Abstract
Background
Hepatitis C virus (HCV) testing guidance issued by the Centers for Disease Control and Prevention in 1998 recommends HCV antibody (anti-HCV) testing for persons with specified risk factors. The purpose of this study was to determine the prevalence and predictors of anti-HCV positivity among primary care outpatients and estimate the proportion of unidentified anti-HCV-positive (anti-HCV+) persons using risk-based testing.
Methods
We analyzed electronic medical record data from a 4-site retrospective study. Patients were aged ≥18 years, utilized ≥1 outpatient primary care service(s) between 2005 and 2010, and had no documented evidence of prior HCV diagnosis. Among persons tested for anti-HCV, we fit a multilevel logistic regression model to identify patient-level independent predictors of anti-HCV positivity. We estimated the proportion of unidentified anti-HCV+ persons by using multiple imputation to assign anti-HCV results to untested patients.
Results
We observed 209 076 patients for a median of 5 months (interquartile range, 1–23 months). Among 17 464 (8.4%) patients who were tested for anti-HCV, 6.4% (n = 1115) were positive. We identified history of injection drug use (adjusted odds ratio [95% confidence interval], 6.3 [5.2–7.6]), 1945–1965 birth cohort (4.4 [3.8–5.1]), and elevated alanine aminotransferase levels (4.8 [4.2–5.6]) as independently associated with anti-HCV positivity. We estimated that 81.5% (n = 4890/6005) of anti-HCV+ patients were unidentified using risk-based testing.
Conclusions
In these outpatient primary care settings, risk-based testing may have missed 4 of 5 newly enrolled patients who are anti-HCV+. Without knowing their status, unidentified anti-HCV+ persons cannot receive further clinical evaluation or antiviral treatment, and are unlikely to benefit from secondary prevention recommendations to limit disease progression and mortality.
Keywords: hepatitis C virus (HCV), anti-HCV prevalence, predictors of HCV positivity, outpatient primary care, HCV testing
Hepatitis C virus (HCV) infection is an urgent health problem, with an estimated 4 million persons ever infected and approximately 3 million chronically infected in the United States [1,2].It is estimated that 45%–60% of adults with HCV infection will develop cirrhosis over the next 2–5 decades [3, 4]. In 2010, HCV infection was the underlying or contributing cause of >16 500 deaths in the United States [5].Advances in antiviral therapy have resulted in the potential for dramatic reductions in morbidity and mortality among persons treated for HCV infection [6–9], but care and treatment can only be provided to those infected persons who are tested and identified.
In 1998, Centers for Disease Control and Prevention (CDC) recommended routine HCV antibody (anti-HCV) testing for persons with specified risk factors and medical indications including history of injection drug use (IDU), persistently elevated alanine aminotransferase (ALT) levels, blood transfusion or organ transplant before 1992, clotting factor concentrates before 1987, and long-term hemodialysis [10]; in 1999, the US Public Health Service and the Infectious Diseases Society of America jointly recommended HCV testing for all human immunodeficiency virus (HIV)-infected adults [11]. Several noninterventional studies have since examined the effectiveness of risk-based testing among persons receiving care in clinical settings or persons with access to healthcare; the studies have compared the overall observed anti-HCV prevalence—defined as the number of identified persons divided by the total population of patients (tested and not)— to the expected prevalence as determined by population surveys or modeling [12–15]. Among patients enrolled in an urban managed care organization, Roblin et al determined that the overall observed anti-HCV prevalence was 0.2% [12]; using a population of patients enrolled in 4 private healthcare organizations, Spradling et al found an overall observed prevalence of approximately 0.9% [13]; other researchers have found higher observed prevalence rates ranging from 2.5% to 4.6% among predominantly African American and Hispanic populations receiving routine care in urban primary care settings [14, 15]. Unanimously, the overall observed anti-HCV prevalence in these settings has been described as lower than expected and the proportion of unidentified anti-HCV positive persons has been estimated as ranging from 40% to 85% [12–16].
In 2012, CDC augmented the risk factor– and medical indication–based HCV screening strategies by issuing a recommendation to test all persons born during 1945–1965 without prior risk ascertainment [17]. Persons in this cohort account for an estimated three-quarters of all HCV infections and HCV-associated deaths in the United States [18, 19]. In 2013, the US Preventive Services Task Force also revised its recommendations to include this birth cohort based on their review of the available evidence [20].
This study sought to determine anti-HCV prevalence and factors associated with anti-HCV positivity in a large geographically and demographically diverse cohort of adult primary care outpatients with no previous diagnosis of HCV. We also aimed to estimate the proportion of unidentified anti-HCV-positive (anti-HCV+) persons within this cohort.
METHODS
Study Design
We conducted a cross-sectional analysis using electronic medical record (EMR) data of patients from the Birth Cohort Evaluation to Advance Screening and Testing for HCV (BEST-C), a multisite retrospective cohort study.
