Context: The long-term survival of patients with cirrhosis of the liver admitted to the ICU has not been described.
Objective: The main objectives were to determine the 1-year and 5-year mortality rates of a cohort of patients admitted to a medical ICU (MICU), and to identify the risk factors that may predict long-term outcomes.
Design: This is a cohort study. We used a model-building (MB) and model validation (MV) procedure that has previously been described to determine the risk factors for overall mortality.
Settings: An MICU in a major referral medical center.
Patients: Four hundred twenty consecutive patients admitted to the ICU from January 1, 1993, through October 31, 1998, met the inclusion criteria of diagnosis of liver failure, cirrhosis, chronic liver disease, variceal bleeding, hepatic encephalopathy, or hepatorenal syndrome. Patients with acute liver failure who had undergone liver transplantation, or candidates for orthotopic liver transplantation were excluded.
Results: The 1-year mortality rate was 69%, and the 5-year mortality rate was 77%. The median survival time was 1 month. The independent predictors of mortality in patients in the MB group who retained their significance in the MV group were as follows: an acute physiology, age, and chronic health evaluation (APACHE) III score of [greater than or equal to] 90 (hazard ratio [HR], 2.2; 95% confidence interval [CI], 1.6 to 2.8; p < 0.0001), the use of pressors (HR, 2.5; 95% CI, 1.9 to 3.2; p < 0.0001), and jaundice (HR, 1.7; 95% CI, 1.4 to 2.2; p < 0.0001). Patients with all three risk factors (ie, APACHE III score [greater than or equal to] 90, use of pressors, and jaundice) had a 92% 1-month mortality rate compared to 11.2% for patients with no risk factors.
Conclusions: Patients admitted to an MICU with underlying cirrhosis who arc not eligible for liver transplantation have a poor long-term prognosis, even if they survive the ICU admission, particularly as the number of risk factors increases.
Key words: acute physiology and chronic health evaluation II; cirrhosis; ICU; mortality; prognosis
Abbreviations: APACHE = acute physiology, age, and chronic health evaluation; CI = confidence interval; HR = hazard ratio; MB = model building; MICU = medical ICU; MV = model validation
Cirrhosis and chronic liver diseases are important causes of morbidity and mortality in the United States. Cirrhosis is the 12th most common cause of death and the most common cause of non-malignancy-related death among patients with diseases of the digestive tract. (1) Critically ill patients with chronic liver diseases admitted to the ICU have a low survival rate, with mortality rates ranging from 50 to 100%. (2-5) The mortality rate is higher in those who develop multiple organ dysfunction, and in patients in septic shock a mortality rate of 100% has been reported. (2-6) In addition to the high mortality rate, the management of these patients is resource-intensive, and a significant percentage (36.7%) of the total cost of ICU care for cirrhotic patients is spent on those who do not survive. (7,8)
Several prediction models of mortality in the hospital and ICU have been published. (4,8-11) In a previous study (4) using a model-building (MB) and model validation (MV) analysis, we reported that an acute physiology, age and chronic health evaluation (APACHE) III (12) score of [greater than or equal to]90, the use of pressors, mechanical ventilation, and the presence of renal failure were predictors of short-term medical ICU (MICU) and hospital mortality. However, the long-term survival of patients with cirrhosis of the liver who have been admitted to the ICU and are not candidates for orthotopic liver transplantation has not been studied appropriately. In a recent multicenter study of 538 patients admitted to five hospitals as part of the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment study, the 1-year death rate was 72%. (9) In this study, we assessed the 1-year and 5-year mortality rates for patients admitted to an MICU with cirrhosis and identified the risk factors during the ICU admission that predict long-term mortality in those who survive the index admission.
MATERIALS AND METHODS
The setting and subjects of this retrospective cohort of study have been described previously. (4) This was a cohort of 420 consecutive patients admitted to an MICU from January 1, 1993, through October 31, 1998, who had a diagnosis of chronic liver failure, cirrhosis, chronic liver disease, variceal bleeding, hepatic encephalopathy, or hepatorenal syndrome. The patients were identified from a database maintained by the Critical Care Medicine section at the Cleveland Clinic Foundation. All patients were [greater than or equal to] 18 years of age and had histologically proven or clinically diagnosed cirrhosis (eg, portal hypertension with ascites, esophageal varices, or encephalopathy). No patient had acute fialminant liver failure or had undergone a liver transplantation, and none was a candidate for orthotopic liver transplantation at the time of their ICU admission.
Main Outcome Measure
The main outcome measure of this study was the all-cause mortality rate at 1 and 5 years. The Social Security Death Index (13) was used to identify the date of death and therefore the length of survival beyond the index MICU admission. This search engine is a validated, free-of-charge method of determining the vital status of subjects. (14) The cause of death was not assessed. The Cleveland Clinic Institutional Review Board waived the need for informed consent. Individual patient consent was not sought.
