Society: AASLD

Background and Aims: Screening tools are needed for early detection of NAFLD-fibrosis. We aimed to identify risk factors for NAFLD-fibrosis in the multiethnic population of the U.S. and thus accelerate development of equitable screening tools.
Methods: National Health and Nutrition Examination Survey (NHANES) 2017-March 2020 datasets were analyzed. The study group included 6,661 adults (age ≥ 20 years) with vibration controlled transient elastography (VCTE) data. NAFLD-fibrosis thresholds were set by optimizing the sensitivity and specificity of the VCTE-CAP score and the stiffness score for differentiating steatosis grade 0 from grades 1-3 and distinguishing fibrosis stages 0-1 from stages 2-4. Logistic regression was used to determine odds ratios (ORs). Independent variables included age, sex, body mass index (BMI), smoking history, alcohol use, poverty, diabetes, and hypertension. Diabetes was defined by self-report, and/or hemoglobin A1c (HbA1c) ≥ 6.5%, and/or fasting plasma glucose ≥ 126 mg/dL. Participants with past and current viral hepatitis and alcohol-associated liver disease were excluded.
Results: NHANES 2017-March 2020 had 372 individuals with NAFLD-fibrosis, defined as VCTE-CAP ≥ 285 dB/m and stiffness ≥ 8.6 kPa. Non-Hispanic Black persons had a lower prevalence of NAFLD-fibrosis than non-Hispanic Whites (Fig.1A). The age-standardized weighted prevalence of NAFLD-fibrosis was about 7-fold higher in those with diabetes in the total population and in non-Hispanic White, Mexican American and Other race, but not in non-Hispanic Black (Fig.1C). The prevalence of NAFLD-fibrosis did not differ significantly between non-Hispanic Black persons with and without diabetes. After adjusting for age, sex, BMI, alcohol use, smoking status, and hypertension in multivariate logistic regression models, diabetes was not significantly associated with NAFLD-fibrosis in non-Hispanic Black persons (OR=1.26, 95%CI: 0.57-2.77), but it was associated in non-Hispanic White (OR=4.46, 95%CI: 2.87-6.91) and Mexican American persons (OR=5.61, 95%CI: 2.61-12.04) (Table 1). Similar results were obtained when NAFLD-fibrosis was defined by liver stiffness ≥8.6 kPa and CAP ≥ 263 dB/m, which had 90% sensitivity for detecting steatosis grade 0 vs.1-3. Of interest, non-Hispanic Black persons had a higher prevalence of diabetes than whites (Fig.1B). Thus, the disconnection between diabetes and NAFLD-fibrosis in non-Hispanic Black persons does not stem from a low prevalence of diabetes.
Conclusions: NHANES provides nationally representative data about the residential population of the U.S. NAFLD-fibrosis was not associated with diabetes in non-Hispanic Black persons, indicating that screening algorithms that rely on diabetes to identify people at high risk for NAFLD-fibrosis may inadvertently disadvantage non-Hispanic Black individuals (Supported in part by Pfizer).

Introduction: Overt hepatic encephalopathy (OHE) is a serious clinical complication of cirrhosis. Risk of recurrence is estimated to be as high as 40% within one year of a first OHE episode, resulting in costly and burdensome readmissions. Among patients with an initial OHE hospitalization, little is known about the impact of rifaximin use post-discharge on OHE-related rehospitalization and associated costs in real-world practice.

Methods: Claims data from MarketScan® Commercial Subset (10/01/2015–03/31/2020) were used to identify adults (18-64 years) with an OHE hospitalization (index hospitalization). Based on treatment following index hospitalization, patients were classified into mutually exclusive cohorts: rifaximin treated (with/without lactulose) and not rifaximin treated. The two cohorts were further stratified into four subgroups representing decreasing quality of care (QOC): Type 1 received rifaximin without any gap in treatment following index hospitalization; Type 2 received rifaximin within 30 days post-discharge; Type 3 received lactulose within 30 days post-discharge; Type 4 received no treatment. The length of stay (LOS) for index hospitalization, 30-day OHE-related rehospitalizations, and all-cause healthcare costs were reported. The association of rifaximin use with 30-day OHE-related rehospitalizations and costs were assessed.

