2% in groups A and B, respectively (Supplementary Table 1).

Table 7.

Comparison of Outcomes in Group A and Group B.

Variable	Intention to treat analysis			Per protocol analysis
	Group A N = 116	Group B N = 58	P value	Group A N = 103	Group B N = 55	P value
In-hospital mortality	11 (9.4%)	6 (10.34%)	0.965a	11 (10.67%)	6.0 (10.09%)	0.965a
Median duration of hospital stay, days (IQR)	8.0 (5–11)	8.0 (5.5–11)	0.921c	8.0 (6–11)	8.0 (6–11.25)	0.451c
Mortality by day 30	23 (19.8%)	14 (24.1%)	0.659a	23 (22.3%)	14 (25.4%)	0.659a
Hypotension improved	6/13 (46.1%)	0/3 (0)	0.250b	6/12 (50%)	0/3 (0%)	0.229b
Sensorium improved	20/34 (58.8%)	14/22 (63.6%)	0.719a	20/34 (58.8%)	14/22 (63.6%)	0.719a
Serum creatinine improved	68 (58.6%)	36 (62.0%)	0.813a	68 (66.01%)	36 (65.45%)	0.813a
INR improved	71 (61.2%)	38 (65.5%)	0.849a	71 (68.9%)	38 (69.09%)	0.984a
Serum bilirubin improved	64 (55.1%)	37 (63.7%)	0.522a	64 (62.1%)	37 (67.2%)	0.522a

INR = International normalized ratio; IQR = Interquartile range.

^aChi square test.

^bFischer's exact test.

^cMann Whitney U Test.

The median duration of hospitalization was eight days in both groups, which was comparable (Table 7). The remaining clinical secondary outcomes, 30-day mortality, improvement in hypotension, and sensorium (58.8% and 63.6% in groups A and B, respectively) were comparable in both groups (Table 7). Likewise, when compared to baseline values, absolute improvements in serum creatinine, INR, and serum bilirubin were also comparable in both groups (Table 7).

A total of 21 variables were included in the univariate analysis to predict variables associated with in-hospital mortality (Supplementary Tables 2 and 3). Univariate analysis showed that lower respiratory tract infection, nonalcoholic steatohepatitis (NASH) as the etiology of cirrhosis, and MAP at baseline reached statistical significance (P < 0.05) and were therefore selected for multivariate analysis as independent predictors of in-hospital mortality. Interestingly, the dose of albumin used didn't influence the mortality rate (Supplementary Tables 2 and 3).

After adjusting for potential confounders, in perprotocol multivariate analysis, two variables were independently associated with in-hospital mortality: LRTI as the focus of infection at admission (OR 6.237; 95% CI 1.392–27.951; P = 0.017) and NASH as the etiology of cirrhosis (OR 7.022; 95% CI 1.238–39.825; P = 0.028) (Table 8).

Table 8.

Multivariate Analysis of Factors Affecting In-hospital Mortality According to Per-protocol Analysis.

Variable	P-value	OR	95% C. I for OR
			Lower	Upper
LRTI at admission	0.017	6.237	1.392	27.951
NASH at admission	0.028	7.022	1.238	39.825
MAP on day 1	0.106	0.926	0.844	1.016

1 = day 1; MAP = Mean arterial pressure; NASH = Non-alcoholic steatohepatitis; LRTI = Lower respiratory tract infection.

Bold signifies value < 0.05.

