Sunday Poster Sessions, October 29, 2006
Portal Hypertension and Other Complications
T. D. Boyer; J. Henderson;
A. Heerey; K. Abu-Elmagd ;
J. Galloway; L. F. Rikkers; L. Jeffers.
Introduction:
DSRS (5.5%) and TIPS (10.5%) have
been shown in a randomized controlled trial to be equally effective in
preventing rebleeding from esophageal varices
(Gastroenterology 130:1643;2006). We have examined the cost-effectiveness of
each of these two approaches.
Methods:
Quality of life (QOL) was measured
using SF-36 preceding randomization and yearly thereafter and converted to
utility using the SF-6D. Cost utility analysis performed using Data Triage ®.
All data were collected prospectively. DRG and CPT codes for both in- and
out-patient events and interventions were obtained for each patient. Cost of
medications and laboratory tests were also determined. Medicare costs from 2003
were used and inflated to 2004 costs using the medical care inflation index
from the Bureau of Labor Statistics. Costs using
coated stents were estimated using different rates of
stenosis. Incremental cost effectiveness ratios (ICERs) were determined at 1, 3 and 5 years.
Results:
Cost of TIPS exceed the cost of
DSRS at all time points but not significantly ($44,772 vs. $41,941; $67,760 vs.
$57,921; $84,033 vs. $66,685). For the patients who survived, the out-patient
cost of TIPS was significantly greater than DSRS due to re-interventions. A
trend towards better survival for TIPS early in the study influenced the cost
utility analysis in favor of TIPS for the ICERs per quality adjusted life year (QALY) saved (year 1
$101,470; year 2 $140,357; year 5 $61,000). By year 5 TIPS was only marginally
more cost effective, using a threshold of $50,000/QALY. Using modeling for
coated stents to assess effect of lowering
re-intervention rates, the cost-effectiveness of TIPS increased only slightly.
Summary: TIPS cost more than DSRS, especially in the out-patient setting. These
costs reflect the higher rates of re-intervention seen with TIPS. However,
health related QOL measured in QALYS were greater at all time points for the
TIPS group due to the early, but nonsignificant,
survival advantage observed in the study. The ICER favored
TIPS and at 5 years TIPS would be considered moderately cost-effective. However
due to the wide confidence intervals there was no significant difference in
cost-effectiveness at any time point.
Conclusion:
This is the first study that
defines prospectively the costs of preventing rebleeding with TIPS or DSRS in
Child’s A/B patients who have failed pharmacologic and endoscopic therapy.
Although TIPS was more expensive there was no significant difference in the
cost-effectiveness of either approach.
Supported by a grant from NIDDK: DK050680.
Authors have no conflicts to report.
P.
Angeli; L. Dallagnese; S. Fasolato; E. Mazza; F. Salinas;
S. Donŕ; S. Fagiuoli; A. Sticca; G. Zanus; U. Cillo; C. Destro; A. Gatta.
Introduction:
The role of bacterial infections in
the pathogenesis of renal failure has been assessed mainly in patients with
SBP.
Aim:
The aim of the study was to
investigate the prevalence, the clinical course, and the outcome of renal
failure induced by all the types of bacterial infections in patients with
cirrhosis and ascites.
Methods:
399 patients, who had been
consecutively admitted to the three major hospitals of Padova
(Italy) during the first 6 months of 2005 were included in the study.
Results:
233 patients out of 309 patients
(75.4%) had evidence of ascites on admission. In 104 out of 233 (44.6%)
cirrhotic patients with ascites a bacterial infection was diagnosed. UTI and
pneumonia were the most frequent types of infections followed by SBP. A
bacterial infection-induced renal failure was observed in 35 out of 104 (33.6
%). The prevalence of bacterial infection-induced renal failure was
significantly higher in sub-diaphragmatic bacterial infections than in supradiaphragmatic infections or in other infections
(p<0.0001). The progressive form of bacterial infection-induced renal
failure was only precipitated by the sub-diaphragmatic bacterial infections in
these patients (p<0.0001). Either the stable or the progressive form of
renal failure was induced by all types of sub-diaphragmatic bacterial infection
and not only by PBS in cirrhotic patients with ascites. Finally, renal failure
was induced by sub-diaphragmatic bacterial infections in spite of the
resolution of the infection in these patients. In a multivariate analysis only
MELD score (p=0.001), the peak count of neutrophyl
leukocyte in blood (p=0.04) and the resolution of infection (p=0.03) showed an
independent value when predicting the occurrence of bacterial infection-induced
renal failure in cirrhotic patients with ascites. On the other hand, MELD score
(p=0.002) and the resolution of infection (p=0.04) were the only variables
which showed an independent prognostic value on survival in cirrhotic patients
with ascites and bacterial infection.