Results: Baseline demographics and clinical characteristics were well balanced between the rifaximin treated (N=1,452; Type 1: N=1,138, Type 2: N=314) and not rifaximin treated (N=560; Type 3: N=337, Type 4: N=223) cohorts. Mean age was 54.2 and 55.2 years and 39.3% and 42.7% were female, respectively. The index OHE hospitalization of the rifaximin treated cohort had an average LOS of 10.8 and an average cost of $48,225 compared to 8.0 days and $29,108 for the not rifaximin treated cohort. The 30-day risk of OHE-related rehospitalization following the index hospitalization was lower for the rifaximin treated cohort (6.8%) vs not rifaximin treated (10.5%; adjusted odds ratio (OR) 0.56, p<0.01; Figure 1). Following the index OHE hospitalization, decreasing QOC (from Type 1 to Type 4) resulted in higher 30-day risk of OHE-related rehospitalization; compared to Type 1, the ORs were 1.31 for Type 2 (p=0.28), 1.74 for Type 3 (p=0.01), and 1.90 for Type 4 (p=0.01; Figure 2). Among the rifaximin treated cohort, increased pharmacy costs were offset by an equivalent reduction in medical costs, which led to no significant annual cost differences between the two cohorts.

Conclusions: Patients receiving rifaximin at discharge of their OHE hospitalization were less likely to experience 30-day OHE-related rehospitalizations compared to patients who did not receive rifaximin. The reduction in medical costs offset the increased pharmacy costs among the rifaximin treated cohort, demonstrating that rifaximin use is cost neutral.

<i><b>Figure 1. 30-day risk OHE-related rehospitalization (%)</b></i><br /> <i>*Significant at the 5% level</i><br /> <b>Note: </b><br /> [1] An odds ratio less than 1 indicates that the odds of experiencing an OHE-related rehospitalization in rifaximin treated patients are lower than those of not rifaximin treated patients.

Figure 1. 30-day risk OHE-related rehospitalization (%)
*Significant at the 5% level
Note:
[1] An odds ratio less than 1 indicates that the odds of experiencing an OHE-related rehospitalization in rifaximin treated patients are lower than those of not rifaximin treated patients.

<i><b>Figure 2. Adjusted 30-day risk of OHE-related rehospitalization (%)</b></i><br /> <i>*Significant at the 5% level</i><br /> <b>Notes: </b><br /> [1] An odds ratio greater than 1 indicates that the odds of experiencing an OHE-related rehospitalization in patients with Type 2, 3, or 4 care are greater than those of patients with Type 1 care.<br /> [2] The four subgroups represent decreasing quality of care and are defined as follows: Type 1 received rifaximin without any gap in treatment following hospitalization; Type 2 received rifaximin within 30 days post-discharge; Type 3 received lactulose within 30 days post-discharge; Type 4 received no treatment.

Figure 2. Adjusted 30-day risk of OHE-related rehospitalization (%)
*Significant at the 5% level
Notes:
[1] An odds ratio greater than 1 indicates that the odds of experiencing an OHE-related rehospitalization in patients with Type 2, 3, or 4 care are greater than those of patients with Type 1 care.
[2] The four subgroups represent decreasing quality of care and are defined as follows: Type 1 received rifaximin without any gap in treatment following hospitalization; Type 2 received rifaximin within 30 days post-discharge; Type 3 received lactulose within 30 days post-discharge; Type 4 received no treatment.

Background: Social determinants of health (SDOH) impact cancer risk, access to screening, treatment (Tx) patterns, and health outcomes. We aimed to examine the impact of demographics, clinical factors, and SDOH on Tx patterns and healthcare costs among newly diagnosed patients (pts) with hepatocellular carcinoma (HCC).

