Hepatitis C virus (HCV) is the most common

Transcription

1 Meeting Vaccination Quality Measures for Hepatitis A and B Virus in Patients with Chronic Hepatitis C Infection Jennifer R. Kramer, 1,2 Christine Y. Hachem, 5 Fasiha Kanwal, 4,5 Minghua Mei, 1,2 and Hashem B. El-Serag 1,2,3 Coinfection with hepatitis A virus (HAV) or hepatitis B virus (HBV) in patients with chronic hepatitis C virus (HCV) is associated with increased morbidity and mortality. The Center for Medicare and Medicaid Services has identified HAV and HBV vaccination as a priority area for quality measurement in HCV. It is unclear to what extent patients with HCV meet these recommendations. We used national data from the Department of Veterans Affairs HCV Clinical Case Registry to evaluate the prevalence and predictors of meeting the quality measure (QM) of receiving vaccination or documented immunity to HAV and HBV in patients with chronic HCV. We identified 88,456 patients who had overall vaccination rates of 21.9% and 20.7% for HBV and HAV, respectively. The QM rates were 57.0% and 45.5% for HBV and HAV, respectively. Patients who were nonwhite or who had elevated alanine aminotransferase levels, cirrhosis, or human immunodeficiency virus were more likely to meet the HBV QM. Factors related to HCV care were also determinants of meeting the HBV QM. These factors included receiving a specialist consult, genotype testing, or HCV treatment. Patients who were older, had psychosis, and had a higher comorbidity score were less likely to meet the HBV QM. With a few exceptions, similar variables were related to meeting the HAV QM. The incidence of superinfection with acute HBV and HAV was low, but it was significantly lower in patients who received vaccination than in those who did not. Conclusion: Quality measure rates for HAV and HBV are suboptimal for patients with chronic HCV. In addition, several patient-related factors and receiving HCV-related care are associated with a higher likelihood of meeting QMs. (HEPATOLOGY 2011;53:42-52) Abbreviations: ALT, alanine aminotransferase; CCR, Clinical Case Registry; CI, confidence interval; DRG, diagnosis-related group; ESRD, end stage renal disease; HAV, hepatitis A virus; HBV, hepatitis B virus; HCV, hepatitis C virus; HIV, human immunodeficiency virus; ICD-9, International Statistical Classification of Diseases and Related Health Problems, 9th edition; Ig, immunoglobulin; IRR, incidence rate ratio; QM, quality measure; VA, Veterans Affairs. From the 1 Houston VA Health Services Research & Development Center of Excellence, Michael E. DeBakey Veterans Affairs Medical Center, Houston, TX; 2 Health Services Research, and 3 Gastroenterology, Department of Medicine, Baylor College of Medicine, Houston, TX; and the 4 John Cochran VA Medical Center, St. Louis, MO; and 5 Department of Gastroenterology and Hepatology, Saint Louis University School of Medicine, St. Louis, MO. Received March 26, 2010; accepted September 23, The research reported here was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service Grant, MRP (to J.R.K) and a grant from Bristol- Meyers Squibb as a part of the Virology Fellows Research Training Program (to C. Y. H.). Dr. El-Serag is supported by National Institutes of Health Grant K24DK and The Texas Gulf Coast Digestive Diseases Center (National Institutes of Health Grant P50 DK56338). Address reprint requests to: Jennifer R. Kramer, PhD., M.P.H., 2002 Holcombe Boulevard (152), Houston, TX jkramer@bcm.edu; Fax: Copyright VC 2010 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /hep Potential conflict of interest: Nothing to report. Hepatitis C virus (HCV) is the most common blood-borne chronic infection in the United States, with 4.1 million (1.6%) individuals infected, of whom 3.2 million (1.3%) have chronic HCV. 1 Veterans have been disproportionately affected by HCV, with reported prevalence rates ranging from 5% to 35%, 2-5 resulting in a substantial increase in the burden of liver disease in the Department of Veterans Affairs (VA) health care system. 6 Multiple guidelines and consensus panels recommend vaccination for hepatitis A virus (HAV) and hepatitis B virus (HBV) in patients with chronic HCV infection The Center for Medicare and Medicaid Services has identified HAV and HBV vaccination rates as one of the priority areas for quality measurement in the management of individuals with HCV infection. HAV and HBV vaccination rates in individuals with HCV are now part of Medicare s Physician Quality Reporting Initiative a voluntary program that ties a reimbursement incentive to compliance with performance on a range of quality measures (QMs). 12 Medicare defines vaccination 42

