The impact of socioeconomic status on outcomes in hepatocellular carcinoma: Inferences from primary insurance

Abstract Background To investigate the impact of insurance status on outcomes in patients with hepatocellular carcinoma (HCC). Methods Patients diagnosed with HCC in the cancer registry from 2005 to 2016 were retrospectively stratified by insurance group. Overall survival was assessed via Kaplan‐Meier curves and Cox proportional hazard models including potential confounders in multivariable analyses. Results Seven hundred and sixty‐nine patients met inclusion criteria (median age 63 years, 78.8% male, 65.9% Caucasian). 44.5% had private insurance (n = 342), 29.1% had Medicare (n = 224), and 26.4% had Medicaid (n = 203). At diagnosis, Medicaid patients had higher rates of Child‐Pugh B (32.0%) and C disease (23.6%) vs Medicare (28.6% and 9.8%) and private insurance (26.9% and 6.7%, P < 0.0001) and higher MELD scores (median 11.0) vs Medicare (9.0) and private insurance (9.0, P = 0.0266). Across insurance groups, patients had similar distribution of American Joint Committee on Cancer stage, tumor size, and multifocal tumor burden. Patients with private insurance had the highest survival (median OS 21.9 months) vs Medicare (17.7 months) and Medicaid (13.0 months, overall P = 0.0061). On univariate analysis, Medicaid patients demonstrated decreased survival vs private insurance (HR 1.40, 95% CI: 1.146‐1.715, P = 0.0011). After adjustment for liver disease factors, this survival difference lost statistical significance (Medicaid vs private insurance, HR 1.02, 95% CI: 0.819‐1.266, P = 0.8596). Conclusion Medicaid was associated with advanced liver disease at HCC diagnosis; however, insurance status is not an independent predictor of HCC survival.


| INTRODUCTION
Hepatocellular carcinoma (HCC) accounts for 70%-85% of primary liver cancers; liver cancer is the fifth most common cancer worldwide and the third leading cause of cancer death. [1][2][3] Approximately 60% of HCC cases are attributable to infection with hepatitis B (HBV) or hepatitis C (HCV) virus. [4][5][6] Two other main causes of HCC are cirrhosis secondary to chronic alcohol consumption and nonalcoholic steatohepatitis (NASH). In the United States, rates of NASH-driven HCC are increasing. 7 The predominant curative therapies for HCC are liver resection or liver transplantation, although recent studies have reported that radiofrequency ablation may be as effective as resection in treating small solitary HCC lesions. [8][9][10] Locoregional therapies (LRT), such as thermal ablation and transarterial chemoembolization, may be used as bridging therapy to transplant, to downstage disease, or as a palliative option. Treatment allocation is influenced by a variety of factors, including extent of cancer and the severity of liver disease. Patients with decompensated liver cirrhosis, poor hepatic synthetic function, and/or other serious comorbidities are less likely to receive treatment with curative intent. However, while patients with advanced cirrhosis are typically excluded from resection, these same patients are often prioritized for transplant.
Race, ethnicity, and socioeconomic status (SES) have been found to have significant effects on HCC incidence, overall survival, and treatment allocation. [11][12][13][14][15][16][17] In the Swiss Hepatitis C Cohort study, low SES was associated with the development of HCC. 18 Insurance status often correlates with SES. The purpose of our study was to examine the impact of primary insurance payer on outcomes in an inner-city tertiary care hospital population, with the hypothesis that insurance would reflect SES and that patients with Medicaid coverage would have poorer survival compared to Medicare or private insurance.

| Study population
Institutional review board approval was obtained prior to the study. We identified adult patients from the Smilow Cancer Center cancer registry at Yale New Haven Hospital who were diagnosed with HCC via either radiologic or histopathologic criteria between 2005 and 2016 and followed through 2017. Patients with unknown treatment status or who received the majority of their treatment at another hospital were excluded.
Patients with no or unknown insurance and patients with VA, Tricare, or Indian Health Service insurance were also excluded, as these groups were too small for statistical analysis. Patients who received liver transplantation and patients with combined HCC and cholangiocarcinoma were not included in this study.

| Data collection
Data available from the registry included age, gender, ethnicity, primary insurance, American Joint Committee on Cancer (AJCC) staging, and treatment status. Further data, including baseline laboratory values, tumor imaging, and detailed treatment course, was acquired through electronic medical record review. Child-Pugh score, Model of End-stage Liver Disease (MELD) score, Barcelona Clinic Liver Cancer (BCLC) stage, and Charlson Comorbidity index (CCI) were calculated using baseline laboratory values, patient characteristics, and imaging reports.