Methods: We identified patients diagnosed with HCC from 1/1/2017-12/31/2020 using 100% Medicare Fee-For-Service claims and a commercial multi-payor claims database. Using home ZIP code from enrollment records, pts were matched to near-neighborhood SDOH characteristics. Cost and Tx outcomes were assessed post-diagnosis among pts with ≥1 month follow-up. Tx was classified as curative if there was any surgical Tx or ablation alone. Locoregional or systemic Tx were considered non-curative. Surveillance receipt was defined as ultrasound (US) and/or alpha-fetoprotein (AFP), or contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI), during the 12 months pre-diagnosis. Multinomial logistic regression was used to assess the impact of these factors on Tx (curative vs. non-curative). Multivariable generalized linear regression was used to identify factors associated with total healthcare costs.

Results: The 34,068 pts with HCC were mostly male (67%), White (72%), with mean age 68 years. Only half (51%) of pts with at least 30 days of follow-up received any Tx and 70% of pts received any surveillance. Curative treatment receipt was higher among pts with surveillance (24% with CT/MRI surveillance, 18% with US/AFP surveillance vs 9% with no surveillance; Fig 1A). Multivariable analyses showed pts who received surveillance were significantly more likely to receive curative Tx, while Black pts, pts with lower education level and pts diagnosed in 2020 (COVID-19 year) were significantly less likely to receive curative Tx (Fig. 1B). Pts with CT/MRI and those with US/AFP surveillance had 25% (p<0.01) and 3% lower (p=0.07) adjusted healthcare costs post-diagnosis than those without surveillance, respectively. Pts with CT/MRI and those with US/AFP surveillance had 16% lower and 5% higher unadjusted healthcare costs post-diagnosis than those without surveillance, respectively (Fig 2A). Other factors associated with higher costs were Black race, low English proficiency, living alone, and diagnosis in 2020 (Fig. 2B).

Conclusion: Pts with surveillance were significantly more likely to receive Tx and had lower post-diagnosis healthcare costs. Increasing access to HCC surveillance could improve Tx outcomes, reduce health disparities, and reduce total cost of care through early cancer detection.

Background
Biannual ultrasonography and alpha-fetoprotein (AFP) test are recommended for hepatocellular carcinoma (HCC) surveillance. However, ultrasonography has a lower sensitivity for detecting early-stage HCC than computed tomography and magnetic resonance imaging (MRI), particularly in cirrhotic patients. Recently, non-contrast abbreviated MRI (aMRI) demonstrated comparable performance to MRI without the risk of contrast media exposure and at a lower cost. We aimed to investigate the cost-effectiveness of non-contrast aMRI for HCC surveillance in cirrhotic patients, using ultrasonography with AFP as reference.
Methods
Cost-utility analysis was performed using Markov model with deterministic analysis and patient-level microsimulation in Thailand and the United States (US) as representatives of countries with low and high medical costs, respectively. The model represented the course of disease from cirrhosis to HCC development and post-HCC treatments (Figure). HCC states were classified as small HCC (tumor size <2 cm) and large HCC (tumor size ≥2 cm). Cirrhosis states were categorized into Albumin-Bilirubin (ALBI) grade 1 (early-stage cirrhosis), ALBI grade 2-3 (late-stage cirrhosis), and end-stage cirrhosis. A simulated cohort of 100,000 cirrhotic patients was entered into the model. During a 6-month cycle, the cohort underwent HCC surveillance with ultrasonography plus AFP followed by a confirmatory full diagnostic MRI, or aMRI with confirmatory full diagnostic MRI. Surveillance protocols were considered cost-effective based on a willingness-to-pay of $4,665 in Thailand and $50,000 in the US. One-way sensitivity analyses and probabilistic sensitivity analyses were performed to evaluate the robustness of the model.
Results
aMRI was cost-effective in both countries with incremental cost-effectiveness ratios (ICERs) of $4,418/quality-adjusted life year (QALY) in Thailand and $39,513/QALY in the US. A patient-level microsimulation showed consistent findings that aMRI was cost-effective in both countries (Table). In probabilistic sensitivity analysis, aMRI was more cost-effective than combined ultrasonography and AFP with a probability of 0.66 in Thailand and 1.00 in the US. By sensitivity analyses, the cost-effectiveness of aMRI increased in settings with a higher HCC incidence. In the US, adjusting the performance of surveillance modalities did not affect the cost-effectiveness of aMRI. However, in Thailand, aMRI was not cost-effective when the sensitivity of aMRI was lower than 0.71 or the sensitivity of ultrasonography with AFP was greater than 0.57 for small HCC detection.
Conclusions
Non-contrast abbreviated MRI is more cost-effective than ultrasonography with AFP for HCC surveillance and may be considered as the first option for HCC surveillance in cirrhotic patients.