A retrospective database consisting of 202 variables that assessed hospital admission status and hospital course, in addition to mortality outcome, hospital utilization, and hospital charges, was utilized for the analyses. The same clinical variables as those used in a previous study (4) were applied to determine the long-term mortality rate of those who survived the initial admission. Patients without evidence of death were censored at the time they were last known to be alive.
We used the MB and MV procedure previously described to determine the risk factors for overall mortality. (4) Cases were randomized into two groups, MB and MV, and were stratified by calendar year of hospitalization. Of the 236 patients who survived the initial hospitalization, 100 were included in the MB group and 136 in the MV group. Forward stepwise selection was used to determine potential risk factors in the MB group, and then they were included in the final model if they remained significant in the MV group. This allowed a reduction of type I error rates (the type I error was 0.00125 per variable using these methods), while providing the flexibility to build models in the MB group. Cox proportional hazards regression was used for this analysis of long-term mortality rates (ie, from the date of MICU admission to the date of the last follow-up), with resulting hazard ratios (HRs) and 95% confidence intervals (CIs) calculated.
The subset of patients discharged from the hospital after the initial admission was analyzed separately to determine whether risk factors could be identified for post-hospital discharge mortality risk (using MICU discharge as the initial time point) because much of the overall mortality analysis was based on in-hospital mortality. In addition to the baseline and hospital course variables included in the analysis of survival from hospital admission, risk factors for this analysis also included hospital charges and hospital length of stay (both were log-transformed) to determine whether utilization was related to subsequent mortality for patients who were discharged from the MICU. Kaplan-Meier survival curves were used to illustrate survival trends. All calculations were performed with a statistical software package (SAS, version 8.1; SAS Institute Inc; Cary, NC).
The characteristics of the patients are shown in Table 1. (15) Of the 420 patients included in the analysis, 184 (44%) died in the hospital during the index MICU admission (ie, the first MICU admission for each patient during the study period) and another 139 (33%) died during the follow-up period, for a total of 323 deaths. The follow-up period was between 3 and 8 years after the index admission. Thus, the 5-year observed mortality rate was 77% (total deaths/total patients [unadjusted for censoring]), and the Kaplan-Meier estimate of the 5-year mortality rate was > 90%. Of the 323 deaths, 290 (69%) occurred within 1 year of the index MICU admission (Fig 1). Median survival time was 1 month (Fig 1).
[FIGURE 1 OMITTED]
We analyzed the predictors of mortality (ie, in-hospital and post-MICU discharge mortality) that were present during MICU admission for those patients who survived to be discharged from the MICU. The following three independent predictors of mortality in the MB group kept their significance in the MV group: APACHE III score of [greater than or equal to] 90 measured on day 1 (HR, 2.2; 95% CI, 1.6 to 2.8; p < 0.0001); the use of pressors at any time during the ICU stay (HR, 2.5; 95% CI, 1.9 to 3.2; p < 0.0001), and the presence of jaundice (HR, 1.7; 95% CI, 1.4 to 2.2; p < 0.0001). Jaundice had not been a risk factor in the previous study of MICU or hospital mortality (4) but became a predictor of long-term mortality for patients after MICU or hospital discharge. We further assessed the association between jaundice and the level of total bilirubin (Wilcoxon rank-sum test) for the 405 patients for whom we had complete information. The 199 patients who had jaundice recorded in the physical examination had significantly higher bilirubin levels (median total bilirubin level, 8.7 mg/dL; 25th to 75th percentile, 4.2 to 19.7) vs the 206 patients in whom jaundice had been absent or was not recorded during the physical examination (2.2 mg/dL; 25th to 75th pereentile, l.2 to 4.1; p < 0.001).
To quantify the effect of the multiple risk factors, patients were classified from no risk factor to up to three risk factors. The three factors could be added in such a simple fashion because the HRs associated with the presence of each of the three risk factors were approximately equal, with a great deal of overlap in their CIs. The survival of patients was decreased as the number of risk factors increased, and each incremental increase of the number of risk factors was statistically significant (Table 2). Patients with all three risk factors (ie, APACHE III score of [greater than or equal to] 90, use of pressors, and jaundice) had a 92% 1-month mortality rate compared to an 11.2% 1-month mortality rate if the patients had no risk factors (Fig 2). The etiologies of the liver disease in patients with all three risk factors are given in Table 3. This group was similar to the overall population in the study.
