If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Hospital Universitari Germans Trias i Pujol, Liver Unit, Gastroenterology, Departament of Medicine, Universitat Autònoma de Barcelona, Badalona, Barcelona, Spain
Hospital Universitari Germans Trias i Pujol, Liver Unit, Gastroenterology, Departament of Medicine, Universitat Autònoma de Barcelona, Badalona, Barcelona, SpainCentro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, CIBERHED, Barcelona, Spain
Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, CIBERHED, Barcelona, SpainFundació Germans Trias i Pujol, Gastroenterology, Badalona, Spain
Hospital Universitari Germans Trias i Pujol, Liver Unit, Gastroenterology, Departament of Medicine, Universitat Autònoma de Barcelona, Badalona, Barcelona, SpainCentro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, CIBERHED, Barcelona, Spain
Hospital Universitari Germans Trias i Pujol, Liver Unit, Gastroenterology, Departament of Medicine, Universitat Autònoma de Barcelona, Badalona, Barcelona, SpainCentro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, CIBERHED, Barcelona, Spain
Hospital Universitari Germans Trias i Pujol, Liver Unit, Gastroenterology, Departament of Medicine, Universitat Autònoma de Barcelona, Badalona, Barcelona, SpainCentro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, CIBERHED, Barcelona, Spain
Hospital Universitari Germans Trias i Pujol, Preventive Medicine and Epidemiology Department, Autonomous University of Barcelona, Badalona, Barcelona, Spain
Hospital Universitari Germans Trias i Pujol, Liver Unit, Gastroenterology, Departament of Medicine, Universitat Autònoma de Barcelona, Badalona, Barcelona, Spain
The early hospital readmission of patients with decompensated cirrhosis is a current problem. A study is presented on the incidence, the impact on mortality, and the predictive factors of early hospital readmission.
Patients and methods
On the study included 112 cirrhotic patients, discharged after some decompensation between January 2013 and May 2014. Multivariate analyses were performed to identify predictors of early readmission and mortality.
Results
The early readmission rate was 29.5%. The predictive factors were male gender (OR: 2.81; 95% CI: 1.07–7.35), Model for End-Stage Liver Disease-sodium score ≥15 (OR: 3.79; 95% CI 1.48–9.64), and Charlson index ≥7 (OR: 4.34, 95% CI 1.65–11.4). This model enabled patients to be classified into low or high risk of early readmissions (13.6% vs. 52.2%). The mortality rate was significantly higher among patients with early readmission (73% vs. 35%) (p< .0001). After adjusting for the Model for End-Stage Liver Disease-sodium score, Charlson index, dependence in activities of daily living, educational status, and number of medications on discharge, the early readmission was independently associated with mortality.
Conclusions
Early hospital readmission is common, and is independently associated with mortality. Male gender, MELD-Na ≥15, and Charlson index ≥7 are predictors of early readmission. These results could be used to develop future strategies to reduce early readmission.
Cirrhosis is the end stage of chronic liver disease, is associated with high mortality, and is the second leading cause of digestive tract disease-related death (after colorectal cancer) [
]. Cirrhosis is also responsible for significant morbidity and health-care costs. It leads to more than 150,000 hospital admissions, costing nearly 4 billion dollars each year in the United States [
]. When a patient is hospitalized for decompensated cirrhosis, the risk of readmission is very high, with an overall actuarial probability of readmission at 1 year of 45% [
Readmission within 30 days after hospital discharge (early readmission) among the Medicare population often leads to high health care costs, and also has become a measurement of quality of health care [
]. Of those discharged from an acute-care hospital in the United States between 2003 and 2004, 19.6% were readmitted within 1 month, at a cost of more than 17 billion dollars, which represents nearly 20% of the Medicare budget [
]. Moreover, these were conducted in the United States. Therefore, the aims of this study were to identify the incidence of early readmission (≤30 days) in patients with decompensated cirrhosis, its impact on mortality, and the predictive factors that will allow us to classify the patients in risk groups of early readmission in our population.