2 HEPATOLOGY, Vol. 53, No. 1, 2011 KRAMER ET AL. 43 QMs as the percentage of patients aged 18 years and older with a diagnosis of hepatitis C who received at least one injection of hepatitis A (or B) vaccine, or who have documented immunity to hepatitis A (or B). With the advent of these changes, it is important to understand the current process of HAV and HBV vaccination in HCV and to measure whether there are systematic deficiencies in the receipt of such care. Few existing data indicate wide variation in these vaccination rates in patients with chronic HCV Combining vaccination rates with the evidence of previous immunity, a measure that is consistent with the aforementioned Medicareendorsed performance indicators, Hernandez et al. 14 found that 30% of HCV patients had no documentation of vaccination or prior immunity to HAV and HBV. Although important, these studies have focused on single centers and have had a limited ability to adjust for the range of factors associated with lack of vaccination. In this study, we determined the proportions of patients who met the Medicare QM for vaccination for HAV and HBV in a national cohort of more than 80,000 HCV-infected patients who received care in VA facilities throughout the United States. We also examined the patient- and facility-level determinants of the HAV and HBV vaccination QM rates. Patients and Methods Data Sources We used data from the VA HCV Clinical Case Registry (CCR), which contains health-related information on all known HCV-infected VA users nationwide. The CCR automatically identifies patients from HCV antibody testing results as well as International Statistical Classification of Diseases and Related Health Problems, 9th edition (ICD-9) codes. Data elements in the CCR include demographics, laboratory test results, outpatient and inpatient VA pharmacy data, and inpatient and outpatient diagnoses and procedure codes. Additional details of the CCR data are published elsewhere. 17 We examined data sets obtained from the VA HCV CCR database for patients diagnosed with HCV between January 1, 2000, and January 1, Study Population Patients with chronic HCV, as indicated by at least one positive RNA or genotype test for HCV and at least one outpatient visit, were included in the cohort. The index date for HCV diagnosis reflected the date of the earliest positive test for HCV (antibody, polymerase chain reaction, or genotype), or first appearance of an ICD-9 code for HCV. We excluded patients who had metastatic cancer (ICD-9 code ) or who were in a hospice bed section within 1 year before the HCV index date and up to 2 years after the index date, and patients who died or had more than 500 visits in the 2 years following the index date. The study protocol was reviewed and approved by the Institutional Review Board at the Baylor College of Medicine. Definition of Outcome Variables We used a combination of serology and vaccination information to define the outcome variables. HAV past or present infection was defined by the presence of a positive serology test for HAV antibody (immunoglobulin [Ig] G or IgM). HBV past or active infection was defined by the presence of a positive serology test for HBV surface antigen, HBV surface antibody, HBV core antibody (total, IgG, or IgM), HBV DNA, HBV e antibody, or HBV e antigen. Presence of HAV or HBV vaccinations in the CCR were identified by Current Procedural Terminology codes for hepatitis A vaccination (90632, 90636, or 90730) and hepatitis B vaccination (90636, 90740, 90746, 90747, 90748, or G0010). The main outcome of interest was meeting the physician-derived QMs for HAV and HBV vaccination in HCV-infected patients. 18 Meeting the QM was defined by the presence of a positive serology test (in the absence of vaccination) or a vaccination regardless of serology anytime before or within the 2 years after the HCV index date. We also examined the possible effect of vaccination on acute infection with HBV and HAV (i.e., superinfection with HBV or HAV). For HBV, we examined a cohort of patients who were not acutely or chronically infected with HBV or had past exposure to HBV; this was defined with a baseline negative HBV surface antigen and a negative HBV core antibody test within 1 year of each other. In this cohort, we examined the subsequent positive tests for HBV surface antigen or HBV DNA and compared the incidence of acute HBV superinfection among those with at least one HBV vaccination compared with those without HBV vaccination. For HAV, we examined all patients with a baseline negative test for any antibodies to HAV. We then examined subsequent HAV IgM tests to compare the incidence of HAV superinfection (defined as a positive HAV IgM test) among patients with at least one HAV vaccination versus those without HAV vaccination. Definition of Exposure Variables Patient Factors. We identified several sociodemographic factors including age, race (black, Hispanic,

