Worldwide, approximately 257 million persons are infected with hepatitis B virus (HBV) (1). In the United States, an estimated 0.84–1.59 million adults have chronic hepatitis B (CHB) infection (2,3), which is well known to cause cirrhosis and hepatocellular carcinoma (HCC) (4–6). As such, CHB is a substantial public health burden of morbidity, mortality, and economic costs, with patients with CHB infection having substantially higher healthcare resource utilization and cost compared with the population without CHB infection (7).
Although there is currently no cure for CHB, treatment can prevent 15%–25% of premature deaths from cirrhosis or HCC, but CHB spontaneously fluctuates between different phases in its natural history and not all phases require treatment (8). As a result, close monitoring is essential even when treatment is not needed because delayed initiation of antiviral therapy (if indicated) can leave patients at higher risk of disease progression and HCC development (9–11).
The American Association for the Study of Liver Diseases (AASLD) has released guidelines since 2001 recommending monitoring for disease progression to allow for timely initiation of HBV treatment in appropriate candidates (12–17). These guidelines also recommend HCC surveillance for patients with CHB infection because previous studies have reported a considerable survival advantage in patients who are adherent to HCC surveillance (18,19).
Despite the significant burden of disease and existence of management guidelines, previous studies have also suggested poor adherence to recommended care (20–22), although previous studies were limited by small sample size, single-center/single integrated care system, and/or studies focusing primarily in 1 population such as the population seeking care at the Veterans Affairs Health Care System (20–22). Therefore, the goals of this study were to examine the percentage of patients who adhered to AASLD guidelines for initial evaluation and long-term monitoring of patients with CHB infection in a nationally representative sample of patients with CHB infection with commercial insurance and Medicare with private insurance supplement and to identify potential predictors of adherence to the guidelines.
Study design and population
We performed a retrospective study of adult patients with CHB infection using the Truven Health MarketScan Research Database (Truven) housed in the Population Health Science Center at Stanford University, Palo Alto, California. The Truven database is a large national administrative claims database with data on approximately 139 million Americans with private health insurance and Medicare coverage between January 2007 and December 2014. We identified patients who had at least 1 inpatient or 2 outpatient International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes for HBV (070.22, 070.23, 070.32, or 070.33). We excluded patients younger than 18 years and those with less than 6 months of follow-up from the overall cohort. For analysis of long-term monitoring, we also excluded patients with less than 12 months of follow-up and those with HCC development or liver transplantation during the initial 12 months of follow-up. The index date was defined as the first CHB diagnosis date. All medical diagnoses and tests, procedures, or clinical visits were determined by ICD-9-CM and Current Procedural Terminology codes, respectively. This study was conducted in accordance to the tenets of the 1975 Declaration of Helsinki (23), and the study protocol was approved by the institutional review board at Stanford University, Stanford, California.
For initial evaluation, we reported the percentages of patients with testing performed as recommended by the AASLD. Over the years, the AASLD has made changes on treatment recommendation but the recommendations for baseline evaluation have remained relatively unchanged as follows: (i) tests to assess the state of liver disease including complete blood count (CBC), alanine aminotransferase (ALT), and prothrombin time (PT)/international normalized ratio (INR); (ii) hepatitis B e antigen (HBeAg)/hepatitis B e antibody (anti-HBe) and HBV DNA as a measure of HBV replication; (iii) tests to rule out coinfection: hepatitis C virus antibodies (anti-HCV), hepatitis D virus antibodies (anti-HDV), and human immunodeficiency virus antibodies (anti-HIV); (iv) assessment for hepatitis A immunity to identify those who need vaccination; (v) and ultrasound with or without α-fetoprotein (AFP) for initial HCC screening (12–15). For patients with cirrhosis (as defined by ICD-9 diagnosis codes as previously described) (see Supplementary Table 1, Supplementary Digital Content 1, http://links.lww.com/AJG/B992) (7), esophagogastroduodenoscopy (EGD) to screen for varices and creatinine testing are also recommended (24,25).
Regarding ALT, if a patient had a single test for ALT, a liver panel, or a comprehensive metabolic panel performed, it was counted in the ALT testing analysis. Similarly, the analysis for creatinine also includes renal panels and comprehensive metabolic panels.