Conclusion
In conclusion, the results of the
study show that:
a)
all types of bacterial infections can precipitate
renal failure in cirrhotic patients with ascites,
b)
the development of renal failure is more frequent in
sub-diaphragmatic bacterial infections,
c)
a progressive form of renal failure can be
precipitated only by sub-diaphragmatic infections and
d)
the stable or the progressive form of renal failure
can be induced by all types of sub-diaphragmatic bacterial infections and not
only by PBS. The probability of developing bacterial infection-induced renal
failure is related to the MELD score, the severity of the infection, and to the
resolution of the infection.
A.
V. Longacre; G. Garcia-Tsao;
L. Fraenkel.
Background:
Endoscopic variceal ligation (EVL) and
non-selective beta-blockers (BB) are both effective for primary prophylaxis of
variceal hemorrhage, however the route of
administration and side effects of these treatments are distinct. Patient
preferences, an important part of medical decision-making, have not been
evaluated and are the objective of this study.
Methods:
Untreated patients with newly
diagnosed esophageal varices underwent a standardized
educational session. Patients with contraindications to either EVL or BB were
excluded. Patient preferences for treatment were evaluated using an interactive
computer task in which patients compared treatments using Adaptive Conjoint
Analysis (ACA), a validated method that effectively describes preferences when
competing options exist. Treatment characteristics were based on published
literature and included route of administration, risk of fatigue, sexual
dysfunction, dysphagia, shortness of breath and/or
hypotension, procedure-related bleeding and perforation. Demographics, type and
severity of liver disease and level of physician trust were recorded by written
survey. Subjects were contacted one month after ACA to determine the treatment
prescribed.
Results:
49 subjects were enrolled with a
median age of 56 years (range 24-74); 78% were male, 94% Caucasian, 45% had
some college education, 43% were Child B/C and the median MELD score was 11
(6-29). Subjects reported strong preferences for receiving complete medical
information and preferred a shared decision-making role with their physician.
Based on the ACA, 31 (63%) subjects preferred EVL. Risks of shortness of breath
and/or hypotension, fatigue, and procedure-related bleeding were most important
to subjects. 92% felt the ACA correctly predicted the importance of each
treatment characteristic. BB were preferred by older patients (p=0.007) and
those preferring medications over procedures (p=0.0003), whereas EVL was
preferred by those with a college education (p=0.03) or higher annual income
(p=0.007). In a linear regression model only age and preference for medication
over procedures remained significantly associated with treatment preference. 44
subjects were prescribed BB; 3 were lost to followup,
2 were not given prophylaxis and none received EVL.
Conclusion:
Our results suggest that although
patient preferences for prophylaxis of variceal bleed are variable, many
patients prefer EVL over BB. Given patients’ desire for full disclosure and an
active role in shared decision-making, both EVL and BB should be discussed with
cirrhotic patients requiring primary prophylaxis for variceal hemorrhage and patient preference should be considered when
starting primary prophylaxis.
T.
D. Boyer; J. Henderson; S. Arrigain; J. Connor.
Introduction:
Decompensation in the cirrhotic is
said to occur with development of variceal bleeding and/or ascites,
encephalopathy or icterus and is associated with a
worsened prognosis. In a recently completed trial comparing TIPS to DSRS in
Child’s A/B cirrhotics refractory to medical or
endoscopic therapy we found 2 years survival rates of 88% and 81%, respectively
(Gastroenterology 130:1643;2006). This excellent survival with follow-up of 46
months provided the opportunity to examine which features at randomization were
predictive of outcome.
Methods:
Seven categorical and 19 continuous
variables collected at time of randomization were used to compare the
relationship between the variable and time to death using the Log-rank test for
categorical variables and Cox proportional hazards model for continuous
variables. Interactions between individual variables were assessed by using a
Cox proportional hazards model. A multivariable model was obtained using a
bootstrap-bagging procedure on the Cox proportional hazards model.
Results:
By univariate
analysis: age-hazard ratio (HR-95%CI) per 5 years increase = 1.15 (1.004,1.31);
MELD score-HR per 5 unit increase = 1.74 (1.01,2.99); prior bleeds excluding
current-HR per 1 bleed = 1.19 (1.1,1.28); PT-HR per 0.1 increase = 1.11
(1.001,1.22); and creatinine-HR per 0.1 increase = 1.16(1.011,1.34) were all
significantly associated with increased risk of dying during the period of
observation. Gender and Child's score were also predictive of increased
mortality though these terms did not achieve significance at the alpha = 0.05
level. By multivariate analysis only number of prior bleeds and level of PT
were associated significantly with increased risk of dying-Table.