3 44 KRAMER ET AL. HEPATOLOGY, January 2011 Table 1. Operational Definitions for Clinical and Laboratory Variables Criteria Major, uncontrolled depressive illness Current users of illicit drugs Current users of alcohol HIV coinfection Cirrhosis Anxiety/posttraumatic stress disorder Bipolar Psychosis Nonmetastatic cancer Chronic renal disease Elevated ALT HCV genotype HCV treatment Operational Definitions Any hospital admission with a diagnosis of depression or 3 outpatient visits with diagnosis of depression; must have occurred before HCV treatment in patients who received treatment Inpatient primary ICD-9 OR 2 outpatient ICD-9 codes for drug or alcohol dependence/abuse Inpatient primary ICD-9 code for alcohol or 2 outpatient ICD-9 codes for alcohol dependence/abuse Any positive blood test (antibody, western blot, or viral load) or ICD-9 code for HIV Any ICD-9 code for cirrhosis or its complications (ascites, varices, encephalopathy, hepatitis with coma, and portal hypertension) in inpatient or outpatient files Any primary inpatient ICD-9 code for anxiety/posttraumatic stress disorder Any primary inpatient ICD-9 code for bipolar disorder Any primary inpatient ICD-9 code for psychosis Any inpatient or outpatient ICD-9 code for nonmetastatic cancer during the entire time before the HCV index date and 2 years after Any inpatient or outpatient code for end stage renal disease ALT >40 IU/mL for all tests within the 2 years before or after HCV index date Genotype 1 or 4, genotype 2 or 3, or not tested Presence of at least one filled prescription of interferon, pegylated interferon, or ribavirin in the outpatient pharmacy files in the 2 years after the HCV index date For ICD-9 codes, we used the Agency for Healthcare Research and Quality clinical classification system to classify all patient ICD-9 codes into the relevant diagnosis. With the exception of variables with timing mentioned above, the remaining variables were defined as the presence of the code 2 years before or after the HCV index date. other, white, and unknown), sex, and period of service (Vietnam era versus others). We also examined HCV genotype, receipt of HCV treatment, and persistently high (>55 IU/mL) alanine aminotransferase (ALT) levels. Comorbidities defined by ICD-9 codes using the Agency for Healthcare Research and Quality criteria included cirrhosis, alcohol and drug abuse, psychiatric illnesses, human immunodeficiency virus (HIV) infection, and end stage renal disease (ESRD) (see Table 1 for definitions). We also calculated a scaled relative risk score to adjust for presence and severity of several additional comorbidities. This score has been validated to predict mortality in VA patients. 19 We examined whether or not patients visited a VA HCV specialist within 2 years before or after their HCV index date. We defined HCV specialists as providers (physicians, nurse practitioners, physician assistants, or pharmacists) associated with gastroenterology/ hepatology or infectious disease clinics. We also examined the mean number of inpatient and outpatient visits in the 2 years after the HCV index date and receipt of HCV antiviral treatment within 2 years after their index date. HCV antiviral treatment was defined as the presence of a prescription for interferon or ribavirin in the pharmacy files of the CCR data. Facility Variables. We assigned patients to the VA facility where the majority (>50% of outpatient or inpatient encounters) of their care was obtained during the 2 years after their HCV index year. We evaluated several facility level variables. The first variable represented the overall patient burden and complexity of illness seen at that facility, measured using an average scaled relative risk score created using DxCG software. 