2. Patients and methods
2.1 Patients
This retrospective, longitudinal, and observational study, was conducted in the Hepatology Unit of the University Hospital Germans Trias i Pujol (Badalona, Spain). It included cirrhotic patients discharged between January 2013 and May 2014, after being admitted due to some decompensation:
•
Hepatic encephalopathy (EH), defined as a brain dysfunction caused by liver insufficiency and/or portosystemic shunting (PSS) [
Hepatic encephalopathy in chronic liver disease: 2014 practice guideline by the European Association for the Study of the Liver and the American Association for the Study of Liver Diseases.
]. Patients with EH grade I to grade IV according to West Haven criteria were included.
•
Portal hypertensive bleeding diagnosed by emergency endoscopy on observing one of the following: (1) active bleeding from a varix; (2) white nipple or clot adherent to a varix; (3) when varices are the only lesion found, and if blood is present in the stomach or endoscopy is performed after 24 h of hemorrhage [
Portal hypertensive bleeding in cirrhosis: risk stratification, diagnosis, and management: 2016 practice guidance by the American Association for the study of liver diseases.
Spontaneous bacterial peritonitis (SBP), defined as >250 polymor-phonuclear cells per high-power field and/or monomicrobial culture in the ascitic fluid [
Clinically evident ascites documented on physical examination: moderate (grade 2) or large ascites (grade 3), according to the classification of The International Ascites Club [
The follow-up was until death, or the end of the study period in December 2014.
Exclusion criteria were the following: decompensated cirrhosis patients who died, or had voluntary discharge during the index hospitalization, and patients with HIV infection.
Data collected as regards the index hospitalization included: demographic data (gender, age, dependence in activities of daily living (ADLs), educational status, living alone), etiology of cirrhosis, history of admissions for decompensated cirrhosis in the previous year, presence of hepatocellular carcinoma, Barcelona Clinic liver cancer (BCLC) staging, Body Mass Index (BMI), Charlson index (estimated survival at 10 years, depending on the age and comorbidities of the patient), the main cause of admission, presence of infection (non-SBP infections), acute kidney injury (AKI) during the index hospitalization, length of hospital stay, discharge location (nursing facility or home) and most recent values on discharge, including Model for End-stage Liver disease-sodium (MELD-Na) score, laboratory values (platelet count, serum bilirubin, prothrombin time, serum albumin, serum sodium, and serum creatinine), and number of medications.
We evaluated readmissions for decompensated cirrhosis (HE, ascites, with or without renal failure, portal hypertensive bleeding, or SBP) or any other cause, but complicated for some decompensation of cirrhosis in our hospita,l or at other institutions during the study period, time to first readmission (<or> 30 days), and the total number of hospital readmissions at the end of follow-up. In patients with multiple admissions, only the first admission was considered as the index hospitalization, and the others were considered readmissions.
In addition, an analysis was made on the number of emergency service consultations post-discharge, and the mortality at the end of study.
When there was more than one decompensation of cirrhosis at admission, the criteria used to define the main cause of admission were the following: (1) Whenever portal hypertensive bleeding was coincident with another complication, it was considered the main cause of admission, because it frequently causes bacterial infection or hepatic encephalopathy; (2) When bleeding was not present at admission, infection (SBP) was considered the main cause of hospitalization; and (3) In patients with hepatic encephalopathy and ascites, the main cause was the former.
Data were compiled by examining the computerized medical record for each patient and by telephone follow-ups. The study protocol was in compliance with the ethical standards of the Helsinki Declaration of 1975 and was approved by the Ethics Committee of University Hospital Germans Trias i Pujol as part of a larger study of a program to improve the post-discharge support of decompensated cirrhotic patients. Informed consent was obtained from each patient.
2.2 Statistical analysis
To evaluate the predictive factors of early hospital readmission we compared the two groups (patients with and without early hospital readmission) using the Chi-square test for qualitative variables [continuous variables were categorized according to the best cut-off values using receiver operating characteristic (ROC) curve analysis]. Variables that were found to be different (p < .05) in the univariate analysis were used to create different nested models by logistic regression and were then compared by Likelihood ratio tests (LRTs) to select the significant variables and to determine the best model. This procedure was performed because the number of significant variables was greater than the number of events, and all of them have been associated with re-hospitalization in previous studies. The results were presented as odds-ratios with a 95% confidence interval. Model calibration was evaluated using the Hosmer–Lemeshow goodness of fit test. Model discrimination was assessed by the c-statistic using ROC curve analysis. Logistic regression was used as the time of readmission was not taken into account. The final model obtained, allowed us to classify the patients into early readmission risk groups.
