HEPATOCELLULAR CARCINOMA SURVEILLANCE TRENDS IN INDIVIDUALS WITH CIRRHOSIS IN NORTH CAROLINA, 2010-2018

Cirrhosis is the 11th leading cause of death in the United States1 and the 2nd leading cause of nonmalignant gastrointestinal mortality in the United States2, with an age-adjusted mortality rate that has been on the rise over the past decade.3 Given the elevated risk of HCC in patients with cirrhosis, the American Association for the Study of Liver Disease and the National Comprehensive Cancer Network recommend that individuals with cirrhosis receive HCC surveillance using abdominal ultrasound, with or without alpha-fetoprotein (AFP) test, every six months.4,5 Guideline-concordant HCC surveillance is associated with earlier stage at HCC diagnosis, increased receipt of curative treatment, and improved survival.6,7 Despite these benefits, prior reports indicate that fewer than 25% of adults with cirrhosis receive HCC surveillance as recommended.8–14 Furthermore, the epidemiology of cirrhosis and HCC in the U.S. have shifted in recent years, with an increase in alcohol-related liver disease (ALD) and non-alcoholic liver disease (NAFLD)-related cirrhosis, along with improvements in hepatitis C virus (HCV) screening and treatment.15–17 Despite clear shifts in chronic liver disease epidemiology in the U.S., little is known about contemporary patterns of HCC surveillance. Additionally, prior studies have shown that HCC surveillance utilization is influenced by a variety of patient-, provider-, and area-level factors. At the patient-level, studies have suggested that individuals with NAFLD or ALD have lower rates of HCC surveillance than those with hepatitis B virus (HBV) and HCV cirrhosis etiologies.8,11,18–20 Insurance type has also been found to be associated with receiving HCC surveillance, where patients with Medicaid and Medicare co-insurance had higher surveillance rates than those with Medicaid alone.19 At the provider-level, several studies have reported that individuals receiving gastroenterology or hepatology specialty care are more likely to receive HCC surveillance.8,9,11,20–22 Therefore, the aims of this dissertation were (1) to describe patterns of HCC surveillance utilization by payer, cirrhosis etiology, and calendar year and (2) to examine patient-, provider-, and area-level factors associated with HCC surveillance among individuals with cirrhosis in North Carolina, using multi-payer claims data. We conducted a retrospective cohort study using claims data from Medicare, Medicaid, and a large private insurer in North Carolina. We included individuals 18 years or older with a first claims-based cirrhosis diagnosis between January 1, 2010, and June 30, 2018, determined using validated coding algorithms. Individuals were required to have continuous insurance enrollment for at least 12 months before the first claim with a cirrhosis diagnosis. We excluded individuals with liver transplant or HCC history and individuals with an HCC diagnosis in the six months after cirrhosis diagnosis. The outcome was HCC surveillance by abdominal ultrasound, computed tomography scan, or magnetic resonance imaging. For Aim 1, we estimated 1- and 2-year cumulative incidences for HCC surveillance and assessed longitudinal adherence to surveillance by computing the proportion of time covered (PTC). For Aim 2, we used a cause-specific proportional hazards model to assess associations between patient-, provider-, and area-level covariates and 2-year cumulative incidence of surveillance. In our final cohort of 46,052 eligible individuals (71% Medicare, 15% Medicaid, 14% private insurance), 54% were male and the mean age was 63 years (standard deviation, 14 years). The majority of individuals were White Non-Hispanic race/ethnicity (73%), followed by Black Non-Hispanic (23%), and Other or Hispanic (4%). Over half (59%) did not have decompensated cirrhosis at the index date. For Aim 1, from 2010 to 2016, the overall 2-year cumulative incidence of HCC surveillance was 60%, ranging from 59% in 2010 to 64% in 2016. The 2-year cumulative incidence of HCC surveillance increased the most for patients with Medicaid, from 57% in 2010 to 71% in 2016. For those with an initial screen in the first six months of their cirrhosis diagnosis, the median was 50% (Q1, 25%; Q3, 75%). The mean PTC for all surveillance modalities for the cohort was 29.9% (std, 34.5%), and the median was 16.3% (Q1, 0; Q3, 52.2%)For Aim 2, individuals with alcohol-related liver disease (ALD) (cause-specific hazards ratio (HR) 0.89, 95%CI 0.86-0.93) and non-alcoholic fatty liver disease (NAFLD) (HR 0.95, 95%CI 0.92-0.99) cirrhosis etiologies were less likely to receive HCC surveillance than those with hepatitis C virus (HCV). Individuals with decompensated cirrhosis at the time of diagnosis were more likely to receive HCC surveillance in the first two years after the index date than individuals who had compensated cirrhosis (HR 1.33, 95%CI 1.30-1.37), as were individuals insured through Medicare compared to those insured through Medicaid (HR 1.11, 95%CI 1.07-1.16). At the provider-level, individuals who only had a claim from a primary care provider were less likely to receive an HCC surveillance exam than those who had a claim from a gastroenterologist or hepatologist in the year before their cirrhosis diagnosis (HR 1.21, 95%CI 1.18-1.24). In conclusion, in our first aim we found that HCC surveillance initiation after cirrhosis diagnosis remains low, though it has improved slightly over time, particularly among those with Medicaid. Given the continued rise in cirrhosis prevalence, efforts should be made to increase uptake and improve longitudinal adherence to HCC surveillance. Our findings in our second aim suggest that individuals with viral hepatitis etiologies, Medicare insurance, decompensated cirrhosis, and care from a GI/hepatologist are more likely to receive an initial HCC surveillance exam in the first two years after the index date than their counterparts. Findings from this study may help guide future interventions in increasing HCC surveillance rates for individuals with cirrhosis.