20,21 The second variable reflected facility complexity as measured by diagnosis-related group (DRG) index, which is the product of the relative weights of the DRGs multiplied by the volume of patients seen at the facility in each DRG. The third variable was the number of hospital beds in each facility. The fourth variable measured the facility s load of HCV-infected patients, which was defined as the total number of HCV RNA positive patients seen at each facility during Statistical Analyses We plotted the overall proportion of patients who met the QM for HBV and HAV vaccination by facility with their corresponding 95% confidence intervals. Possible determinants of meeting the QMs for HAV and HBV vaccination were assessed by chi-square tests for categorical variables and Wilcoxon tests for continuous variables. Adjusted odds ratios and Wald 95% confidence intervals were estimated using hierarchical logistic regression models with random effects for the clustering of patients within facilities. The contribution of clustering within facilities to the amount of variation in HCV treatment was measured by empirical Bayes estimates of the cluster random effects. Model building followed a forward selection approach, in which the Akaike information criterion was minimized subject to the constraint that Wald P values for independent variables did not exceed Finally, we estimated the proportion of variation explained by a given model using the Efron pseudo-r 2 definition. 22,23 We

4 HEPATOLOGY, Vol. 53, No. 1, 2011 KRAMER ET AL. 45 Table 2. Characteristics of 88,456 HCV RNA Positive Patients and Relationship with QMs for HBV and HAV Vaccination HBV Vaccination HAV Vaccination Variable No. (%) Percent (%) Percent (%) Not Percent (%) Percent (%) Not Demographics Race/ethnicity White (non-hispanic) 37,429 (42.3) 41.1* * 43.0 Black 22,319 (25.2) Hispanic 3,737 (4.2) Other 1,145 (1.3) Unknown 23,826 (27.0) Male sex 85,652 (96.8) Vietnam era of service 58,801 (66.5) 68.6* * 65.7 Age (years) <45 13,239 (15.0) 14.1* * ,237 (60.2) ,758 (18.9) ,222 (5.9) Laboratory tests HCV genotype 1 or 4 44,502 (50.3) 55.1* * or 3 11,359 (12.8) Not tested 32,595 (36.9) High ALT 46,785 (52.9) 53.5* * 52.1 Clinical diagnoses Cirrhosis 5,931 (6.7) 7.6* * 5.7 Alcohol abuse 15,995 (18.1) 18.6* * 19.0 Drug abuse 16,200 (18.3) 19.5* * 19.2 Anxiety or PTSD 1,851 (2.1) 2.3* Bipolar disorder 990 (1.1) * 1.2 Severe depression 6,322 (7.1) 7.7* * 6.9 Psychosis 1,861 (2.1) 1.2* * 2.4 HIV 2,208 (2.5) 3.4* * 1.8 ESRD 1,644 (1.9) * 2.0 Scaled relative risk score, median (IQR) 0.3 ( ) 0.30* * 0.30 Additional HCV care Specialist consult 69,689 (78.8) 85.0* * 71.8 Treatment for HCV 12,597 (14.2) 17.5* * 9.9 Number of total visits, median (IQR) 39 (21-76) 44* 33 44* 35 QM is defined as having immunity or being vaccinated irrespective of immunity status. Abbreviations: IQR, interquartile range; PTSD, posttraumatic stress disorder. *P < P < calculated these R 2 values first for the facility level random effects, and subsequently for a model that included patient level fixed effects, then facility level fixed effects, and finally for a full model with both facility level random and fixed effects and patient level fixed effects. Results Study Population Characteristics. We identified 88,456 patients with chronic HCV who fulfilled our inclusion/exclusion criteria. The median age of the HCV-infected patients was 51 years old, and 96.8% were male (Table 2). Approximately 42.3% were white, 25.2% were African American, 4.2% were Hispanic, 1.3% were of another race/ethnicity, and 27% were of unknown race/ethnicity. The majority of patients (66.5%) served during the Vietnam era. Of patients who received an HCV genotype test (63%), most had genotype 1 or 4 (79.7%). Approximately half the patients had persistently high ALT tests, and 6.7% had cirrhosis. Vaccination and Quality Measure Rates. Of the 88,456 patients, 82.2% and 68.6% were tested for HBV or HAV serology, respectively (Fig. 1A,B). Of the patients tested, 57.2% and 63.8% had no immunity to HBV or HAV, respectively, and 41.7% and 38.2% of these were subsequently vaccinated for HBV or HAV, respectively. Of patients not tested for HBV or HAV serology, only 12.6% and 12.8% were vaccinated for HBV and HAV, respectively. The overall vaccination rates were 21.9% for HBV and 20.7% for

5 46 KRAMER ET AL. HEPATOLOGY, January 2011 Fig. 1. (A) HBV and (B) HAV serology test and vaccination status of 88,456 veterans with chronic hepatitis C virus. *Included in calculation of meeting overall QM. HAV. The QM rates were 57.0% for HBV and 45.5% for HAV. The overall rates were slightly higher in patients with cirrhosis: 25.4% and 23.7% for vaccination rates for HBV and HAV, and 64.9% and 53.7% for QM rates for HBV and HAV (Fig. 2A,B). Patient Characteristics Associated with HBV and HAV Vaccination Quality Measures. For both HAV and HBV vaccination, patients who met the QMs via immune status or receiving vaccination regardless of immune status were significantly more likely to have genotype 1 or 4, persistently high ALT levels, psychiatric diagnoses (alcohol or drug abuse, severe depression, and psychosis), clinical