[FIGURE 2 OMITTED]
Two hundred thirty-six patients (56%) survived the hospital admission. Forty-seven patients (20%) were discharged to a facility other than home, and 189 patients (80%) were discharged to home. The MB and MV groups using only the hospital survivors indicated that APACHE III score and discharge location were the two significant risk factors for post-hospital survival risk factors (Table 4). Patients who were discharged to home and had an APACHE III score of < 90 had a 1-year mortality rate of 33%. The 1-year mortality rate for the patients who were discharged to home, but had an APACHE III score of [is greater than or equal to] 90, was 67%. Similarly, the 1-year mortality rate for the patients who were discharged to somewhere other than home was also 67%. The hospital discharge disposition and the APACHE III score were significantly related (p < 0.001) to post-hospital discharge survival. The median survival time for patients with an APACHE III score of < 90 who had been discharged to home was 17 months. The median survival time for patients with an APACHE III score of [is greater than or equal to] 90 and had been discharged to a facility other than home was 4 months (Fig 3). Hospital length of stays and charges were not significantly related to post-hospital discharge survival.
[FIGURE 3 OMITTED]
The most important finding of this retrospective study of 420 patients who were admitted to an MICU was that the observed mortality rates at 1 and 5 years were 69% and 77%, respectively. The median survival time was 1 month (Fig 1). The impact of this figure represents early mortality for the patients in the MICU (36.6%) and hospital (44%) prior to discharge. In those patients who survived the initial admission to the ICU, the combination of an APACHE III score on the day of MICU admission of [is greater than or equal to] 90, the use of pressors at any time, and the presence of jaundice was associated with a 92% 1-month mortality rate. Those patients with an APACHE III score of < 90 who had been discharged to home had a median survival time of 17 months.
The presence of jaundice in combination with an APACHE III score of [is greater than or equal to] 90 and the use of pressors were associated with high short-term and long-term mortality rates. Patients who had jaundice recorded during the physical examination had a higher level of total bilirubin, probably making it easier to be noticed. Furthermore, a higher level of total bilirubin may represent more advanced cirrhosis or a superimposed liver injury caused by multisystem organ dysfunction. Although the use of pressors at baseline may not intuitively seem as if it should be related to long-term survival, the use of pressors may be a surrogate measure of cardiovascular function rather than simply a concomitant use of medication. It is interesting that the presence of jaundice noted clinically and the use of pressors were highly significant, indicating that the APACHE III components related to these factors were perhaps underweighted in this group of patients.
Patients admitted to an MICU with underlying cirrhosis but who were not eligible for liver transplantation have poor prognosis even if they survive the ICU admission. The short-term mortality rate of our population is similar to that from previous reports. (3) Wehler et al (8) reported a 36% ICU mortality rate and a 46% hospital mortality rate in the same study period in patients with disease severity that was similar to that seen in our patients. In cirrhotic patients admitted to the ICU in shock, the mortality rate has been reported to be as high as 100%. (6) The poor long-term survival rate reported in this study is hardly unique to patients with cirrhosis. Recently, several small studies (16-19) have reported mortality rates up to 100% in patients with idiopathic pulmonary fibrosis who had been admitted to the ICU and who required invasive mechanical ventilation.
Besides poor hospital survival, evidence suggests that patients who are discharged to home have a poor quality of life. Roth et al (20) assessed the relative quality of life of patients with end-stage liver disease with cirrhosis from the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment database. Seventy-two percent of the patients who were alive at 1 year after hospitalization reported their quality of life as fair or poor. Even though pain is not considered to be a major symptom of end-stage liver disease, one third of those patients reported serious pain. In addition, three quarters of the patients stated that they would rather die than be in a coma, and one third stated that they would rather die than be in pain or confused for the rest of their lives. (20) In contrast, one study (21) of unselected patients surviving admission to an ICU found that many had resumed baseline quality of life and would undergo ICU admission again if necessary.
The patients in the cohort reported on here were not candidates for liver transplantation. Although our patients were not candidates or listed for transplantation at the time of admission to the ICU, and the role of transplant is not fully explored in our article, the long-term survival of liver transplant recipients has been well-documented. We recently reported (22) on the outcomes of patients who had undergone a liver transplantation and had been admitted at any time to the ICU during the hospitalization when they received the liver transplant. Patients and patient/ graft survival at the 4-year follow-up were similar to the survival of patients who were ambulatory at the time of the liver transplant and to patients who were hospitalized but did not require ICU admission.
One of the limitations of our study was that our patient population was subject to a referral bias because we are a tertiary referral center that can perform high-risk endoscopic procedures, transjugular intrahepatic portosystemic shunt procedures, and liver transplants. The second important limitation was that this is a retrospective analysis from a single institution. Prospective studies from different institutions to confirm these findings are needed.
Our study nonetheless emphasizes that when patients with cirrhosis become sick enough to require ICU care and are not candidates for liver transplantation, they have a very high likelihood of dying even after the ICU is no longer needed. This information can be used to give patients and families an estimate of prognosis based on commonly and easily gathered data in an ICU setting.