We reported tests performed during the first year after the CHB diagnosis index date for initial evaluation assessment to allow for a grace period. For the purposes of this study, an optimal initial evaluation was defined as having ALT, CBC, HBV DNA, HBeAg, anti-HCV, anti-HAV, liver imaging, and AFP testing during the first 6–12 months of follow-up.
During this study period, the AASLD 2004 guidelines for long-term monitoring recommended (i) ALT every 3–6 months for HBeAg-positive patients with HBV DNA >105 copies/mL and normal ALT, (ii) ALT every 1–3 months if HBeAg-positive and ALT >1–2 × upper limit of normal (ULN), and (iii) ALT every 6–12 months for patients in the inactive HBsAg carrier state. In 2007, recommendations for the HBeAg-positive group with elevated ALT was stratified by ALT levels, with suggested ALT testing every 3 months if HBeAg-positive and ALT 1–2 × ULN and ALT every 1–3 months if ALT >2 × ULN. Similarly, recommendations for the inactive HBsAg carrier state group was also broken into 3 categories: (i) ALT every 3 months 3 times, then every 6–12 months if ALT <1 × ULN and HBV DNA <2,000 IU/mL, (ii) ALT and HBV DNA every 3 months if ALT 1–2 × ULN and HBV DNA 2,000–20,000 IU/mL, and (iii) treatment if ALT ≥2 × ULN and HBV DNA ≥20,000 IU/mL. For the purposes of this study, we used the least strict frequencies: every 6 months (with a 1-month grace period) and every 12 months (with a 3-month grace period) because the recommended frequency of ALT and HBV DNA testing depends on HBeAg status, which may not be performed for many patients.
For patients with cirrhosis, particularly, we also examined the frequency of additional tests: CBC, PT/INR, creatinine, and EGD. To assess the state of liver disease, we reported the frequency of patients having their CBC and PT/INR checked. Because patients with cirrhosis frequently show an impairment of renal function, we also reported the frequency at which patients with cirrhosis had their serum creatinine level checked (26). Regarding EGD, the AASLD recommends that among patients with cirrhosis and no varices, those who are compensated are recommended to undergo EGD screening every 3 years, whereas those who are decompensated are recommended to undergo annual EGD screening (24). For the purposes of this study, we used the least strict frequency, every 3 years, with a 1-year grace period.
To examine whether low testing rates were due to insufficient clinic visits, we also reported the frequencies of clinic visits. “Optimal” long-term monitoring was defined loosely as a minimum of ALT, HBV DNA, and HCC surveillance every 12 months with a 3-month grace period because we expected the adherence to strict guidelines would be very low and there may be a delay between a physician generating orders and the patient performing the tests. However, for selected outcomes, we also performed sensitivity analysis assessing testing frequency every 6 months. Additional details and rationales for the frequency of HCC surveillance are described further.
Regarding HCC surveillance, the AASLD recommended imaging every 6–12 months in their 2005 guidelines and revised their recommendation to every 6 months in 2011. Therefore, we used the least strict recommendation and defined adherence as every 6–12 months during this period, with a 3-month grace period, although we also assessed the rate of imaging at 6-month intervals as a sensitivity analysis (27,28). We evaluated the frequency of initial surveillance during the first 2 years of follow-up; in those who had at least 1 imaging test annually for 2 years, we also evaluated the persistence of surveillance afterward.
Although AASLD guidelines recommend the use of ultrasound for HCC surveillance, computed tomography and MRI are also occasionally used in clinical practice in selected patients, such as when ultrasound imaging is suboptimal, as is frequently the case with cirrhotic livers (27). As such, surveillance was defined as undergoing any liver imaging test (ultrasound, computed tomography, or MRI).
AASLD guidelines recommend HCC surveillance for Asian male HBV carriers older than 40 years, Asian female HBV carriers older than 50 years, HBV carriers with a family history of HCC, African/North American Blacks with CHB infection, and cirrhotic HBV carriers. However, because data on ethnicity and family history were unavailable, we defined “high risk” in our analysis as patients with cirrhosis, noncirrhotic male subjects older than 40 years, and noncirrhotic female subjects older than 50 years and compared rates of HCC surveillance between the high-risk group and those not considered to be at high risk.
Among patients with pharmacy claims data available, we reported the rate of ever receiving treatment.