comorbidities (cirrhosis, end stage renal disease, and HIV), or HCV treatment (Table 2). These patients had a significantly higher scaled relative risk score and had more visits to the VA (44 versus 33) than patients who did not meet the QMs. An exception was anxiety/ posttraumatic stress disorder, which predicted meeting the HBV vaccination QM, but not the HAV vaccination QM. Provider and Facility Characteristics Associated with HBV and HAV Vaccination Quality Measures. Most (79%) patients had seen an HCV specialist within the 2 years before or 2 years after their HCV index date (Table 2). However, patients who met the QMs for both HBV and HAV vaccination were significantly more likely to have seen an HCV specialist. Figure 3A,B shows the variation in vaccination QM rates for HBV and HAV vaccination across the 127 VA facilities included in this analysis. The range of HBV QM rates by facility was 22.5%-87.3% (median, 58.5%). However, the interquartile range was 50.0%- 66.9%. For HAV, the range was 9.6%-82.9% (median, 45.4%). The interquartile range was 35.1%-54.1%. The patients who met the HBV QM were more likely to receive care at facilities with a higher DRG index than patients who were not immune or vaccinated, indicating fewer complicated procedures (Table 3). They were also more likely to be associated with facilities with

6 HEPATOLOGY, Vol. 53, No. 1, 2011 KRAMER ET AL. 47 Fig. 2. (A) HBV and (B) HAV serology test and vaccination status of 5,931 veterans with chronic hepatitis C virus and cirrhosis. *Included in calculation of meeting overall QM. more hospital beds and a higher HCV-infected patient load. For the HAV QM, a higher DRG index and HCV-infected patient load were also significant facility predictors. In addition, patients who received care at facilities with higher relative risk scores, indicating a sicker patient population, were more likely to be immune or receive vaccination for HAV. Multivariate Predictors of HBV and HAV Vaccination Quality Measures. In the multivariate hierarchical model, patients were more likely to meet the HBV QM if they were black, Hispanic, or other race compared with white race, middle age (45-54 and versus <45), or service during the Vietnam era (Table 4). Patients with elevated ALT, cirrhosis, or HIV were also more likely to meet the HBV QM. Additional factors related to HCV care such as specialist consult, receipt of HCV treatment, and number of inpatient/outpatient visits were also predictive of meeting HBV QM. Older patients (65), patients with psychosis, and a higher relative risk score (reflecting more comorbidities) were less likely to meet the HBV QM. For HAV, similar variables and magnitude were related to meeting the HAV QM, with a few exceptions. These included patients 65 years of age who were more likely to meet the HAV QM, patients serving in the Vietnam era who were less likely to meet this QM, and those with a few additional clinical characteristics, such as alcoholism, drug abuse, and end stage renal disease, who were less likely to meet this QM. In addition, persistently elevated ALT was not significant for the HAV QM. VariationinHBVandHAVQualityMeasuresRates Attributed to Provider and Facility Characteristics. The facility random effect model explained 3.5% and 8.4% of the variability for HBV and HAV QMs, respectively (Table 5). This means that more variation was

7 48 KRAMER ET AL. HEPATOLOGY, January 2011 Fig. 3. Variation of (A) HBV and (B) HAV vaccination QM rates across 127 VA facilities. explained by facility for HAV than for HBV vaccination. The patient level effects explained roughly 7% of the variation for both HBV and HAV QMs. The full model with patient level fixed effects and facility random effects explained 13% and 16.7% of the variation for HBV and HAV QMs, respectively. Association Between Vaccination and Acute Infection with HBV and HAV. A total of 35,435 patients had a negative HBV surface antigen test and HBV core antibody test within 1 year of each other. Only 39.0% had a subsequent test for HBV surface antigen. The incidence of HBV superinfection in the unvaccinated group was 1.64%, whereas it was only 0.38% in the vaccinated group (incidence rate ratio [IRR] 4.34; 95% confidence interval [CI] ; P < ). In a sensitivity analysis that included patients with no subsequent test as negative, the incidence of HBV superinfection changes to 0.67% for the unvaccinated group versus 0.16% in the vaccinated group (IRR 4.13; 95% CI ; P < ). The results did not change when we examined these IIRs among patients with a negative HBV surface antigen test regardless of core status (data not shown).