Summary:
Refractory variceal bleeding
treated effectively with TIPS or DSRS is associated with excellent survival.
Although a higher MELD score was predictive of outcome by univariate
analysis it was not predictive by multivariate analysis. Only number of prior
bleeds and PT were predictive of outcome in these well compensated patients.
Conclusion: Variceal bleeding in Child’s A/B cirrhotics
is not a sign of decompensation when treated with TIPS or DSRS, especially when
recurrent bleeding is prevented. In addition, MELD score is of limited
predictive value in patients with compensated disease who bleed once number of
prior bleeds and PT are considered.
Supported by grant from NIDDK #
DK050680 Authors have no conflicts to report.
Table 1
|
Variable |
Hazard Ratio (95%
CI) |
Cox proportional
Hazards P-value |
|
DSRS vs. TIPS |
1.28 (0.74, 2.23) |
0.38 |
|
Number prior bleeds if no prior bleed then 0
(HR per 1 bleed) |
1.21 (1.12, 1.31) |
<.0001 |
|
Pt (HR per 0.10 increase) |
1.13 (1.02, 1.24) |
0.017 |
V.
Mallet; D. Lasne; H. Fontaine; P. Blanchard; A. Vallet-Pichard; J. Serpaggi; S. Pol.
Background:
In the era of highly active
antiretroviral therapy (HAART), chronic liver disease is responsible for an
important and increasing number of deaths among human immunodeficiency virus
(HIV)-infected patients. In some patients the etiology of the liver damage
remains unknown. HIV-associated idiopathic liver disease has not been fully
described: mechanisms involved, its clinical importance and prognosis are for
the moment unknown and probably underestimated. Recent reports indicate that
HIV-infected patients are at increased risk for the development of thrombosis.
Among other possibilities, an acquired deficiency of protein S (PS), one of the
plasma’s natural anticoagulants might explain this tendency.
Objective:
To describe and explain
non-cirrhotic portal hypertension in HIV-infected patients.
Patients:
After our first observation, eight
consecutive patients (n=9) were referred to our unit from January 2003 to May
2006 for clinical and/or biological symptoms of chronic liver disease of
unexplained origin (two patients with positive HCV-RNA were excluded). All
patients underwent liver biopsy and were screened for thrombophilia
(including antithrombin, protein C and protein S deficiency,
factor V Leiden and II G20210A mutations, lupus anticoagulant, and antiphospholipid antibodies).
Results:
All patients had symptomatic portal
hypertension, with a history of esophageal bleeding
in six of them. At time of diagnosis, all were treated by HAART with an
efficient immune restoration. All were receiving or had received didanosine. Biopsy-proven nodular regenerative hyperplasia
(NRH) was observed in six patients, and suggested in one (sinusoidal dilatation
in a clinical context of portal hypertension without overt liver disease). Out
of the 7 patients, 7 were found to have one or more coagulation abnormalities
inducing an increased risk of thrombosis: PS deficiency in the absence of vitamine K deficiency, 7 patients ; factor V Leiden mutation,
1 patient, factor II G20210A, 1 patient In two patients, more than one prothrombotic state were identified.
Conclusions:
NRH appears to be a new cause of
non-cirrhotic portal hypertension in HIV-infected patients and is linked with a
thrombotic state, potentially HIV-induced and/or to
HAART toxicity.
F.
Kanwal; B. M. Spiegel; R. Bolus; R. D. Hays; S. J.
Kim; I. M. Gralnek.
Background:
Despite the realization that health
related quality of life (HRQOL) is an important outcome in patients with
advanced liver disease, clinicians rarely assess HRQOL of the liver disease
patients. This disconnect may reflect the perceived respondent burden related
to the length of available HRQOL instruments. LDQOL1.0 is a reliable and valid
instrument in patients with advanced liver disease. However, it has 75
disease-targeted items grouped in 12 scales. Our objective was to develop and
validate a Short Form (SF) of LDQOL 1.0.
Methods:
Using HRQOL data from the
validation study of LDQOL 1.0, we selected 36 items based on content coverage
and internal consistency reliability coefficients (cronbach
α) of original scales. We then administered the resulting questionnaire to
a cohort of 157 patients with advanced liver disease. We used both multi-trait
scaling and factor analysis to test our hypotheses regarding HRQOL domains. We
measured the reliability for the resulting scales using Cronbach
α, and assessed the construct validity by correlating the SF-LDQOL scores
with several anchors, including SF-36, symptom severity, disability days, and
overall health.
Results:
36 items were grouped into 9
disease-targeted scales (Table 1). When administered together with SF-36, the
mean (SD) completion time was 18 (+9) for SF vs. 38 (+20) min for LDQOL 1.0.