Discussion

The European Association for the Study of liver (EASL)⁹ guidelines recommends intravenous albumin at a dose of 1.5 g/kg body weight and 1 g/kg body weight on days 1 and 3, respectively, for patients with cirrhosis and SBP (3). This recommendation was based on a study by P. Sort et al.,⁶ where patients with cirrhosis and SBP were treated with antibiotics, and this regimen of intravenous albumin infusion had a 30% reduction in type 1 HRS and a 10% reduction in mortality when compared to antibiotics alone. The study made no mention of how the dose of intravenous albumin was determined.⁶ It appears that the selection of this dose was arbitrary.⁹ The use of albumin infusion in patients with non-SBP infections is not well established at present. However, a recent open-label randomized study using albumin in patients with cirrhosis and non-SBP infection showed that mortality was similar in patients irrespective of the use of albumin. However, in patients on albumin infusion, there was improvement in ACLF. Also, albumin infusion resulted in the significant reduction in the incidence of hospital-acquired infections.¹ Moreover, a recent meta-analysis has also shown recovery from ACLF with the use of albumin, though there was no effect on renal impairment or mortality. Rather the use of albumin may worsen pulmonary edema in these patients.⁵ Based on these conflicting findings, the routine use of albumin infusion has not been advocated in these patients. Moreover, the use of albumin in these high doses is not cost-effective and may even be hazardous (pulmonary edema and fluid overload).¹⁶ However, use of a lower dose of albumin may be the preferred option and may actually have a cost-benefit effect in a resource- rich country like India. Also the use of albumin has been shown to promote resolution of ACLF and reduce levels of proinflammatory cytokines. The risk of developing nosocomial infections may also be reduced with an albumin infusion.^4,5 Our study showed that a regime using a lower dose of albumin infusion is not inferior to a recommended dose of albumin infusion when used with antibiotics to treat cirrhosis with infections.

The most common focus of infection in our study was UTI, followed by LRTI, SBP, skin, and soft tissue infections. In a study by Rajiv Baijal et al.,¹⁷ SBP (33%), followed by UTI (23%), was predominant. This could be because we defined UTI as ≥ 10 pus cells/HPF in this study. Another reason for the low incidence of SBP in our study could be the prior use of antibiotics before admission to our hospital. The observed infection pattern in our study was similar to the study by Merli et al. {UTI (35%), lower respiratory tract infection (22%), and SBP (8%)}.¹⁸

The in-hospital mortality in our study was 11 (10.67%) in the low-dose albumin group (group A) and 6 (10.09%) in the recommended-dose albumin group (group B), respectively. In univariate analysis, the dose of albumin didn't affect in-hospital mortality. The in-hospital mortality of patients with infection and cirrhosis varies between 10 and 30% in various studies, with more recent studies showing relatively lower mortality compared to the older studies.^{19, 20, 21} In a study by P. Sort et al. (including only patients with SBP),⁶ the in-hospital mortality was 10% in the recommended dose of albumin plus antibiotic group. Our study shows that in-hospital mortality in the low-dose albumin group was similar to that in the recommended-dose albumin group of other Indian and western studies,^{5,16, 17, 18, 19} proving our hypothesis that low-dose intravenous albumin has similar efficacy to the recommended albumin dose in patients with cirrhosis with infections. In agreement with this hypothesis, another recent study has also failed to show any survival benefit of albumin infusion to maintain serum albumin at more than 3 gm/dL in patients with decompensated cirrhosis.¹⁶

The duration of hospitalization was eight days in both groups which was similar (6.4 ± 3.4 days) to the Indian study done by Sharma BC et al.²² In contrast, it was 14 days in the albumin plus antibiotic group in the study by P. Sort et al.⁶ The prolonged duration of hospitalization could be due to underlying advanced cirrhosis in most of these patients. This study was also relatively older compared to ours. Recent advances in intensive care have resulted in significant improvements in morbidity and mortality. The 30-day mortality was 22.3% and 25.4% in groups A and B, respectively. The 30-day mortality in a study by Uday Kumar et al.²³ was 46.72%, almost double our study. The lower percentage in the present study could be due to a predominant UTI and a smaller number of advanced cirrhosis cases in our cohort. Both duration of hospitalization and mortality by day 30 were lower in our lower-dose intravenous albumin group, reinforcing our hypothesis that the recommended dose of intravenous albumin is arbitrary in patients with cirrhosis and infection.

The improvement in serum creatinine was seen in 66.01% and 65.45% of patients in groups A and B, respectively. In a prospective RCT study by Guevara M. et al.,²⁴ the authors showed an improvement in serum creatinine to baseline in about 70% of patients who received albumin. Improvement in sensorium was seen in 58.8% and 63.6% of groups A and B, respectively, which was similar to other studies.^22,25,26 There was an improvement in serum INR and serum bilirubin in both groups, which were comparable.