Descriptive statistics were reported as proportions (%) for categorical variables and mean ± SD or median (interquartile range) for continuous variables. Comparative analysis between groups was performed using the χ2 test for categorical variables. For continuous variables, the Student t-test was used to evaluate normally distributed continuous variables, and the Wilcoxon rank-sum test was used to evaluate continuous variables that were not normally distributed. In addition to analysis of the overall cohort, we performed subgroup analysis by provider specialty (gastroenterology/infectious disease [GI/ID] vs non-GI/ID), liver disease severity (no cirrhosis, compensated cirrhosis, and decompensated cirrhosis), sex, and age.
Stepwise multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs), relating potential predictors to the outcome of optimal initial evaluation and long-term monitoring including HCC surveillance. We adjusted for the following potential confounders: age, sex, insurance plan types (PPO, preferred provider organization; HMO, health maintenance organization), out-of-pocket (OOP) expenditures, provider specialty, number of clinic visits per year, liver disease severity (no cirrhosis, compensated cirrhosis, and decompensated cirrhosis), and patient overall health status using the Deyo-Charlson Comorbidity Index (DCCI) (29). We chose to consider and adjust for DCCI scores as a general marker of how many comorbidities the patients may have to adjust for potential surveillance bias, in which sicker patients may seek care more often for a particular disease and end up getting surveyed for other diseases more often as a result (30).
Among those who were adherent during the first 2 years of follow-up, the persistence of HCC surveillance was estimated by the Kaplan-Meier method, and Kaplan-Meier curves were compared by the log-rank test. Time zero was 2 years after the CHB index date, and persistency duration was until adherence loss occurred or patients were censored because of lost to follow-up, liver transplantation, or incident HCC, whichever occurred first. Statistical significance was defined as a 2-tailed P value < 0.05, and statistical analysis was performed using R (3.5.0) and Stata 14 (Stata, College Station, TX) software.
After excluding patients younger than 18 years, those with less than 6 months of follow-up, and those without at least 1 inpatient or 2 outpatient CHB claims, we identified 55,317 eligible patients with CHB infection for analysis of initial CHB evaluation. Patient demographic and clinical characteristics are presented in Table 1. The overall mean age at baseline was 46 ± 12 years, with a little over half being men (57.6%). The Southern and the Western parts of the United States each accounted for a third of the patients (29.6% and 33.6%, respectively), a quarter were from the Northeast (23.6%), and 11.5% were from the North Central region. The most common insurance plan types were PPO plans (53.6%), followed by HMO (24.2%). Most (85.3%) of the patients did not have cirrhosis, 4.7% had compensated cirrhosis, and 10.1% had decompensated cirrhosis. Among all patients in the study, only 23.5% ever received at least 1 prescription for CHB medication. The mean follow-up duration was 3.2 ± 2.3 years.
Slightly over half (55.8%) of the patients were seen by GI or ID specialists. Patients seen by GI/ID were older (46.0 vs 45.6 years), more likely to be men (58.6% vs 56.5%), and more likely to have PPO insurance plans than those not seen by GI/ID (58.0% vs 48.0%). GI/ID patients were also more likely to have cirrhosis (18.7% vs 9.6%), with more comorbid conditions (diabetes, 14.6% vs 12.7%; cardiovascular disease, 10.6% vs 8.3%; and renal impairment, 5.1% vs 3.9%) (Table 1).
By provider specialty.
We found low rates of testing among both patients seen by GI/ID and those not seen by GI/ID, although significantly higher among GI/ID group. Only 64.5% of patients seen by GI/ID had both ALT and HBV DNA testing, compared with 51.6% of those not seen by GI/ID (P < 0.001). Similarly, 38.0% of GI/ID group had ALT, HBV DNA, and HBeAg testing vs 25.9% of the non-GI/ID group (P < 0.001). When each test was examined individually, a similar trend of low testing rates, although higher among patients seen by GI/ID, was also seen: 79.8% vs 75.6% for CBC, 82.2% vs 75.1% for ALT, 69.3% vs 59.0% for HBV DNA, and particularly low at 43.3% vs 32.8% for HBeAg and 39.7% vs 28.9% for anti-HBe, respectively (all P < 0.001) (Figure 1a).
