Decompensated Cirrhosis Outcomes Ascertainment and Adjudication in the VACS

1
Decompensated Cirrhosis Outcomes Ascertainment
and Adjudication in the VACS

• Vincent Lo Re, MD, MSCE
• Division of Infectious Diseases
• Center for Clinical Epidemiology and
  Biostatistics
• University of Pennsylvania School of Medicine
• Philadelphia VA Medical Center

2
Decompensated Cirrhosis as a Clinical Outcome

• Few studies examined decompensated cirrhosis
• Lacked formal validation methods
• Evaluation of clinical outcome important
• Does not require liver biopsy
• Allows inclusion of all patients with chronic
  liver disease

3
Specific Aims

• Establish case definition of decompensated
  cirrhosis
• Outline process for ascertainment and
  classification of decompensated cirrhosis

4
Study Design / Setting

• Observational cohort study
• Setting VACS-8
• Large number with chronic liver diseases
• Data relevant to decompensated cirrhosis
  collected
• Follow-up in the VACS sufficiently long to allow
  for development of outcomes

5
Study Subjects

• Eligible All subjects enrolled in VACS
• Identification and selection
• Based on ICD-9 codes for decompensated cirrhosis
• ICD-9 codes recorded up to 1 yr before through 6
  mo after VACS enrollment
• Subjects with at least one ICD-9 code for
  decompensation were included in study sample

6
Outcome Hepatic Decompensation
Note Presence of any one diagnosis in record
represented an outcome
7
Data Collection Medical Record Data

• Subjects with ICD-9 codes ? chart review
• Medical records downloaded from VA centers
• Text files created for
• Radiographic studies
• Surgical pathology reports
• Laboratory results
• Progress notes
• Files imported into Access database for review
• Database placed on secure VACS server

8
Data CollectionStandardized Data Forms / Review

• Detailed search instructions
• Guide how and what to abstract
• Abstraction -1 year earliest ICD-9 date ?
  follow-up
• Abstract relevant data, dates recorded
• Imaging studies (abdo ultrasound, CT, MRI
  results)
• Ascites (quantity), liver masses (number,
  dimensions)
• Lab data AFP, ammonia, peritoneal fluid
• Liver biopsy stage, cirrhosis, cancer
• Endoscopy varices (location, bleeding, banded)
• Notes encephalopathy, variceal bleed, asterixis

9
Data CollectionAdjudication of the Outcome

• To facilitate review, data forms were
• Scanned in pdf format
• Downloaded onto secure VACS server
• Two endpoints arbitrators reviewed data forms
• Specialized in care of pts with chronic liver
  disease
• Determined whether event occurred / date
• In event of disagreement, third arbitrator acted
  as tie-breaker

10
Results Subject Selection
5,970 VACS Subjects
82 Subjects had relevant ICD-9 code -1 yr
through 6 mo after VACS enrollment
2 excluded No records available
80 Subjects Included in the Study
11
Results Patient Characteristics
12
Results ICD-9 Codes
Subjects may have had more than one ICD-9 code
recorded
13
Results Procedures

• Avg. time to review record 11 min. (range,
  3-38 min.)

14
Results Diagnoses
15
Results Endpoint Adjudication

• Arbitrator 1 47 cases / 80 (59)
• Arbitrator 2 46 cases / 80 (58)
• Rate of agreement 78/80 98
• Kappa 0.91 (substantial agreement)

Arbitrator 2
16
Conclusions

• Established working definition for decompensated
  cirrhosis in the VACS
• Used ICD-9 codes to screen for cases
• Demonstrated feasibility of centralized medical
  record review for abstractions
• Created standardized method to adjudicate
  decompensated cirrhosis outcomes

17
Future Work

• Improve screening with lab data
• Total bilirubin albumin INR
• APRI FIB-4
• Gradation of outcomes definite, probable, no
• Identify and adjudicate incident cases
• K award Specific Aims
• Examine risk factors for decompensated cirrhosis
• Predictive index to stratify HIV/HCV pts by risk
  of progression to decompensated cirrhosis
• Use in future studies?

18
Acknowledgements

• VACS Liver Group
• Amy Justice
• Joseph Lim
• Matthew Goetz
• Joseph Goulet
• Shawn Fultz
• Adeel Butt
• Marina Klein
• David Rimland
• Maria Rodriguez-Barradas
• Cynthia Gibert
• Kendall Bryant

• VACS
• Cynthia Brandt
• Woody Levin
• Farah Kidwai
• Michael Rambo
• Melissa Skanderson
• Kristin Mattocks
• Kathleen McGinnis
• Faith Whitsett
• Penn
• Brian Strom
• Russell Localio
• K. Rajender Reddy
• Jay Kostman
• NIH K01 AI 07001