The Model for Endstage Liver Disease MELD Score is Predictive of ICU Mortality in Patients with Cirr

1
The Model for End-stage Liver Disease (MELD)
Score is Predictive of ICU Mortality in Patients
with Cirrhosis

• Ali Massoumi, MD
• Division of Internal Medicine, Georgetown
  University Hospital

2
Background

• Chronic liver disease, including cirrhosis
  remains one of the major causes of death in the
  United States
• Based on CDC results, it is the ninth leading
  cause of death and one of the major causes of
  hospitalization
• Cirrhotics admitted to the ICU have been shown to
  have poor prognosis with a high rate of mortality

3
Background

• Prognostic models have been developed as a means
  of estimating the survival based on disease
  severity
• Often these models incorporate various critical
  laboratory values to compute scores with their
  respective prognostic scores
• Of these models, the Acute Physiology and Chronic
  Health Evaluation System (APACHE) has now become
  one of the standard indicators of prognosis of
  patients admitted to the intensive care unit
  (ICU) regardless of disease state

4
Background

• APACHE II includes underlying diseases and
  chronic health status, however it has been
  criticized as its mortality predictions have been
  less than observed.
• While the APACHE II scores can be used as a
  general prognostic model, few groups have
  investigated the role specific markers for
  cirrhotics admitted to the ICU
• Furthermore, APACHE II cannot accurately predict
  outcome for specific patient subgroups, such as
  liver failure

5
APACHE II Requirements

• Temp / MAP / HR / RR / A-a gradient
• PaO2 / HCO3/ pH / Na
• GCS / Creatinine / K
• Hct / Age / WBC
• Presence of chronic disease (e.g. COPD, class IV
  HF, portal hypertension)

6
Background

• Other models are more disease specific, as is
  the case for Model for End-Stage Liver Disease
  (MELD)
• MELD scores were developed to determine the
  severity of liver disease based on the patients
  serum bilirubin, serum creatinine, and the
  international normalized ration (INR)
• MELD Score 10 0.957 Ln(Serum creatinine mg/dL)
  0.378 Ln(Total bilirubin mg/dL) 1.12 Ln(INR)
  0.643.

7
Background

• The MELD score is currently a well accepted
  scoring system of chronic liver disease in order
  to predict survival
• The score is then used by the UNOS (United
  Network of Organ Sharing) in the allocation of
  livers for transplantation
• 3 month mortality If MELD lt 9? lt2 MELD
  gt40?71
• Although the MELD is now the accepted criteria
  for determining prognosis of overall mortality,
  we found no studies using the MELD score as an
  indicator of survival in the ICU for those not
  eligible for a transplant

8
Background

• Aggarwal et al evaluated cirrhotic patients
  admitted to the ICU in order to determine
  independent risk factors for prognosis
• The group found an overall MICU mortality of
  36.6, with independent predictors of mortality
  in the MICU found to be an APACHE III score gt90,
  the use of mechanical ventilation, and
  vasopressors
• Child-Pugh score has not been shown to be an
  independent predictive of mortality in the ICU

9
Background

• Intensive care unit (ICU) admission is associated
  with a high mortality rate in those with
  cirrhosis
• ICU specific prognostic scores are derived from
  analyses comprising of few cirrhotic patients,
  and thus may not be generalizable to this patient
  population

10
Goals of Study

• Compare the model for end stage liver disease
  (MELD) to the Acute Physiology and Chronic Health
  Evaluation II (APACHE II) score in predicting
  mortality.
• Define predictors of mortality in a cohort of
  cirrhotics admitted to an ICU

11
Methods

• We proceeded with a retrospective chart review
  of cirrhotic patients admitted to the Georgetown
  University ICU (SICU or MICU) over the span of
  seven years (1998-2005).
• Attention was paid to
• MELD score upon admission
• Interventions implemented during the ICU course
• Need for vasopressors or mechanical ventilation
• Presence of co-morbidities (cardiac disease,
  malignancies, renal failure)
• Patient outcome (death, withdrawal of care,
  discharge from the ICU).

12
Methods

• We excluded the following patients
• under the age of 18
• those with acute liver failure
• prior liver transplant (LT)
• those who received a LT during their index
  hospitalization

13
Methods

• Multivariate logistic regression was performed
  to evaluate admission factors associated with
  survival.
• The MELD and APACHE II scores were compared by
  the area under receiver operating characteristic
  curves (AUROC).