Regarding coinfections (HCV, HDV, and human immunodeficiency virus), less than 20% of either group were assessed for each coinfection (all P < 0.001). Similarly, less than 20% were tested for hepatitis A immunity, and less than 5% of those without a previous infection or immunity for HAV ICD-9 diagnosis codes received the HAV vaccine, although testing and vaccination rates were also higher among patients seen by GI/ID (all P < 0.001) (Figure 1a).
Regarding HCC screening, 61.2% of patients seen by GI/ID underwent liver imaging, although the number was higher than that of the non-GI/ID group (53.3%), and approximately 59.9% of the GI/ID group and 57.0% of the non-GI/ID group had an AFP test completed (P < 0.001) (Figure 1a). Among cirrhotic patients, 54.2% of patients seen by GI/ID vs 44.0% of the non-GI/ID group had PT/INR tests performed, whereas more than 70% had their creatinine checked in both groups, with significantly higher testing rates among patients seen by GI/ID (all P < 0.001) (Figure 1a). In summary, although higher testing rates were seen among patients seen by GI/ID, the testing rates were still suboptimal for both the GI/ID and non-GI/ID groups.
By liver disease severity.
Figure 1b displays the percentage of completed tests among those with no cirrhosis, compensated cirrhosis, and decompensated cirrhosis. For each group, the testing rate was less than 50% for most tests and dismal at approximately 20% or less for coinfection and HAV immunity screening. Among the group with the most severe disease stage, that is, the decompensated cirrhosis group, only 39.6% had both ALT and HBV DNA testing, which was even lower than the testing rate of patients with compensated cirrhosis (59.5%) and patients without cirrhosis (61.0%) (P < 0.001). Similarly, only 24.0% of patients with decompensated cirrhosis had ALT, HBV DNA, and HBeAg checked, which was worse than the percentage of patients with compensated cirrhosis (34.5%) and patients without cirrhosis (33.6%) who had all 3 tests performed (P < 0.001). Furthermore, among patients decompensated cirrhosis, testing rates were 78.3% for CBC, 74.2% for creatinine, and 49.1% for INR (Figure 1b).
By sex and age.
Men were slightly more likely to have both ALT and HBV DNA testing (60.0% vs 57.2%, P < 0.001), but the rate of testing for ALT, HBV DNA, and HBeAg was similar between men and women (32.8% vs 32.4%) (P = 0.32) (see Supplementary Figure 1A, Supplementary Digital Content 1, http://links.lww.com/AJG/B992).
By age, patients aged 65 years and older were less likely than younger patients to undergo all the required tests with only 23.3% having both ALT and HBV DNA checked, when compared with those younger than 45 years (64.7%) and the 45- to 64-year-old group (55.9%) (P < 0.001). Similarly, only 10.0% of patients aged 65 years and older had ALT, HBV DNA, and HBeAg checked, when compared with patients younger than 45 years (39.2%) and patients aged 45–64 years (27.8%) (P < 0.001) (see Supplementary Figure 1B, Supplementary Digital Content 1, http://links.lww.com/AJG/B992).
Factors associated with undergoing optimal initial evaluation.
Optimal initial evaluation was defined as having ALT, CBC, HBV DNA, HBeAg, anti-HCV, anti-HAV, liver imaging, and AFP testing during the first year of follow-up. As noted in Table 2, on multivariable logistic regression adjusted for age, sex, insurance types, OOP expenditures, provider specialty, number of clinic visits per year, liver disease severity, and DCCI scores, being seen by GI/ID (OR: 1.86, 95% CI 1.70–2.05, P < 0.001) was associated with almost 2-fold the likelihood of having an optimal initial evaluation. Those with compensated cirrhosis (OR: 1.60, 95% CI 1.34–1.91) or decompensated cirrhosis (OR: 1.67, 95% CI 1.45–1.91) were also more likely to receive an optimal evaluation (all P < 0.001) (Table 2).
After patients who were seen for less than 1 year or experienced HCC or orthotopic liver transplantation during the first year were excluded from the initial evaluation cohort, 45,435 patients were available for evaluation of their long-term monitoring. Overall, 58.8% had at least annual clinic visits, but only 29.2% were seen at least every 6 months. Although 40.3% had their ALT level checked every 12 months, the rate of HBV DNA testing was lower at 28.0%. Among patients with cirrhosis, only 30.7% had a CBC, 20.6% had PT/INR, and 40.6% had creatinine checked every 12 months. Less than 15% had any of these tests every 6 months (Table 3, A).