8 HEPATOLOGY, Vol. 53, No. 1, 2011 KRAMER ET AL. 49 Table 3. Facility Characteristics of Primary Facility for 88,456 RNA-Positive Patients and Relationship with QMs for HBV and HAV Vaccination HBV Vaccination HAV Vaccination Facility factors Median (IQR) Median for Median for Not Median for Median for Not Average scaled relative risk score 1.11 ( ) * 1.11 DRG index 202 (0-550) 243* * 125 No. of hospital beds 171 ( ) 172* Facility HCV patient load 6,494 (3,762-8,903) 6,885* 6,289 6,957* 6,440 QM is defined as having immunity or being vaccinated irrespective of immunity status. Abbreviations: IQR, interquartile range. *P < For HAV, 53,965 had a negative test for HAV antibodies. Only 36.7% had a subsequent test for HAV. The incidence of HAV superinfection in the unvaccinated group was 0.16%, whereas it was only 0.01% in those vaccinated (IRR 14.25; 95% CI ; P ¼ ). In a sensitivity analysis that considered patients with no subsequent HAV tests as negative for HAV, the incidence of HAV superinfection drops to 0.05% in the unvaccinated group versus 0.004% in the vaccinated group (IRR 12.58; 95% CI ; P ¼ ). Discussion This is the first study to investigate HBV and HAV vaccination QMs in a national health care system. We found that vaccination rates for HBV and HAV in patients with chronic HCV are relatively low (21.9% Table 4. Predictors of HBV and HAV Vaccination QM Estimates from Hierarchical Logistic Regression of 88,456 HCV RNA Positive Patients HBV Vaccination QM HAV Vaccination QM Variable Adjusted Odds Ratio (95% CI) P Value Adjusted Odds Ratio (95% CI) P Value Demographic characteristics Race/ethnicity (versus white) Black 1.24 ( ) < ( ) Hispanic 1.10 ( ) ( ) < Other 1.13 ( ) ( ) Unknown 1.04 ( ) ( ) Age, years (versus <45) ( ) < ( ) < ( ) < ( ) < ( ) < ( ) < Vietnam era 1.08 ( ) ( ) < Laboratory tests HCV genotype (versus not done) 1 or ( ) < ( ) < or ( ) < ( ) < Elevated ALT 1.04 ( ) Clinical characteristics Cirrhosis 1.28 ( ) < ( ) < Alcohol abuse 0.92 ( ) Drug abuse 0.86 ( ) < Psychosis 0.81 ( ) ( ) HIV 2.42 ( ) < ( ) < End stage renal disease 0.76 ( ) < Relative risk score 0.98 ( ) ( ) < Additional HCV care Specialist consult 1.72 ( ) < ( ) < Treatment for HCV 1.71 ( ) < ( ) < Number of total visits 1.00 ( ) < ( ) < QM is defined as having immunity or being vaccinated irrespective of immunity status.

9 50 KRAMER ET AL. HEPATOLOGY, January 2011 Table 5. Relative Contribution of Facility and Patient Factors in Explaining Percent Variation in HBV and HAV Vaccination QMs in Patients with Chronic HCV Model and Effects HBV Vaccination QM Efron R 2 HAV Vaccination QM Empty model Facility random effect 3.51% 8.37% Full model Patient level fixed effects only 7.21% 7.13% Patient level fixed effects and facility random effect 13.05% 16.72% QM is defined as having immunity or being vaccinated irrespective of immunity status. for HBV and 20.7% for HAV). However, when we take into account patients with positive serology tests, the percentage meeting the QM (immune or received vaccination) for HBV and HAV increases to 57.0% and 45.5%, respectively. This indicates that 43% and 54.5% of patients were neither tested nor vaccinated, or were tested and known to be susceptible but were not vaccinated for HBV and HAV. Several important patient factors predict meeting the QMs for HBV and HAV vaccination. In addition, receiving other aspects of HCV-related care (such as seeing a specialist, receiving genotype testing or HCV treatment) is associated with a higher likelihood of meeting QMs for vaccination. We did not find any fixed facility effects that were associated with QM rates; however, there is some variability explained by the random facility effects, especially for HAV. We also found that whereas the overall incidence of HBV and HAV superinfection was low in this group, the incidence was significantly lower in those with a vaccine compared with those without a vaccine. Within the limitations of this study, we felt that these data support the AASLD guidelines recommending that patients with chronic HCV infection who lack antibodies to HAV and HBV should be vaccinated against these infections. This is the first study of this size to examine the effect of vaccination on superinfection with HAV and HBV in patients with chronic HCV. However, this analysis was limited by the fact that we defined vaccination as the presence of at least one vaccine Current Procedural Terminology code, did not examine vaccination completion, and by the lack of systematic surveillance or testing for HBV or HAV following the baseline measurement. Many patients (60%) did not have subsequent HBV or HAV testing. In addition, due to the small numbers of incident cases and the lack of detailed clinical information, we were unable to examine the clinical impact of these superinfections in this study. Larger numbers, longer follow-up time, and more detailed review of medical records are required to better assess the effect of superinfection on outcomes such as the development of fulminant hepatic failure, liver cancer, and/or death. Several previous studies examined HAV vaccination rates in single VA medical centers and found variable results. Shim et al. 15 reported a 7.9% vaccination rate and 35% QM rate in the VA New York Harbor Healthcare System, while Hernandez