14
Results

• A total of 322 patients were initially screened.
  104 patients were studied, of whom 65 were male,
  and 32 had Hepatitis C
• The principal indication for ICU admission was
  gastrointestinal bleeding (34) and the majority
  of patients (71) had renal failure

15
Etiology of Cirrhosis
16
Admitting Diagnosis
17
Results

• Nineteen percent of patients did not survive the
  ICU admission, and a further 50 were withdrawn
  from intensive care based on perceived futility.
• This resulted in an overall survival rate of 31.

18
Outcome
19
Variables During ICU Course
20
Results

• Of those that did not survive, the average MELD
  score was 25 compared to 21 in those who survived
  (p0.02)
• The MELD score was shown to be an independent
  predictor of mortality (OR 0.824 95 CI,
  0.761-0.893 Plt0.0001)

21
(No Transcript)
22
Results

• Of those that did not survive, the average APACHE
  II score was 21, compared to 11 to those who did
  survive
• The APACHE II score was shown to be an
  independent predictor of mortality (OR 0.789 95
  CI, 0.715-0.871 Plt0.0001),

23
(No Transcript)
24
(No Transcript)
25
MELD vs APACHE II

• MELD score was shown to be superior to the APACHE
  II score
• (AU-ROC 0.82 vs 0.71, respectively)

26
Results

• MELD score vs Survival after adjusting for other
  variables
• Including etiology, age, sex, sepsis, ARF,
  pressors, admitting dx, co-morbidity,
  interventions
• The probability of survival is related to the
  collection of variables (Plt0.0001)
• However, adjusting for individual predictors, the
  MELD was not predictive of survival (OR1.027,
  95 CI 0.855-1.233, P 0.8197

27
Results

• On multivariate analysis, co-morbidity and MELD
  score gt30 were found to be predictive of
  non-survival.
• Co-morbidity OR0.884, 95 CI 0.787-0.993, P
  0.0374
• Co-morbidity was defined as CAD, malignancies, or
  history of renal failure

28
Limitations / Future Goals

• Mortality in the ICU can be multi-factorial
• Increase sample size
• To further evaluate the role of individual
  predictors in relation to the MELD score
• Expand comparison to other models such as APACHE
  III and SOFA (sequential organ failure assessment
  ) score
• An increase in the SOFA score within the first 48
  hours has been shown to be a poor predictor
• Incorporates bili, MAP, creatinine, GCS,
  respiratory evaluation, platelets

29
Conclusions

• Our study indicates that the admission MELD score
  in cirrhotics is an independent prognostic
  indicator of ICU mortality, and is superior to
  the more cumbersome model such as the APACHE II

30
Conclusions

• Patients with a MELD score exceeding 30, with
  co-morbid illness, are unlikely to survive their
  ICU admission

31
Conclusions

• It is hoped that delineation of a subgroup in
  whom specialized therapies are futile will
  optimize utilization of interventions in patients
  who will derive maximal benefit from such care.

32
Acknowledgements

• Kirti Shetty, MD¹
• Muhannad Hafi
• Amy Lu, MD¹
• Lynt B. Johnson, MD¹
• ¹Division of Transplantation, Georgetown
  University Hospital

33
References

• Deaths and Hospitalizations from Chronic Liver
  Disease and Cirrhosis -- United States, 1980-1989
  CDC Vital and Health Statistics Series. January
  08, 1993 / 41(52)969-973
• Predictors of Mortality and Resource Utilization
  in Cirrhotic Patients Admitted to the Medical ICU
  Anjana Aggarwal
• Pugh RN et al.Transection of the oesophagus for
  bleeding oesophagus varices. Br J Surg.
  1973.60646-649
• Outcome of patients with cirrhosis requiring
  intensive care unit support prospective
  assessment of predictors of mortality. Singh N,
  Gavoski T, Wagener MM, Manno IR
• Predictors of long-term mortality in patients
  with cirrhosis of the liver admitted to a medical
  ICU. Gildea TR, Cook WC, Nelson DR, Aggarwal A,
  Carey W, Younossi ZM, Arroliga AC. Chest. 2004
  Nov126(5)1598-603