(1) Hoyert DL, Arias E, Smith BL, et al. Deaths: final data for 1999. Natl Vital Stat Rep 2001; 49:1-113
(2) Singh N, Gayowski T, Wagener MM, et al. Outcome of patients with cirrhosis requiring intensive care unit support: prospective assessment of predictors of mortality. J Gastroenterol 1998; 33:73-79
(3) Kress JP, Rubin A, Pohlman AS, et al. Outcomes of critically ill patients denied consideration for liver transplantation. Am J Respir Crit Care Med 2000; 162:418-423
(4) Aggarwal A, Ong JP, Younossi ZM, et al. Predictors of mortality and resource utilization in cirrhotic patients admitted to the medical ICU. Chest 2001; 119:1489-1497
(5) Zimmerman JE, Wagner DP, Seneff MG, et al. Intensive care unit admissions with cirrhosis: risk-stratifying patient groups and predicting individual survival. Hepatology 1996; 23:1393-1401
(6) Moreau R, Hadengne A, Soupison T, et al. Septic shock in patients with cirrhosis: hemodynamic and metabolic characteristics and intensive care unit outcome. Crit Care Med 1992; 20:746-750
(7) Wong LL, McFall P, Wong LM. The cost of dying of end-stage liver disease. Arch Intern Med 1997; 157:1429-1432
(8) Wehler M, Kokoska J, Reulbach U, et al. Short-term prognosis in critically ill patients with cirrhosis assessed by prognostic scoring systems. Hepatology 2001; 34:255-261
(9) Cooper GS, Bellamy P, Dawson NV, et al. A prognostic model for patients with end-stage liver disease. Gastroenterology 1997; 113:1278-1288
(10) Zoli M, Cordiani MR, Marchesini G, et al. Prognostic indicators in compensated cirrhosis. Am J Gastroenterol 1991; 86:1508-1513
(11) Butt AK, Khan AA, Alam A, et al. Predicting hospital mortality in cirrhotic patients: comparison of Child-Pugh and Acute Physiology, Age and Chronic Health Evaluation (APACHE III) scoring systems. Am J Gastroenterol 1998; 93:2469-2475
(12) Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991; 100:1619-1636
(13) Social Security Administration. Social Security death index. Available at: http://ssdi.genealogy.rootsweb.com/. Accessed October 15, 2004
(14) Sesso HD, Paffenbarger RS, Lee IM. Comparison of National Death Index and World Wide Web death searches. Am J Epidemiol 2000; 152:107-111
(15) Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373-383
(16) Fumeaux T, Rothmeier C, Jolliet P. Outcome of mechanical ventilation for acute respiratory failure in patients with pulmonary fibrosis. Intensive Care Med 2001; 27:1868-1874
(17) Saydain G, Islam A, Afessa B, et al. Outcome of patients with idiopathic pulmonary fibrosis admitted to the intensive care unit. Am J Respir Crit Care Med 2002; 166:839-842
(18) Stern JB, Mal H, Groussard O, et al. Prognosis of patients with advanced idiopathic pulmonary fibrosis requiring mechanical ventilation for acute respiratory failure. Chest 2001; 120:213-219
(19) Blivet S, Philit F, Sab JM, et al. Outcome of patients with idiopathic pulmonary fibrosis admitted to the ICU for respiratory failure. Chest 2001; 120:209-212
(20) Roth K, Lynn J, Zhong Z, et al. Dying with end stage liver disease with cirrhosis: insights from SUPPORT; Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment. J Am Geriatr Soc 2000; 48:S122-S130
(21) Wehler M, Martus P, Geise A, et al. Changes in quality of life after medical intensive care. Intensive Care Med 2001; 27:154-159
(22) Aggarwal A, Ong JP, Goormastic M, et al. Survival and resource utilization in liver transplant recipients: the impact of admission to the intensive care unit. Transplant Proc 2003; 35:2998-3002
* From the Departments of Pulmonary and Critical Care Medicine (Drs. Gildea and Arroliga), Internal Medicine (Drs. Cook and Aggarwal), Biostatistics (Mr. Nelson) and Gastroenterology (Dr. Carey), Cleveland Clinic Foundation, Cleveland, OH; and the Center for Liver Diseases (Dr. Younossi), Inova Fairfax Hospital, Falls Church, VA.
Manuscript received June 10, 2003; revision accepted May 12, 2004.
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail: firstname.lastname@example.org).
Correspondence to: Alejandro C. Arroliga, MD, FCCP, Department of Pulmonary and Critical Care Medicine, The Cleveland Clinic Foundation, 9500 Euclid Ave, G62, Cleveland, OH 44195; e-mail: email@example.com
COPYRIGHT 2004 American College of Chest Physicians
COPYRIGHT 2004 Gale Group