A univariate and multivariate analyses were used to evaluate if the early hospital readmission is a predictive factor of mortality in a patient with decompensated cirrhosis. They were performed using the Kaplan–Meier method and compared with the log-Rank test. Those variables with a p< .05 were included in a multivariate Cox regression analysis, and the results were presented as a hazard ratio with a 95% confidence interval.
Statistical analysis of the data was performed using the IBM SPSS version 21 statistical software (SPSS Inc. and Microsoft Corp., Chicago, IL).
3. Results
A total of 126 patients met the inclusion criteria, but 14 of them were excluded due to one with a voluntary discharge, and 13 died during the index hospitalization. None were infected with HIV. Therefore, the study finally included a total of 112 patients with a mean follow up of 11.1 ± 7.2 months.
Clinical characteristics of the cohort are shown in Table 1. The mean age of patients was 65.2 ± 11.8 years. The cohort was predominantly male (57.1%), and had a BMI ≥25 kg/m2 (56.3%). The main origins of cirrhosis were hepatitis C virus (HCV) (39.3%) and alcohol (38.4%), and the mean of MELD-Na at discharge was 14.7 ± 4.9. Ascites was the main reason for the index admission (55.4%), and the mean of length of hospital stay was 21.0 ± 15.0 days.
Table 1Demographic, clinical and laboratory data obtained at discharge from index hospitalization.
Variables
Overall (n = 112)
30-day readmission (n = 33)
No 30-day readmission (n = 79)
p Value
Male sex, n (%)
64 (57.1)
24 (72.7)
40 (50.6)
.031
Age, mean ± SD (yr)
65.2 ± 11.8
68.3 ± 10.5
64.0 ± 12.1
.075
Lives alone, n (%)
16 (14.3)
3 (9.1)
13 (16.5)
.310
Dependence in activities of daily living, n (%)
22 (19.6)
7 (21.2)
15 (19.0)
.787
Educational status (Elementary School), n (%)
77 (75.5)
24 (80.0)
53 (73.6)
.494
Admissions in the year before index hospitalization, (sí), n (%)
29 (25.9)
13 (39.4)
16 (20.3)
.035
Etiology of liver disease, n (%)
- Alcohol
43 (38.4)
13 (39.4)
30 (38.0)
.754
- HVC
44 (39.3)
11 (33.3)
33 (41.8)
- HVB
6 (5.4)
2 (6.1)
4 (5.1)
- NASH
5 (4.5)
1 (3.0)
4 (5.1)
- Others
14 (12.5)
6 (18.2)
8 (10.1)
MELD-Na score, mean±SD
14.7 ± 4.9
16.9 ± 5.0
13.8 ± 4.6
.002
BMI (kg/m2), n (%)
- <18.5
9 (8.0)
3 (9.1)
6 (7.6)
.926
- 18.5–24.9
40 (35.7)
11 (33.3)
29 (36.7)
- >25
63 (56.3)
19 (57.6)
44 (55.7)
Hepatocellular carcinoma, n (%)
33 (29.5)
14 (42.4)
19 (24.1)
.052
BCLC staging, n (%)
- BCLC 0
1 (3.0)
1 (7.1)
0 (0)
.120
- BCLC A
6 (18.2)
1 (7.1)
5 (26.3)
.480
- BCLC B
7 (21.2)
4 (28.6)
3 (15.8)
.097
- BCLC C
9 (27.3)
1 (7.1)
8 (42.1)
.208
- BCLC D
10 (30.3)
7 (50)
3 (15.8)
.003
Charlson Index, mean ± SD
6.6 ± 2.8
7.6 ± 2.3
6.2 ± 2.9
.017
Cause of index admission, n (%)
- Ascites
62 (55.4)
18 (54.5)
44 (55.7)
.330
- Portal hypertensive bleed
22 (19.6)
6 (18.2)
16 (20.3)
- Hepatic encephalopathy
15 (13.4)
7 (21.2)
8 (10.1)
- SBP
13 (11.6)
2 (6.1)
11 (13.9)