| Statistical analysis
Treatment was stratified into resection, ablation (percutaneous or laparoscopic), transcatheter LRT, combined ablation and transcatheter LRT (combo LRT), systemic chemotherapy, and palliative care. Therapy status and the temporal sequence of treatments received were explicitly identified via chart review. Patients receiving resection with or without additional therapies were categorized as "resection." Patients receiving a type of LRT as well as chemotherapy or radiation therapy were classified under the type of LRT received. "Curative intent" treatment was defined as patients receiving either resection or ablation of a single lesion with a largest diameter of less than 3 cm.
The database explicitly identified primary insurance payer at the time of diagnosis, which was subsequently divided into three groups: private insurance, Medicare, and Medicaid. Patients with Medicare with supplemental insurance were classified as private insurance.
Categorical variables were compared using the χ 2 test and continuous variables using the Kruskal-Wallis test for nonnormally distributed data to identify key differences between insurance groups. The Kaplan-Meier method was used to estimate median overall survival (OS). Prognostic factors for overall survival rates were compared via univariate (UVA) Cox proportional hazard models, including age, gender, ethnicity, Child-Pugh class, MELD score, liver cancer etiology, BCLC stage, best AJCC stage, largest tumor diameter, tumor cirrhosis, hepatitis C, hepatocellular carcinoma, insurance, socioeconomic status location, multifocal tumor burden, and treatment intent. Factors that were significant on UVA were included in the multivariable (MVA) proportional hazards analysis to account for confounding. An alpha level of 0.05 was chosen to indicate statistical significance. All P-values provided are two-sided. Calculations were performed using JMP Pro v.13.0.0 (SAS Institute Inc, Cary, NC) and R v.3.4.3 (R Core Development Team, Vienna, Austria).

| Disease presentation
In our study, patients with Medicaid presented with more advanced liver disease as represented by Child-Pugh and MELD score than patients with either Medicare or private insurance. However, no significant differences between insurance groups existed for cancer stage, as measured by AJCC stage, tumor size, tumor location, and the presence of multifocal HCC. While BCLC stage differed across insurance groups, this is likely due to BCLC including liver disease and performance status in the staging algorithm, while AJCC staging does not. [19][20][21] It has been noted that HCC patients with Medicaid or no insurance are more likely to present with late-stage cancer vs patients with private insurance. 22,23 Lack of long-term insurance has also been associated with higher prevalence of metastases in both the University Health Consortium [OR 1.9, 95% CI: 1.6-2.2] and Nationwide Inpatient Small [OR 1.6, 95% CI: 1.4-1.9] databases. 24

| Treatment allocation
Curative intent treatment was defined in this study as partial hepatic resection or ablation of a single lesion <3 cm in diameter and occurred less frequently in Medicaid patients than in those with Medicare or private insurance. Furthermore, the These differences in treatment allocation disappeared after substratifying insurance groups by Child-Pugh score. Hence, it can be inferred that treatment differences were due to increased prevalence of severe liver disease in the Medicaid group. This additionally demonstrates that treatment decisions at this institution are independent from insurance status. Primary insurance can serve as a surrogate parameter for SES. Various studies have reported that patients with Medicaid, underinsurance, no insurance, or lower SES are less likely to receive surgical treatment 25,26 or even treatment in general. 23,27,28 After controlling for tumor stage, resection status, and transplant eligibility, Sarpel found that patients with government insurance (Medicaid or Medicare without supplement) were less likely to undergo transplantation for HCC. 17 Similarly, in a study of safety-net hospitals, vulnerable patients (including those with Medicaid and poor SES) had lower rates of curative surgery and poorer short-term outcomes. 29 However, safety-net patients who could endure liver surgery had a similar prognosis as compared with patients at nonsafety net hospitals. This further suggests that survival differences between socioeconomic and insurance groups may be driven more by liver and tumor factors than by treatment decisions.

| Overall survival and insurance status
Kaplan-Meier curves and univariate Cox models demonstrated decreased survival in the Medicaid group as compared to the Medicare and private insurance groups in our cohort, which agreed with our original hypothesis. After analyzing the UNOS (United Network for Organ Sharing) database from 2002 to 2013, Magnetta reported that private payer insurance led to improved overall survival after OLT as compared to public insurance (Medicare/Medicaid) (MVA HR 0.91, 95% CI: 0.88-0.93, P < 0.001). 30 Hoehn similarly noted improved survival with private vs nonprivate insurance on a study of National Cancer Database patients with curable HCC (stage I/II) from 1998 to 2011. 14 Poorer survival has also been associated with lower SES. 11,12,16 In a study examining HCC survival over three decades, Wang et al found inferior survival in the high-poverty group. 31 In addition, Major has reported higher chronic liver disease mortality in areas of socioeconomic deprivation (HR 1.78, 95% CI: 1.34-2.36). 32 Although UVA showed decreased survival in Medicaid patients, primary insurance was not an independent predictor of survival after adjusting for Child-Pugh and MELD scores. This is similar to what was described by Yu et al: while their results also showed that insurance was not an independent predictor of mortality in HCC, it was associated with more advanced liver disease at diagnosis and lower rates of transplantation. 33 The lack of survival differences in our cohort after adjusting for liver disease factors demonstrates that patients with similar liver and performance status received similar treatments at this institution, regardless of insurance or SES. This suggests that while patients with lower SES or less desirable insurance may present with more advanced liver disease or cancer, if patients receive the appropriate treatment for their HCC stage and underlying liver disease, insurance and perhaps even SES may not affect survival.

| Limitations
While this study relied on archived records, we limited the study to patients with complete insurance and treatment data. The degree of missing data in other patient characteristic variables was minimal and was distributed throughout insurance groups. Although our sample size may have led to underpowering of analyses in subgroups, this cohort is among the largest from a single institution to analyze the impact of insurance on survival in HCC. Finally, this study was conducted using a Western (US) population, and therefore results may not be globally generalizable.

| CONCLUSION
Insurance status is not an independent predictor of HCC survival. Medicaid was associated with advanced liver disease at HCC diagnosis. After adjusting for Child-Pugh score, rates of treatment with curative intent were similar between insurance groups.