By provider specialty.
Patients seen by GI/ID were more likely to have a clinic visit at least once every 12 months (59.8% vs 57.5%) and were more likely to undergo the recommended testing at least every 12 months: ALT (41.5% vs 38.6%), HBV DNA (29.8% vs 25.7%), or both ALT and HBV DNA (27.2% vs 22.2%) (P < 0.001 for all). Among patients with cirrhosis, those seen in a GI/ID clinic also had more CBC testing completed every 12 months than the non-GI/ID group (32.7% vs 28.1%, P < 0.001). Less than 16% had any given test completed every 6 months by either provider group (Table 3, A).
By liver disease severity.
There were also significant differences in the frequency of monitoring by liver disease severity, except for the comparison of tests for CBC and PT/INR at the 6-month intervals and creatinine at both the 6-month and 12-month intervals (Table 3, B). Cirrhotic patients had more frequent clinic visits compared with noncirrhotic patients every 6 months (43.4% vs 26.7%) and every 12 months (67.9% vs 57.3%). Although the frequency of clinic visits was similar between patients with compensated and decompensated cirrhosis, those with compensated cirrhosis were more likely to undergo testing for ALT, HBV DNA, or both compared with either the decompensated and the noncirrhotic groups, for both the 6-month and 12-month testing intervals. EGD was performed in approximately one-third of patients with compensated (37.5%) and decompensated (32.9%) cirrhosis and about one-fifth (20.9%) in noncirrhotic patients (P < 0.001); however, data for EGD indication were not available in the Truven databases.
By sex and age.
Although there was no significant difference by sex (every 6 months: 29.2% vs 29.1%; every 12 months: 58.9% vs 58.8%, P > 0.05) in the frequency of clinic visits, older patients (≥65 years) had more frequent clinic visits compared with the younger groups (P < 0.001) (see Supplementary Tables 2A and 2B, Supplementary Digital Content 1, http://links.lww.com/AJG/B992). However, men were more likely to undergo testing for ALT, HBV DNA, or both every 6 or 12 months than women, and those aged 65 years and older were less likely to undergo testing than younger patients (all P < 0.01).
Predictors of optimal long-term monitoring.
We defined optimal long-term monitoring as having ALT and HBV DNA levels checked and liver imaging performed at least every 12 months with a 3-month grace period. On multivariable analysis, being decompensated (OR: 0.67, 95% CI 0.60–0.74, P < 0.001) was associated with being less likely to receive optimal long-term monitoring, whereas being seen by GI/ID (OR: 1.31, 95% CI 1.24–1.39), having compensated cirrhosis (OR: 1.47, 95% CI 1.32–1.65), and more frequent clinic visits (OR: 1.06, 95% CI 1.05–1.07) were associated with optimal long-term monitoring (all P < 0.001) (Table 2).
Overall and by AASLD criteria.
Less than 5% of patients with CHB infection had HCC surveillance every 6 months regardless of provider specialty, liver disease severity, sex, or age (Figures 2a–c and see Supplementary Figures 2A and B, Supplementary Digital Content 1, http://links.lww.com/AJG/B992). Even among those considered high-risk (patients with cirrhosis, noncirrhotic men older than 40 years, and noncirrhotic women older than 50 years), only 40.3% had HCC surveillance every 6 months or every 7–12 months and 25.2% did not have any HCC surveillance (Figure 2a).
By provider specialty.
Among those seen in the clinic by GI/ID, 42.4% underwent HCC surveillance every 6 months or every 7–12 months, compared with 31.2% for those seen outside the GI/ID clinic (P < 0.001). However, 21.9% of patients seen in the GI/ID clinic and 33.7% seen outside the GI/ID clinic never had any HCC surveillance (Figure 2b).
By liver disease severity.
Fifty-six percentage of those with compensated cirrhosis underwent HCC surveillance every 6 months or every 7–12 months compared with 39.2% of those with decompensated cirrhosis and 36.5% without cirrhosis (P < 0.001). However, only 23.3% of those with decompensated cirrhosis, compared with 10.0% of those with compensated cirrhosis and 28.3% of those without cirrhosis, never underwent HCC surveillance (Figure 2c).
By sex and age.