Acute Kidney injury during the index hospitalization, n (%)
53 (47.3)
19 (57.6)
34 (43.0)
.160
Staging of AKI, n (%)
- Stage 1
36 (67.9)
15 (78.9)
21 (61.8)
.294
- Stage 2
8 (15.1)
1 (5.3)
7 (20.6)
- Stage 3
9 (17.0)
3 (15.8)
6 (17.6)
Response to treatment of AKI
- No response
5 (9.4)
2 (10.5)
3 (8.8)
.925
- Parcial response
10 (18.9)
4 (21.1)
6 (17.6)
- Full response
38 (71.7)
13 (68.4)
25 (73.5)
Infection during the index hospitalization, n (%)
53 (47.3)
17 (51.5)
36 (45.6)
.566
Serum sodium (mmol/L), mean ± SD
136.3 ± 4.0
134.6 ± 4.0
137.0 ± 3.7
.004
Serum bilirrubin (mg/dL), mean ± SD
2.4 ± 2.9
3.4 ± 4.2
2.0 ± 1.9
.079
Prothrombin time (%), mean±SD
68.8 ± 16.1
66.8 ± 17.8
69.7 ± 15.3
.402
Serum albumina (g/L), mean±SD
28.7 ± 4.8
28.0 ± 5.0
29.0 ± 4.7
.356
Platelet count (x109/L), mean±SD
105.2 ± 68.7
105.5 ± 87.2
105.1 ± 59.9
.974
Serum creatinine (mg/dL), mean± SD
1.0 ± 0.8
1.0 ± 0.38
1.0 ± 0.97
.730
Lenght of stay (days), mean±SD
21.0 ± 15.0
20.4 ± 9.9
21.2 ± 16.7
.816
Discargue location, n (%)
- Home
107 (95.5)
33 (100)
74 (93.7)
.139
- Nursing facility
5 (4.5)
0 (0)
5 (6.3)
Number of medications on discharge, mean ± SD
7.5 ± 3.3
8.3 ± 3.2
7.1 ± 3.3
.081
Number of emergency service consultations post-discharge, mean±SD
1.7 ± 2.3
2.3 ± 1.9
1.4 ± 2.4
.073
Number of total readmission at the end of follow-up, mean ± SD
1.1 ± 1.4
1.7 ± 0.9
0.9 ± 1.6
.013
Cause of readmission, n (%)
- Hepatic encephalopathy
31 (48.4)
21 (63.6)
10 (32.3)
.004
- Ascites
24 (37.5)
6 (18.2)
18 (58.1)
- Portal hypertensive bleed
6 (9.4)
5 (15.2)
1(3.2)
- SBP
2 (3.1)
0 (0)
2 (6.5)
- Acute Kidney injury
1 (1.6)
1 (3.0)
0 (0)
Mortality at the end of follow-up, n (%)
52 (46.4)
24 (72.7)
28 (35.4)
<.0001
MELD-Na, Model for End-Stage Liver Disease-sodium; BMI, Body Mass Index; SBP, Spontaneous Bacterial Peritonitis; BCLC, Barcelona Clinic liver cancer staging; AKI stage1 (sCr increase ≥0.3 mg/dL or ≥1.5-2 fold from baseline); AKI stage 2 (sCr increase >2-fold to 3-fold from baseline); AKI stage 3 (sCr increase >3-fold from baseline or sCr ≥4.0 mg/dL with an acute increase ≥0.3 mg/dL or initiation of renal replacement therapy.
Sixty-four (57.2%) patients were readmitted during the follow-up period, with a readmission rate of 3.5/person-year. The median time to early readmission was 11 days (range: 8–16). Thirty-three (29.5%) patients were readmitted within 30 days, 31 (27.7%) were readmitted after 30 days, and 48 (42.9%) were not readmitted during the study period.
The main cause of early hospital readmission was hepatic encephalopathy (63.6%), for which the precipitating factors were diuretics overdose (38%), non-SBP infections (38%), constipation (14.5%), and use of benzodiazepines (9.5%). The other causes of readmission in the early readmission group were ascites (18.2%) and GI bleeding (15.2%). In patients readmitted after 30 days, ascites was the main cause of readmission (58.1%), followed by HE (32.3%) and SBP (6.5%) Table 1.
