Severity Index for patients with Liver Cirrhosis disease

1
Severity Index for patients with Liver Cirrhosis
disease

• Health Decision Analysis 730-001
• Yokabid Worku
• Ymelat_at_hotmail.com
• 02-19-04

2
Outline

• Liver Cirrhosis and its measures of damage
• Selecting Attributes in accordance of their
  importance
• Review of Attribute
• Description of Attributes
• Attribute levels and assigning values
• Calculation Attributes weights
• Modeling Severity index and scenarios
• Severity index for Liver Cirrhosis disease
• What the results mean
• Correlation Graph
• Evaluation of the Model

3
Liver Cirrhosis and its measures of damage

• Liver Cirrhosis is the result of chronic liver
  disease that causes scarring of the liver and
  liver dysfunction. Common causes include alcohol
  abuse and hepatitis C infection. Patients with
  Liver disease suffer several illnesses.
• In this search, a value model was used to create
  a severity index for Liver Cirrhosis patients. A
  value model is the best tool to prioritize for
  evaluation of the diseases severity level. This
  model will help quantify preferences. These
  preferences were quantified by examining the
  various attributes of the disease. To clearly
  show the importance of each attribute, a severity
  index was constructed based on the input of
  expert.

4
Selecting Attributes

• After a brief introductory, the expert was
  notified of the models purpose and how it would
  be developed.
• To help understand which attributes could be used
  to indicate severity level, the expert was asked
  to recall the worst and the best attributes of
  the disease.
• By contrasting patients, attributes were
  solicited, thus helping single out those
  attributes that truly affect prognosis. At this
  point, the expert was asked to identify for
  additional attributes.
• All the attributes were arranged in order of
  their hierarchy.

5
Review of Attributes

• Tests were conducted to ensure all attributes
  were exhaustive, important, and not dependent and
  not redundant. Further more, Attributes were
  minimized to achieve independency in preference.
• Discussion with the expert led to the following
  set of eight attributes.
• Leg Edema
• Red Blood Cell Count (Hemoglobin Count)
• Ascites
• Bleeding esophageal varices
• CNS (Hepatic encephalopathy)
• Jaundice (yellow discoloration)
• Platelets count (Thrombocythemia)
• Albumin (protein)

6
Description of Attributes

• Leg Edema- Swelling of the ankles/ feet/ legs,
  caused by abnormal buildup of fluid in these
  areas.
• Red Blood Cell Count (Hemoglobin count)-
  inadequate number of circulating red blood cells,
  caused by premature destruction of red blood
  cells.
• Ascites- excess fluid in the space between the
  lining of the abdomen.
• Bleeding esophageal varices-dilated veins in the
  walls of the lower part of the esophagus and
  sometimes the upper part of the stomach.
• CNS (Hepatic encephalopathy)- is brain and
  nervous system damage characterized by changes in
  consciousness and behavior changes.
• Jaundice (yellow discoloration)- yellow color in
  the skin or the eyes, caused by too many red
  blood cells (bilirubin) retiring.
• Platelets count (Thrombocythemia)- abnormality of
  a very immature cell in the bone marrow,
  producing blood cells affecting the platelets.
• Albumin (protein)- clear fluid portion of blood
  (serum), protein of the highest concentration in
  plasma.

7
Table1 Shows Attribute levels ranging from best
to worst, and the value assigned for each level.
Attributes were listed in the order of their
importance to show severity level for Liver
Cirrhosis disease. Each level was assigned a
value relative to its actual value using same
scale.
8
Table 2 Calculation of attributes weightThe
method used to weigh the attributes was by
assessing the ratio of the importance of two
attributes. The least important attribute was
assigned 10 points. Then, the expert was asked
to estimate the relative importance of the other
attributes in relation to Leg Edema.
9
Cont. on Calculation of Weights

• As shown in Table 2, the weights for the
  remaining attributes were calculated then
  followed by normalizing the total weight to 1.
• The weights correlated quite well with the
  experts judgments of the rank order, that
  confirmed accuracy of the attribute levels and
  their importance.

10
Modeling Severity index and Scenarios

• In order to increase confidence in the model, and
  to establish the validity of a model, the model
  should stimulate the judgment of the expert thus,
  scenarios were generated.
• An attribute level was selected randomly from all
  eight attributes to come up with a scenario.
• Double anchored equation was used to produce one
  overall score for every scenario.
• Then, the expert was asked to score these
  hypothetical case profiles described only by
  variables included in the model.
• The method was used to show relation, to have
  confidence in the model and to achieve face
  validity.

11
Table 4 Severity index for Liver Cirrhosis
disease
12
Cont. from Table 3 What the Result mean

• The model adequately predicted the experts
  judgments, this increased confidence in the
  model.
• Higher scores indicated high severity level thus,
  less survival time.
• Lower scores indicated low severity level thus,
  high survival time.
• The overall model scores calculated showed
  severity level for each scenario.
• This method established validity for this model
  measuring the severity index level for Liver
  Cirrhosis patients.

13
Correlation GraphFigure 1 Shows the
correlation between the expert rating v.s. Model
scores of severity level for eight scenarios.
14
Cont. on Correlation graph

• The correlation between the model score and the
  experts rating was 0.933 suggesting high
  agreement between the model and the experts
  perception.
• A correlation closer to 1.0 implies greater
  agreement proving, the high correlation had
  stimulated the judgment of the expert.

15
Evaluation of the Model

• The expert considered the model simple to use and
  face valid.
• The model had corresponded valid with other
  measures such as, correlation graph which proved
  to have stimulated the experts judgment on the
  given scenarios.
• The model score and the experts rating
  correlated positively. This is true, because as
  the model scores increased, the experts ratings
  showed tendency to increase correspondingly in a
  similar trend.
• Techniques used in modeling values adequately
  served in measuring the severity level for Liver
  Cirrhosis disease by allowing to compare the
  scores between the experts ratings and model
  scores.

16
Reference

• The expert in question is Ambachew Woreta M.D.,
  whose primary practice is in the field of
  internal medicine. Dr. Woretas familiarity with
  this field served as justification for his
  involvement in this project.