et al. 14 reported from the VA Palo Alto Health Care System a 32% vaccination rate and 71% QM rate. Our group also previously reported a vaccination rate of 7.9% for HAV from the Michael E. DeBakey VA Medical Center in Houston, TX. 13 We found rates somewhere in between in this study investigating the vaccination practices nationally in the VA. These differences in the reported rates from single-center studies support our finding of significant variation explained by facilities (while adjusting for patient factors) in HAV vaccination QMs. Hernandez et al. 14 also examined HBV patients and found that 43% were vaccinated and 70% met the QM. This is higher than the 21.9% vaccination rate and 57.0% QM rate we found for HBV in the VA nationally. In addition, the vaccination and QM rates reported here were consistently higher for HBV than HAV. We found several patient-level factors that were significantly associated with meeting QMs for HAV and HBV vaccination. Patients having these characteristics were more likely to have been vaccinated or tested and found to be immune. For example, patients with HIV were two and half times more likely to meet the QM for HBV and almost twice as likely to meet the QM for HAV. Because of the shared risk factors for HBV and HIV, this is likely due to immunity rather than vaccination. Regardless, these patients were tested and found to be protected from the virus. Receiving other aspects of HCV care, such as genotype testing, HCV treatment, and consultation with a specialist, were associated with meeting QMs for vaccination. This was probably due to these patients receiving more comprehensive care for their HCV. Although we adjusted for several potential confounding demographic and clinical factors to minimize selection effects, we cannot exclude the possibility that the patients who were referred to a specialist were systematically different from those who are not referred. Several of the fixed facility factors (e.g., DRG index and facility HCV patient load) that we examined in this study and that were significant in the univariate analysis were no longer significant in the multivariable model that adjusted for demographics and

10 HEPATOLOGY, Vol. 53, No. 1, 2011 KRAMER ET AL. 51 comorbidities. Therefore, the facility variation we found was likely due to other factors that we were not able to measure in this study. These may include variations in the presence of a dedicated HCV clinic, use of template algorithms and clinical reminders in the electronic medical record, and affiliation with academic centers. This study takes advantage of data from the HCV CCR, which is a combination of clinical and VA administrative data. Most QMment initiatives for other conditions have used either medical record or administrative resources. 24,25 Compared with administrative data, medical records contain more complete medical information and are therefore preferable in most situations. However, medical records are costly to obtain and abstract, whereas use of administrative data was more feasible and efficient in terms of time, effort, and resources needed to assess the baseline quality of care as well as to track changes in quality over time. However, there is potential for missing or unknown data as well as the possibility that clinics may code diagnoses differently. Backus et al. 17 described methods used to abstract the electronic data that comprises the HCV CCR. They recognize that patients may be included in the registry if they are not chronically infected with HCV. To account for this issue, we were able to use the available laboratory tests for HCV RNA to ensure that everyone in this study had evidence of chronic HCV infection. Additional validity concerns are partially alleviated by the fact that laboratory values and ICD-9 codes from the CCR data are highly accurate. 13,26 We have shown in a previous study that codes for HAV vaccination have a positive predictive value of 94.6% and negative predictive value of 94.0% and HBV vaccinations have a positive predictive value of 100% and negative predictive value of 94.0%. 13 We were not able to determine whether a test or vaccine was ordered but not adhered to by the patient or was received outside the VA medical system. This would underestimate the true QM. In addition, patients in this study are mostly men who are sicker and of lower socioeconomic status than the general population. 27,28 Thus, the results from this study may not be generalizable to women or patients using other health care systems. However, whereas the electronic health records in the VA allow a comprehensive evaluation of HCV care, similar quality assessment efforts in non-va health care systems may be difficult without an advanced information system. Despite these limitations, we feel that this study was an efficient way of examining QMs for HCV care. In conclusion, this is the largest study to date to examine vaccination and QM rates as well as the first to examine patient and facility predictors of those rates in patients with chronic HCV. It highlights that there is still a substantial proportion of chronically infected patients with HCV that are susceptible to HBV and HAV infection. Some solutions include a targeted quality improvement program, such as improving referral to a