Patients with early hospital readmission had a higher number of readmissions during the follow-up period than patients with no early readmission (mean 1.7 ± .9 vs. .9 ± 1.6; p = .013). Emergency service consultations post-discharge were also higher in this group of patients (mean of 2.3 ± 1.9 vs. 1.4 ± 2.4), but the difference was not statistically significant (p = .073).
3.1 Predictors of early hospital readmission (Table 2)
The variables associated with 30-day readmission (p < .05) in the univariate analysis were: age ≥63 years (p = .014), male gender (p = .031), history of admissions for decompensated cirrhosis in the previous year (p = .035), MELD-Na ≥15 at discharge (p = .005), Charlson index ≥7 (p = .002), BCLC stage D hepatocellular carcinoma (p = .003), hyponatremia ≤135 mmol/L (p = .002), serum creatinine ≥.9 mg/dL (p< .001), and the number of medications at discharge ≥7 (p = .007). Of these, the serum sodium and creatinine were excluded from nested models because they were calculated in the MELD-Na score. The age and BCLC stage D hepatocellular carcinoma variables were also excluded because they were included in the Charlson index.
Considering that the number of variables (5) exceeds the number of outcome events (n = 33) and the variables have been associated with the early hospital readmission in patients with and without cirrhosis, we performed nested models by regression logistic and compared with Likelihood ratio tests (LRTs).
A total of five nested models were created. The initial model included the Charlson index, taking into account their statistical significance and OR, for which LRTs were 125.3. The following models were created, including the Charlson index and the other variables. The number of medications at discharge ≥7 and the history of admissions for decompensated cirrhosis in the previous year were not significant and were removed from the model Table 3.
Table 2Predictors of early hospital readmission.
Variables
Number of variable
Univariate analysis
Final model
OR
CI 95%
p Value
OR
CI 95%
p Value
Charlson index ≥7
V1
3.92
(1.57–9.77)
.002
4.34
(1.65–11.4)
.003
MELD-NA score ≥15
V2
3.37
(1.41–8.04)
.005
3.79
(1.48–9.64)
.005
Number of medications on discharge ≥7
V3
3.37
(1.35–8.37)
.007
1.88
(.64–5.53)
.165
Male sex
V4
2.60
(1.07–6.29)
.031
2.81
(1.07–7.35)
.035
Admissions for decompensated cirrhosis in the previous year
Variables excluded from nested models: the creatinine and serum sodium because they were included in the MELD-Na score and the BCLC stage D and age because they were included in the Charlson index.
Variables excluded from nested models: the creatinine and serum sodium because they were included in the MELD-Na score and the BCLC stage D and age because they were included in the Charlson index.
Variables excluded from nested models: the creatinine and serum sodium because they were included in the MELD-Na score and the BCLC stage D and age because they were included in the Charlson index.
Variables excluded from nested models: the creatinine and serum sodium because they were included in the MELD-Na score and the BCLC stage D and age because they were included in the Charlson index.
V9
3.04
(1.22–7.57)
.014
–
–
–
OR, Odds Ratio; CI, Confidence interval. BCLC, Barcelona Clinic liver cancer staging HCC, Hepatocellular carcinoma.
Continuous variables were categorized according to the best cut-offs identified by ROC statistics.
a Variables excluded from nested models: the creatinine and serum sodium because they were included in the MELD-Na score and the BCLC stage D and age because they were included in the Charlson index.
The final model of the predictors of early hospital readmission included: male gender (OR: 2.81; 95% CI: 1.07–7.35; p = .035), MELD-Na score ≥15 (OR: 3.79; 95% CI 1.48–9.64; p = .005) and Charlson index ≥7 (OR: 4.34, 95% CI 1.65–11.4; p = .003) with a LRTs of 112.46, Hosmer–Lemeshow test of 0.657 and a predictive capability with a c-statistic of .76 (95% CI: .66–.86).