Multi-trait scaling showed that item-scale correlation was higher for the
hypothesized scales than for other scales. Factor analysis confirmed these
results. Table 1 displays the Cronbach-α and
supporting evidence of construct validity. The Cronbach-α
for 8 of 9 scales was >0.70. The direction of the correlation coefficients
(r) for all comparisons was consistent with a-priori hypotheses. However, the
magnitude of r was lower than hypothesized for the Hopelessness, Distress, and
Loneliness scales.
Conclusion:
We have developed and validated the
SF of LDQOL (36 items, 9 scales). 3 of the 9 scales will require further
modification. We believe that SF LDQOL will reduce respondent burden without
significantly compromising the reliability.
Table 1: Reliability and Construct Validation
of Short Form LDQOL
|
|
Items, N |
Cronbach-α |
SF-PCS |
SF-MCS |
Symptom Severity |
Disability Days |
Overall health |
|
Symptoms
of liver disease |
6 |
0.74 |
0.49 |
0.46 |
-0.49 |
-0.38 |
0.45 |
|
Effects
of liver disease |
3 |
0.72 |
0.53 |
0.57 |
-0.55 |
-0.41 |
0.51 |
|
Memory/Concentration
|
4 |
0.92 |
0.33 |
0.49 |
-0.48 |
-0.33 |
0.45 |
|
Sleep
|
5 |
0.71 |
0.44 |
0.47 |
-0.52 |
-0.41 |
0.43 |
|
Hopelessness
|
2 |
0.50 |
0.07 |
0.22 |
-0.01 |
-0.11 |
0.18 |
|
Distress
|
2 |
0.83 |
0.10 |
0.15 |
-0.01 |
0.06 |
0.08 |
|
Loneliness
|
5 |
0.70 |
0.14 |
0.31 |
-0.08 |
-0.01 |
0.24 |
|
Stigma
of liver disease |
4 |
0.80 |
0.30 |
0.43 |
-0.45 |
-0.27 |
0.43 |
|
Sexual
functioning/ problems |
4 |
0.83 |
0.39 |
0.41 |
-0.43 |
-0.19 |
0.51 |
M.
Austin; W. Bernal; J. Wendon.
Background:
Outcome of patients with advanced
chronic liver disease (CLD) admitted to intensive care are often poor. The
early and accurate assessment of likely outcome is important for patient
information, relatives expectation and appropriate use of resources. Organ
failure (OF) scoring systems in this regard are often complex with limited
clinical utility. In a large cohort of patients admitted to a liver ICU (LICU)
we examined ICU predictors of survival, on admission and developed and
validated a simple and practical bedside survival score, comparing accuracy
with other scoring systems.
Methods:
Consecutive patients with CLD
admitted to LICU between 01/1999-03/2005 formed a score derivation group, with
a further prospective validation set of admissions between 03/2005-03/2006.
Demographic features, indication for admission and severity of CLD and OF were
examined. Logistic regression analysis was used to determine variables
independently predictive of ICU survival. Comparison with discriminative power
of admission MELD, Child Pugh (CP), APACHEII and SOFA scores were performed
using receiver operating characteristic (AUROC) curve.
Results:
353 patients (61% Male) were
studied in the derivation group; median age 51yrs (interquartile
range 41-58), Child Pugh score 12 (11-15), MELD 25 (18-32) and SOFA 15 (11-18).
Mortality was 62%. Logistic regression analysis identified 5 admission factors
as being independent predictors of ICU outcome: patient age, bilirubin,
requirement for vasopressors or haemofiltration
and encephalopathy of grade 3 and above. Continuous variables were dichotomised
and points assigned in accordance to their prognostic weight to derive the
bedside score. Range of score was 0-23; AUROC was 0.878 (95%CI 0.838-0.917).
SOFA, AUROC 0.796 (0.747-0.842) and MELD 0.706 (0.649-0.763). The threshold
value to best predict non-survival for the bedside score was 12, MELD 23 and
SOFA 12. Bedside score test performance was maintained when applied to the
prospective validation sample (table).
Conclusion:
We describe a simple, accurate and
practical bedside score, which incorporates admission demographic features,
disease severity and organ failure enabling early prediction of short-term
outcome in patients with CLD admitted to an ICU.
|
Score |
Sensitivity |
Specificity |
Accuracy |
PPV |
NPV |
|
Derivation |
sample |
|
|||
|
Bedside>12 |
87 |
77 |
83 |
86 |
78 |
|
MELD>23 |
71 |
63 |
68 |
76 |
57 |
|
SOFA>12 |
84 |
61 |
75 |
78 |
70 |
|
Validation |
sample |
||||