Although differences in the rates of HCC surveillance by sex reached the conventional level of significance, this difference was not large, with 38.5% of men and 36.7% of women undergoing HCC surveillance every 6 months or every 7–12 months and 26.6% of men and 27.3% of women never undergoing HCC surveillance (see Supplementary Figure 2A, Supplementary Digital Content 1, http://links.lww.com/AJG/B992).
Thirty-eight percentage of patients aged 65 years and older underwent HCC surveillance every 6 months or every 7–12 months compared to 34.7% of those under age 45 and 40.8% of those aged 45–64 (P < 0.001). However, only 30.9% of aged 65 years and older, compared with 28.3% of those younger than 45 years and 25.1% of those aged 45–64 years, never underwent HCC surveillance (see Supplementary Figure 2B, Supplementary Digital Content 1, http://links.lww.com/AJG/B992).
Persistency of annual HCC surveillance among adherent patients.
Among those initially adherent to HCC surveillance as per study definition during the first 2 years of follow-up, only 62.0% of patients remained adherent to HCC surveillance at year 3, although by year 5, only 31.5% remained adherent. At year 5, those seen in the GI/ID clinic (32.5%) had slightly higher persistency than those not seen in the GI/ID clinic (29.5%, P < 0.0001), whereas those with compensated cirrhosis (40.9%) had higher persistency than those without cirrhosis (29.7%) or those with decompensated cirrhosis (25.6%, P < 0.001). There was no significant difference in surveillance persistency between men (31.8%) and women (31.1%, P = 0.21).
Predictors of HCC surveillance.
On multivariable regression analysis adjusted for age, sex, insurance type, OOP expenditure, provider specialty, number of clinic visits per year, liver disease severity, and DCCI score (Table 2), having GI/ID visits (OR: 1.14 95% CI 1.09–1.20), compensated cirrhosis (OR: 1.76, 95% CI 1.60–1.93), and more frequent clinic visits (OR: 1.09, 95% CI 1.08–1.10) were all associated with HCC surveillance at least every 12 months (all P < 0.001). On the other hand, decompensation (OR: 0.88 95% CI 0.81–0.95, P < 0.001) was associated with being less likely to undergo annual surveillance (Table 2).
To our knowledge, this is the largest study to date evaluating care delivery to a nationally representative sample of insured Americans with CHB. Previous studies mostly consisted of single centers, large academic centers, or the Veteran's Administration, which all have selection bias for sicker or more motivated patients seeking care at tertiary centers or less diversity in sex and ethnicity, making their findings less generalizable to the overall US population with CHB infection (20–22). We found that, overall, the initial evaluation of CHB was suboptimal in that only 79.1% underwent testing for ALT levels, 64.7% had HBV DNA testing, 38.7% had testing to check for CHB activity status with HBeAg, and less than 20% had testing for viral coinfections. HCC surveillance was also low, with approximately 60% having liver imaging or an AFP test. Testing among those with cirrhosis was still low, with 51.2% having PT/INR and 74.4% having their creatinine evaluated.
The lack of testing, on the other hand, may help explain the low treatment rate of 23.5% because the treatment decision is based on the results of ALT, HBV DNA, and HBeAg testing (17). We found that HBV DNA and HBeAg were not performed in approximately 40% of patients at their initial evaluation. In fact, over the long term, only 59% of patients visited a clinic every 12 months after their initial evaluation, and only 25% had both ALT and HBV DNA levels drawn, laboratory results that are crucial for the management of patients with CHB infection.
Unfortunately, the screening rate for HCC was also low. Four percentage of patients who met AASLD criteria for HCC screening were surveyed every 6 months regardless of whether they were seen at a GI/ID clinic or not. This finding can be partially explained by the fact that, overall, only 29% of patients attended a clinic visit every 6 months after their initial CHB visit. In fact, those with more clinic visits were 9% more likely to undergo HCC surveillance. Furthermore, HCC surveillance is a long-term commitment, but few previous studies have examined persistency rates among those who were initially adherent (31). In this study, we found that the persistency of HCC surveillance was very poor, even among those who had undergone annual surveillance during the first 2 years of follow-up, with a significant drop-off by year 5, when only 31.5% of these patients were still undergoing annual HCC surveillance.