The final model allowed the patients to be stratified into groups of risk. Patients with high risk were readmitted within 30 days in 52.2% (24/46) of cases, while patients with low-risk only 13.6% (9/66) were readmitted early (RR: 3.82; 95% CI: 1.96–7.46; p< .0001) Fig. 1. The mean time to early readmission was 28.7 days in the lowest-risk group, compared with 20.7 days in the highest risk group (p<.0001).
Fig. 1Risk groups of early readmission according the final model. Low-risk group (n = 66): 13.6% of patients were readmitted within 30 days. High risk group (n = 46): 52.2% of patients were early readmitted (p< .0001).
3.2 Impact of early hospital readmission on mortality (Table 4)
The overall mortality at the end of follow-up was 46.4% (52/112). The mortality rate was significantly higher in patients who were readmitted within 30 days than those with no early readmission (72.7% vs. 35.4%) (OR: 4.85; 95% CI 1.98–11.87; p< .0001). Fig. 2 shows the probability of 1-year survival in the two groups of patients.
Fig. 2Survival at the end of follow-up between patients with & without early readmission.
Variables excluded from the multivariate Cox regression analysis: the BCLC stage D HCC and age because they were included in the Charlson index and the serum creatinine because it was included in the MELD-Na score.
Variables excluded from the multivariate Cox regression analysis: the BCLC stage D HCC and age because they were included in the Charlson index and the serum creatinine because it was included in the MELD-Na score.
Variables excluded from the multivariate Cox regression analysis: the BCLC stage D HCC and age because they were included in the Charlson index and the serum creatinine because it was included in the MELD-Na score.
a Variables excluded from the multivariate Cox regression analysis: the BCLC stage D HCC and age because they were included in the Charlson index and the serum creatinine because it was included in the MELD-Na score.
The most frequent causes of death were advanced stage HCC (29.2% of patients in early readmission group vs. 39.3% of patients with no early readmission, p = .444), bacterial infections no-SPB (29.2% vs. 21.4%, p = .521), and gastrointestinal bleeding (16.7% vs. 3.6%, p = .110).
The variables associated with mortality at the end of follow-up (p< .05) in the univariate analysis were: age ≥63 years (p< .0001), BCLC stage D hepatocellular carcinoma (p< .0001), early hospital readmission (p< .0001), Charlson index ≥7 (p< .0001), MELD-Na ≥15 at discharge (p = .014), dependence in activities of daily living (ADLs) (p = .006), elementary school education (p = .044), number of medications on discharge ≥7 (p = .028), and serum creatinine ≥.9 mg/dL (p< .0001).
The BCLC stage D hepatocellular carcinoma and age variables were excluded from the Cox multivariate regression analysis as they were included in the Charlson index, as well as the serum creatinine variable as it was included in the MELD-Na score.
Finally, early hospital readmission was an independent predictor of mortality (HR: 2.40, 95% CI: 1.25–4.61; p = .009), adjusted for a Charlson index ≥7 (HR: 2.47; 95% CI: 1.14–5.32; p = .021), ADLs (HR: 2.27, 95% CI: 1.13–4.38; p = .020), MELD-Na ≥15 (HR: 1.94; 95% CI: 1.04–3.62; p = .036), elementary school education (HR: 1.21, 95% CI: .48–3.09; p = .678), and number of medications on discharge ≥7 (HR: 1.02, 95% CI: .47–2.20; p = .954).
4. Discussion
This study found that early hospital readmission (≤30 days) among patients with decompensated cirrhosis is common, with an incidence of 29%. The early readmission had a negative impact on the survival, since the patients readmitted within 30 days had higher mortality (73% vs. 35%), and was also an independent predictor of mortality. The independent predictors of early hospital readmission were male gender, Charlson index ≥7, and MELD-Na score ≥15 at discharge. These predictors enabled the patients to be classified into two groups, one high risk (52.2% readmitted within 30 days), and one low risk (13.6% readmitted within 30 days).
