PREDICTING 3-MONTH SURVIVAL AFTER TIPS - RETROSPECTIVE EVALUATION OF A PUBLISHED MODEL IN A TERTIARY CARE CENTRE

N Al-Bugami, K Tran¹, J Rawlinson¹, M Marcaccio², D Armstrong

Division of Gastroenterology, Divisions of Diagnostic Imaging¹/Surgery², Hamilton Health Sciences/McMaster University, Hamilton, Ontario

BACKGROUND: Transjugular intrahepatic portosystemic shunt (TIPS) is indicated for portal hypertension variceal bleeding and ascites. Because it is expensive with a 1-month mortality of 10-15% outcome predictors might help optimise resource utilization.
AIM: To assess the accuracy of published predictive model (Malinchoc, Hepatology 2000; 31:864) in assessing 3-month mortality after an elective TIPS procedure.
METHODS: A retrospective chart review of all patients who had undergone TIPS at a single, tertiary care centre serving a population of ~500,000 people between 1990 and 2001; TIPS patients' charts, identified from a diagnostic imaging database, were reviewed to determine baseline values for INR, albumin, creatinine, bilirubin and etiology (alcohol vs other causes) and survival, 3 months post-TIPS. When necessary, missing data were obtained from family physicians or hepatologists. Baseline values were used to calculate the predictive score: a score ³ 1.8 indicated a very poor prognosis.
RESULTS: Complete data were available for 39 of 46 patients identified. Overall 3-month mortality was 20/39 (51 %); 3-month mortality rates were 54% (7/13) in patients with a predictive score ³ 1.8 and 50% (13/26) in patients with a score < 1.8 (p=NS). The predictive model had low sensitivity (35.0%) and specificity (68.4%); positive and negative predictive values were 53.8% and 50.0%, respectively.
SUMMARY: TIPS for variety of indications has 3-month survival of 49%. A previously described predictive model had low sensitivity and specificity for retrospective identification of outcome in a 10-year series of TIPS performed at a tertiary care centre.
CONCLUSION: Accurate prediction of outcome following TIPS will require a prospective study that includes all patients assessed for TIPS eligibility; without further validation, the published model has limited value for predicting early mortality following TIPS in clinical practice.