Although we found suboptimal rates of HCC surveillance overall, patients with compensated cirrhosis had higher rates of adherence, and compensated cirrhosis was a predictive factor for adherence. This could be linked to the fact that physicians perceive only patients with CHB infection with cirrhosis as at risk for HCC and, thus, perform regular imaging follow-up. Similarly, patients with cirrhosis may also perceive themselves to be at high risk of HCC and, thus, adhere to physician recommendations.
Previous studies have shown somewhat of range in the rate of HCC surveillance. For example, a study of patients with cirrhosis seen at Parkland Health and Hospital System reported found that only 13% underwent annual surveillance (32). This could be due to differences in patients' socioeconomic status and education level and between patients with private insurance and those without private insurance. Regarding patients with CHB infection, particularly, a previous study with 4 US study centers (academic and community) reported that 78% of patients with cirrhosis and 60% of those without cirrhosis underwent annual HCC surveillance, compared with 44% of patients with cirrhosis and 36% of patients without cirrhosis in this study (31). This previous study may have found higher rates of surveillance because half of their patients were seen at university clinics. In fact, 71% of those patients seen at university clinics underwent annual surveillance compared with 46% of those seen at community clinics in this previous study, which is similar to rates observed in our study.
Explanations for these findings are multifaceted and complex involving the provider, the patient, and the environment. At the provider level, those seen in a GI/ID clinic were more likely to undergo optimal testing and long-term monitoring, suggesting that there could be lack of provider knowledge, especially in nonsubspecialists, on diagnosis and management of HBV as outlined by the AASLD guidelines (19). One study of primary care providers who cared for a predominantly Asian American population found that only 43% were familiar with guidelines on CHB management (33,34). There may also be lack of time in the primary care setting, when providers have to address multiple issues and cannot focus on CHB care, as is more likely the case in subspecialty GI/ID clinics. There could also be a potential referral bias as to why GI/ID clinics have higher rates of testing, treatment, and surveillance, in that because referral to a specialist can often be a process initiated by a physician, it is possible that only patients perceived by the primary care physician as having more advanced liver disease and/or more in need for treatment or follow-up were referred to the specialty clinic. Even though the DCCI score was not significantly different, those seen in the GI/ID clinic had higher rates of cirrhosis, decompensation, cardiovascular disease, diabetes, and renal impairment. This referral bias might also influence patient behavior in that patients may see themselves as sicker and, thus, become more adherent to physician guidance (35,36). It is disappointing that even though the adherence rate is higher among patients seen in a GI/ID clinic and we did find that being seen by GI/ID was associated with optimal care on multivariable logistic regression, it is still far from perfect.
Another notable finding is that only 56% of our patients were linked to specialist care, and this is relevant because we found that specialist care was associated with better adherent to management guidelines. At the patient level, this could be due to the asymptomatic nature of the disease, poor understanding of the disease, and lack of awareness of long-term benefits such as prevention of cirrhosis or cancer, which may be intangible to patients, but the short-term costs we reported as OOP expenditures are immediate and tenable concerns. The physical discomfort of laboratory draws may also deter patients from following up on their appointments (37,38). In addition, language barriers may exist because many of the patients in this study could be immigrants from areas outside the United States where HBV is endemic, another very relevant point because approximately two-third of CHB in the United States are “imported” (3). In 1 survey, physicians reported the presence of communication barriers even in the presence of a translator or when the language was the physician's native tongue, suggesting the presence of additional complex nonlanguage social and cultural factors (39).
Especially discouraging were the low rates of long-term CHB monitoring and HCC surveillance for those with decompensated cirrhosis, the most advanced stage of CHB. Reasons for this need further investigation but could be that many of the patients were not candidates for transplantation; however, the reasons for their ineligibility would need further elucidation because the average age of our study population was only 46 years. Nevertheless, CHB with decompensated cirrhosis would require close monitoring and management of complications. In addition, it should be noted that only 41.4% of patients with decompensated cirrhosis underwent HBV DNA testing, an especially important test in this population, because antiviral therapy is recommended for viremic decompensated patients because it can significantly improve survival and can even reverse decompensation to compensated disease (9,17).