Simplified Markov model

Deterministic base-case cost-utility analysis and patient-level microsimulations

Deterministic base-case cost-utility analysis and patient-level microsimulations

BACKGROUND
Cirrhosis of the liver is a common condition. Yet, there are limited longitudinal data on the cost of treating patients with cirrhosis. Previous studies were limited by their focus on hospitalized patients or those at the end-of-life, leaving a gap in understanding the long-term costs associated with managing patients with cirrhosis across the full spectrum of their illness.

METHODS
We conducted a retrospective cohort study from 130 facilities in the Veterans Administration among patients with cirrhosis identified based on validated algorithms from 10/1/2011 and 09/30/2015 with follow up until 09/30/2019. We identified a sex- and age-matched control cohort without cirrhosis, selecting up to 4 controls for each patient. We used two-part multivariable models to estimate incremental annual total health care costs attributable to cirrhosis for 4 years overall and in subgroups based on severity (compensated, decompensated), type of cirrhosis complication (ascites, encephalopathy, varices, hepatocellular cancer [HCC]), and comorbidity burden.

RESULTS
We compared 79,664 patients with cirrhosis and 274,745 matched controls. Among patients with cirrhosis (mean [SD] age, 62.8 [11.9] years), 24.9% of patients died within 1 year and 43.9% died within 4 years of the diagnosis. Overall, the incremental total adjusted costs for caring of patients with cirrhosis vs. controls were $27K (95% CI, $26k-$29k) during the first year and ranged from $11k (95% CI, $9k-$12K) to $13K (95% CI, $12K-$14K) in the subsequent 3 years. Most of the cost was related to cirrhosis complications, especially during the first year, with incremental cost of decompensation of $36K (95% CI, $ 33K-$39K). The year 1 incremental cost of managing patients with encephalopathy ($38K; 95% CI, $33K-$43K) was similar to that of managing patients with ascites ($36K, 95% CI, $32K-$40K), both higher than the cost of managing patients with varices ($17K; 95% CI, $12K-$37K). In years 2 to 4, incremental cost of caring for patients with HCC (years 2-4 costs, $41K) exceeded those of managing other complications (years 2-4 costs for ascites and encephalopathy were $28K and $30K, respectively). The incremental cost associated with 3 or more comorbidities in year 1 was $26,107 (95% CI, $19,585-32,629) but only 6% of patients had multi-morbidity.

CONCLUSIONS
In this large study of ambulatory patients, patients with cirrhosis had substantially higher healthcare costs than matched controls. The additional economic burden of cirrhosis was mostly related to cirrhosis complications. Multimorbidity explained a small proportion of incremental cost, suggesting that value-based programs in cirrhosis may not need complex case-mix adjustments. Preventing progression to decompensation and HCC has the largest potential for cost saving and could serve as targets for improvement.

Ruben Hernaez

Baylor College of Medicine

Appearances