specialist, creation of dedicated HCV care clinics, and improving specialists adherence to QMs, including vaccination for HAV and HBV in susceptible patients. Consideration should also be given to the chronic disease management framework to improve health care delivery in these chronically ill patients. Guided by this model, changes in delivery system design (such as an automatic reminder for vaccination that occurs routinely) could be developed. Future research in this area should focus on developing and testing some of these interventions to see if they improve adherence to the vaccination QM in patients with chronic HCV. Acknowledgment: We thank Laura Petersen, M.D., M.P.H., for contributing the facility measures developed by her and her study team. References 1. Armstrong GL, Wasley A, Simard EP, McQuillan GM, Kuhnert WL, Alter MJ. The prevalence of hepatitis C virus infection in the United States, 1999 through Ann Intern Med 2006;144: Austin GE, Jensen B, Leete J, De L Aune W, Bhatnagar J, Racine M, et al. Prevalence of hepatitis C virus seropositivity among hospitalized US veterans. Am J Med Sci 2000;319: Briggs M, Baker C, Hall R, Gaziano J, Gagnon D, Bzowej N, et al. Prevalence and risk factors for hepatitis C virus infection at an urban Veterans Administration medical center. HEPATOLOGY 2001;34: Cheung RC. Epidemiology of hepatitis C virus infection in American veterans. Am J Gastroenterol 2000;95: Dominitz JA, Boyko EJ, Koepsell TD, Heagerty PJ, Maynard C, Sporleder JL, et al. Elevated prevalence of hepatitis C infection in users of United States veterans medical centers. HEPATOLOGY 2005;41: El-Serag HB. Risk factors for the rising rates of primary liver cancer in the United States. Arch Intern Med 2000;160: Public health control of hepatitis A: memorandum from a WHO meeting. Bull World Health Organ 1995;73: National Institutes of Health. National Institutes of Health Consensus Development Conference Statement: management of hepatitis C: 2002 June 10-12, HEPATOLOGY 2002;36(Suppl):S3-S Department of Veterans Affairs. Hepatitis C testing and prevention counseling guidelines for VA health care practitioners. Available at: Accessed September 25, Fiore AE, Wasley A, Bell BP. Prevention of hepatitis A through active or passive immunization: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 2006;55: Ghany MG, Strader DB, Thomas DL, Seeff LB. Diagnosis, management, and treatment of hepatitis C: an update. HEPATOLOGY 2009;49:

11 52 KRAMER ET AL. HEPATOLOGY, January Medicare s Physician Quality Reporting Initiative. Available at: www cms hhs gov/pqri/downloads/2010_pqri_measureslist_ pdf. Accessed November 13, Hachem CY, Kramer JR, Kanwal F, El-Serag HB. Hepatitis vaccination in patients with hepatitis C: practice and validation of codes at a large Veterans Administration Medical Centre. Aliment Pharmacol Ther 2008;28: Hernandez B, Hasson NK, Cheung R. Hepatitis C performance measure on hepatitis A and B vaccination: missed opportunities? Am J Gastroenterol 2009;104: Shim M, Khaykis I, Park J, Bini EJ. Susceptibility to hepatitis A in patients with chronic liver disease due to hepatitis C virus infection: missed opportunities for vaccination. HEPATOLOGY 2005;42: Wong V, Wreghitt TG, Alexander GJ. Prospective study of hepatitis B vaccination in patients with chronic hepatitis C. BMJ 1996;312: Backus LI, Gavrilov S, Loomis TP, Halloran JP, Phillips BR, Belperio PS, et al. Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009;16: Keeffe EB, Williams J, the Hepatitis C Work Group. Hepatitis C physician performance measure set. Available at: ama1/pub/upload/mm/370/hepcws pdf. Accessed March 1, Petersen LA, Pietz K, Woodard LD, Byrne M. Comparison of the predictive validity of diagnosis-based risk adjusters for clinical outcomes. Med Care 2005;43: Byrne MM, Daw CN, Nelson HA, Urech TH, Pietz K, Petersen LA. Method to develop health care peer groups for quality and financial comparisons across hospitals. Health Serv Res 2009;44: Petersen LA, Pietz K, Woodard LD, Byrne M. Comparison of the predictive validity of diagnosis-based risk adjusters for clinical outcomes. Med Care 2005;43: Brookhart MA, Solomon DH, Wang P, Glynn RJ, Avorn J, Schneeweiss S. Explained variation in a model of therapeutic decision making is partitioned across patient, physician, and clinic factors. J Clin Epidemiol 2006;59: Efron B. Regression and ANOVA with zero-one data measures of residual variation. J Am Stat Assoc 1978;73: Ghali WA, Ash AS, Hall RE, Moskowitz MA. Statewide quality improvement initiatives and mortality after cardiac surgery. JAMA 1997;277: Jencks SF, Cuerdon T, Burwen DR, Fleming B, Houck PM, Kussmaul AE, et al. Quality of medical care delivered to Medicare beneficiaries: a profile at state and national levels. JAMA 2000;284: Kramer JR, Davila JA, Miller ED, Richardson P, Giordano TP, El-Serag HB. The validity of viral hepatitis and chronic liver disease diagnoses in Veterans Affairs administrative databases. Aliment Pharmacol Ther 2008;27: Wilson NJ, Kizer KW. The VA health care system: an unrecognized national safety net. Health Aff (Millwood) 1997;16: Yu W, Ravelo A, Wagner TH, Phibbs CS, Bhandari A, Chen S, et al. Prevalence and costs of chronic conditions in the VA health care system. Med Care Res Rev 2003;60(Suppl):146S-167S.