We acknowledge the following limitations. First, our study population consisted of patients with commercial insurance or Medicare with private insurance supplement, so our findings may not be generalizable to those who receive government-sponsored insurance, those who are uninsured, or other special populations. However, we would suggest that the rates of testing and adherence would be even lower than what we have described for those with insurance because people with insurance with less coverage may have more barriers obtaining the recommended testing and monitoring. Furthermore, patients with private insurance may have higher income and are more educated, and these factors may also increase adherence (32). Second, we might have overestimated the rates of imaging for HCC surveillance because we were unable to discern the indication for each imaging, so some may not have been performed for HCC screening or surveillance, although this would not change the direction of our conclusion. Third, we used a very liberal definition for optimal monitoring, with ALT and DNA testing every 12 months, which also was not stratified according to HBeAg status that was often unavailable, but this would only make our findings even more alarming. Fourth, we also cannot determine the exact indication for the clinic visit. However, we surmise that if more stringent criteria were applied, the adherence rate would be even lower and, again, would not alter the direction of our conclusion. Fifth, data were not available as to why patients were referred to GI/ID specialists, but as summarized in Table 1, patients seen by GI/ID specialists were more likely to have cirrhosis, both compensated and decompensated. As such, there could be referral bias in that sicker patients were referred to GI/ID and these patients perceived themselves as such and were more adherent. In addition, data were not available as to why even patients seen by GI/ID specialists did not have optimal care, but we surmised that this is due to both provider and patient factors. For example, busy providers or providers not familiar with practice guidelines including GI/ID specialists may have failed to order recommended tests, or appropriate tests have been ordered by providers but not followed through by patients. Therefore, additional studies and practice quality measures are needed. Sixth, we cannot be certain that patients did not have a CHB diagnosis before our study period. However, data search occurred over a 7-year study period and required a 12-month observation period. Therefore, if the patients had been diagnosed but did not get these laboratory evaluations or were not monitored, then the patients were not adherent to guidelines; consequently, the direction of our results would not be affected. Finally, because of the retrospective design of our study, we were unable to determine the source of nonadherence, meaning we were unable to determine whether the physician ordered the test and the patient did not follow through or if the test was not ordered.
To our knowledge, this study used the largest and most diverse population with CHB infection to characterize the status of CHB care in the United States and to determine factors associated with optimal initial evaluation and long-term monitoring, including HCC surveillance, for patients with CHB. Overall, we found that initial testing, long-term monitoring, and HCC surveillance were all poor, even in those with cirrhosis. Although patients fared better if seen in a GI/ID clinic, rates of optimal monitoring still fell short of AASLD recommendations. Less than 1 in 4 patients received treatment for CHB, which may be associated with the lack of required testing before consideration of treatment. Compensated cirrhosis, lower out-of-pocket expenses, GI/ID visits, and more frequent clinic visits were all associated with adherence to initial evaluation, long-term monitoring, and HCC surveillance. More work is required to better understand the barriers preventing patients with CHB from receiving optimal care.
CONFLICTS OF INTEREST
Guarantor of the article: Mindie H. Nguyen, MD, MAS.
Specific author contributions: S.T. and M.H.N.: study design. D.J., S.T., and M.H.N.: data collection. D.J., S.T., and M.H.N.: data analysis. S.T., L.H., and M.H.N.: drafting of the manuscript. All authors: data interpretation and review/revision of the manuscript. M.H.N.: study concept, study supervision, and guarantor of the study.
Financial support: None to report.
Potential competing interests: R.C.C.: research grant: Gilead. M.H.N.: research support: Gilead, Pfizer, Enanta, B.K. Kee Foundation, National Cancer Institute; consultant and/or an advisory board: Novartis, Bayer, Eisai, Intercept, Gilead, Janssen, Laboratory of Advanced Medicine, Exact Sciences, and Intercept. All other authors have nothing to disclose.
WHAT IS KNOWN
- ✓ Close monitoring of chronic hepatitis B (CHB) is essential, as CHB spontaneously fluctuates between different phases and not all phases require treatment.
- ✓ Prior studies, mostly small and single-center, have suggested poor adherence to recommended care.
WHAT IS NEW HERE
- ✓ In a large, nationwide cohort of insured patients, only 33% had alanine aminotransferase, hepatitis B virus DNA, and hepatitis B e antigen tests at initial CHB evaluation.
- ✓ For long-term monitoring, only 25% had both alanine aminotransferase and hepatitis B virus DNA tests done annually and less than 40% underwent annual hepatocellular carcinoma surveillance.
- ✓ Gastroenterology/infectious disease visits and compensated cirrhosis were predictors of optimal CHB initial evaluation and long-term monitoring.