In per-protocol multivariate analysis, LRTI infection at admission and NASH etiology at admission were independently associated with in-hospital mortality. The probable reason could be that patients admitted with cirrhosis and LRTI infection have high mortality (up to 41%).^{27, 28, 29} Patients with NASH etiology had a higher mean age (years) of presentation (63.53 ± 10.998) and several comorbidities like diabetes mellitus, hypertension, and coronary artery disease, which may explain a higher in-hospital mortality in these patients. Further, though baseline CTP and MELD Na scores are good at predicting three-month mortality, our patients were not followed for three months to analyze the overall survival.

Large-volume transfusions of albumin (500 ml/24 h) in a single day can result in fluid overload, especially pulmonary edema. The recommended dose of albumin (1.5 g/kg body weight and 1 g/kg body weight on days 1 and 3, respectively) is weight-based, but in the presence of ascites, it is difficult to calculate the adjusted body weight. Also, the guidelines do not say whether we have to consider actual body weight or adjusted body weight for albumin dose infusion. Regarding the cost of albumin infusion, for example, a patient with 70 kg requires 175 g of albumin according to the recommended dose, whereas it was 100 g of albumin with respect to lower doses of albumin therapy, leading to a 42.85% reduction in the cost of albumin therapy.

Patients with decompensated cirrhosis and infection often have associated peripheral arteriolar vasodilation, circulatory dysfunction, a compensatory increase in vasoactive cytokines interleukin - 6, and tumor necrosis factor-α, and increased bacterial translocation.^{30, 31, 32, 33, 34, 35} The beneficial effect of intravenous albumin is possibly due to the prevention of circulatory dysfunction, subsequent activation of the vasoconstrictor system, and decreased cytokine and endotoxin levels in the blood.^{22,32,36, 37, 38} Previous studies have shown that recommended doses of intravenous albumin have improved both mortality and morbidity in patients with cirrhosis with SBP and cirrhosis with sepsis other than SBP.^6,24 However, in our study, we found comparable primary and secondary outcomes in both lower dose and recommended dose intravenous albumin groups in both intentions to treat and per-protocol analyses, which strengthens our rationale for the study in a resource-constrained setting. The authors conclude that the recommended dose of intravenous albumin is not optimal. We suggest multicentric trials with a large population should be done to identify the optimal dose of intravenous albumin required in cirrhosis with infections.

Strengths of the study

Ours is the first study comparing various doses of albumin in infected cirrhosis patients. Sample size has been calculated, which is a large cohort of patients with cirrhosis and infection.

Limitations

Our study is a nonblinded, single-center study, thus these results may not be generalized to other populations. Follow-up data for patients was limited to 30 days, which may make it difficult to interpret the effect of low-dose albumin infusion on long-term survival. A detailed follow-up data up to 90 days would have been better to demonstrate the beneficial effect of albumin in these patients. Finally, there is a scarcity of literature regarding the use of albumin infusion in patients with non-SBP infections. Using a historical cohort (including patients with cirrhosis and non-SBP infections not on albumin infusion) and comparing it with the current cohort would have been a better study.

Our prospective randomized control study concludes that, compared with the recommended dose of intravenous albumin therapy, lower-dose intravenous albumin therapy may have similar outcomes on short-term mortality, the average duration of hospital stay, and morbidity in patients with cirrhosis with underlying infection. However, large multicentric randomized studies should be performed in the future to confirm this finding.

As intravenous albumin is expensive and limited in a resource-limited setting, a lower dose of intravenous albumin may provide similar benefits as a conventional dose of albumin infusion.

Credit authorship contribution statement

VJD: Conceptualization, Data Curation, Writing Original Draft; BM: Supervision, Project Administration, Resources, Formal Analysis, Writing- Review And Editing; SKA: Supervision, Methodology, Writing- Review And Editing; RK: Visualization, Writing- Review And Editing; AT: Visualization, Writing- Review And Editing, DLP: Visualization, Writing- Review And Editing; PN: Visualization, Formal analysis, Writing- Review And Editing; SCP: Visualization, Writing- Review And Editing; ACA: Supervision, Methodology, Writing- Review And Editing, Project Administration; YKC: Supervision,Visualization, Writing- Review And Editing.

Conflicts of interest

The authors have none to declare.

Funding

This research did not receive any specific grant from funding agencies in public, commercial, or not-for-profit sectors.

Appendix A. Supplementary data

The following is the Supplementary data to this article.