Hyponatremia is a prognostic factor in cirrhosis. It has been associated with impaired health related quality life, as well as being a risk factor for increased morbidity and mortality before and after liver transplantation, and also as an increased risk of developing hepato-renal syndrome [
]. The incorporation of serum sodium into the model for end-stage liver disease (MELD-Na score) provided a more accurate survival prediction than the MELD alone in chronic liver disease [
]. Furthermore, the MELD-Na score has also been shown to be a feasible and independent prognostic predictor for both short- and long-term outcomes in HCC patients [
], and in a recent study it was a more valuable model than Maddrey discriminant function index to predict short-term mortality in patients with alcoholic hepatitis [
]. Another predictive role of MELD-Na score was found in our study, since it was associated with increased risk of early hospital readmission due to the poor prognosis of these patients, and to the increased susceptibility of developing complications from liver disease, along with the demand for medical care.
A new aspect in our study is the role of the Charlson Comorbidity index as an independent factor of early hospital readmission in patients with decompensated cirrhosis. It had been previously described as a risk factor, but in patients discharged from medical or surgical departments [
]. It seems reasonable to think that older people with several comorbidities have an increased susceptibility to develop adverse effects due to medication, increased polypharmacy, poor compliance with instructions at discharge, and increased demand for medical care, which may explain the high rate of early readmission. This index was also evaluated in the study by Volk et al. [
], without becoming an independent predictive factor of readmission. This difference could be explained due to the median age of their patients being lower than ours (54 vs. 66 years), which could lead to having fewer comorbidities.
], being male was an independent predictor of early hospital readmission. This could be explained by predominance of male patients in the study population, the differences in the evolution of chronic liver disease between men and women [
], although there were no significant differences between patients with and with no early hospital readmission. However, we thought it would be interesting to perform future, prospective, multicenter studies with a larger number of patients to determine their real impact on the early readmission.
The final model of independent predictors of early readmission had a c-statistic value of .76, indicating moderate predictive ability, which permitted us to create risk groups. In our study, 13.6% of low-risk patients were readmitted within 30-days, while in the high-risk group, 52.2% of patients were readmitted. In studies by Volk [
], risk groups were also created, with similar percentages of early hospital readmission (22% and 20% in low-risk groups, and 55% and 45% in the high risk groups, respectively). Although the risk factors for 30-day readmission found in our study are not modifiable (except MELD-Na in some cases), the identification of these allows us to identify patients at highest risk of early readmission during hospitalization. It could be useful for designing future specific surveillance and follow-up strategies at discharge for this group of patients, as a program of transitional interventions with closer specialized monitoring, in order to reduce the incidence of early hospital readmission, and improve survival and quality of life.
It is interesting to note that the main cause of admission was ascites; however the main cause for early hospital readmission was hepatic encephalopathy (not associated with portal hypertensive bleeding or SPB), which makes us suspect that the diuretic therapy, complications associated with this, and a lack of closer follow-up, with analytical and clinical evaluation of these patients, could be influencing the development of decompensation, especially hepatic encephalopathy. This is a point that should be taken into account in order to improve the management of these patients.
Limitations of this study include its retrospective nature. It was performed in a single hospital, which could limit the generalizability of our results, and, finally, the lack of validation of the final model predictors of early hospital readmission.
In conclusion, this study identified that early readmission is common in decompensated cirrhotic patients, and is associated with increased mortality. Male gender, advanced liver failure (MELD-Na), and Charlson index ≥7 are independent predictors of readmission within 30 days, which enabled patients to be classified into low and high risk with moderate accuracy. These results are potentially useful to guide future interventions aimed at reducing 30-day hospital readmission.
Conflict of interest
None declared.
Disclosure statement
Betty P. Morales receives a grant of “Germans Trias i Pujol” Health Sciences Research Institute (IGTP).
Acknowledgement
Betty P. Morales are supported by a grant of the “Germans Trias i Pujol” Health Sciences Research Institute (IGTP).
References
Everhart J.E.
Ruhl C.E.
Burden of digestive diseases in the United States part I: overall and upper gastrointestinal diseases.
Hepatic encephalopathy in chronic liver disease: 2014 practice guideline by the European Association for the Study of the Liver and the American Association for the Study of Liver Diseases.
Portal hypertensive bleeding in cirrhosis: risk stratification, diagnosis, and management: 2016 practice guidance by the American Association for the study of liver diseases.
30265-7&id=gr1.jpg){kind=link}
30265-7&id=gr2.jpg){kind=link}