Study Setting and Participants
Participants in BEST-C included all newly enrolled patients aged ≥18 years with no previous diagnosis of anti-HCV who utilized at least 1 primary care outpatient service in 4 large healthcare centers (sites): Mount Sinai Medical Center (MSMC); University of Alabama at Birmingham (UAB); University of Texas, Houston (UTH); and Henry Ford Hospital System (HFH). The study spans 6 years from January 2005 to December 2010 (MSMC, UAB, UTH) or from March 2005 to March 2011 (HFH). Participants with missing values for date of first encounter were excluded.
Only new patients with a first-time encounter through primary care were included in this study based on the understanding that health screening is most likely to occur for new patients [16]; patients with documentation of HCV diagnosis at the time of first encounter were excluded. Data on patient demographics, clinical characteristics, and anti-HCV testing were collected at each patient’s first visit and at successive encounters during the study period to create an encounter-level dataset. Encounter-level data were summarized to create a patient-level dataset. The current analysis is based on 209 076 patients. We calculated the period of observation (in months) as the date of last visit minus date of first visit; patients with a single visit were coded as observed for 1 month.
Outcome Variables
The main outcome measure for this analysis was anti-HCV positivity. We defined anti-HCV positivity as a positive test result for antibodies to HCV by enzyme-linked immunoassay at any visit during the study period.
Independent Variables
Independent variables previously associated with anti-HCV positivity were selected for this analysis [12, 14, 21]. Demographic variables examined included birth year, race/ethnicity, sex, marital status, and annual census tract–level income. Census tract income was used as a proxy for household income, which was unavailable. Birth year was dichotomized as born during 1945–1965 or born outside the 1945–1965 birth cohort. Race/ethnicity was categorized as white, black, Hispanic, Asian, or other. Sex was classified as male or female. Marital status was categorized as married, divorced/separated/widowed, or never married. Income was categorized as <$30 000, $30 000–$49 999, $50 000–$69 000, $70 000–$99 000, and ≥$100 000. We also examined the following clinical variables: history of IDU, elevated ALT, hemophilia, HIV infection, and total number of patient encounters with health system during the study period. History of IDU was defined as having an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code for IDU (292, 304, 305, 969.7, 970.8, 965.01) documented in the EMR at any time during the study period. This search strategy was augmented by text query of the EMR for evidence of IDU. Patients were coded as having a history of elevated ALT if a laboratory record for ALT test with abnormal result was associated with the EMR at any visit during the study period. Cutoff levels used to establish elevated ALT were sex- and site-specific. Hemophilia (ICD-9-CM 286.52, 286.5) and HIV infection (ICD-9-CM V08, 042) were similarly identified. Identification of HIV infection was enhanced by search of text notes in the EMR. Data regarding other known covariates of anti-HCV positivity such as hemodialysis and blood transfusion before 1992 were not included in this analysis due to lack of consistency in data collection across sites. History of IDU, hemophilia, HIV infection, and elevated ALT were coded as yes or no/unknown.
Statistical Analysis
Multiple Imputation of Missing Data
Overall, about 51.0% of all patients had missing data for at least 1 demographic variable. To maintain statistical power and to avoid potential bias in statistical estimates [22], we used IVE-ware version 0.1 to replace missing values by multiple imputation using the method of sequential regression multivariate imputation [23]. Imputed datasets were individually analyzed with standard software and summarized per the Rubin method with appropriate adjustment for degrees of freedom [24] (Supplementary Appendix).
Statistical Estimates
Statistical significance for all tests was set at a 2-tailed P value of .05. We calculated proportions to describe the characteristics of the overall study population as well as the subpopulation of patients who were tested for anti-HCV at any time during the study period. Overall observed prevalence [13, 14] was estimated by dividing the total number of anti-HCV+ persons by the total number of all participants (tested or not). This approach assumes that because testing is driven by risk factors and medical conditions, patients who are not tested are more likely to be negative for anti-HCV. Expected prevalence was estimated by extending multiple imputation to anti-HCV status for patients who were not tested (Supplementary Appendix). Specifically, imputed anti-HCV values were assigned to each patient who was not tested, fully conditional on all data observed for that patient. The proportion of unidentified anti-HCV+ persons was subsequently estimated as the difference between expected prevalence and observed prevalence divided by expected prevalence in the study population. We performed sensitivity analyses to assess effect of duration of follow-up on this estimate.
Among patients who were tested for anti-HCV during the 6-year study period, we calculated anti-HCV positivity by dividing the total number of anti-HCV+ patients by the total number of tested patients. To account for clustering of patients within study sites, we used univariate generalized linear mixed models to test for differences in anti-HCV positivity by patient characteristics [25]. We also fit a prespecified multilevel multiple logistic regression model adjusting for the random effect of site to identify independent correlates of anti-HCV positivity among patients tested. The dependent variable was anti-HCV positivity. Patient-level independent variables included the following: birth year, sex, race/ethnicity, marital status, elevated ALT, IDU, hemophilia, HIV status, and total number of visits. Study site was modeled as a second-level random intercept. Census tract income was not included in this model as the variable was not measured at either the site level or the patient level. Sensitivity analysis was performed by comparing estimates from the multiply imputed data to results of complete case analysis. Unless otherwise indicated, we used SU-DAAN (version 10.0.1) and SAS (version 9.3) software to analyze data.
RESULTS
Characteristics of Study Population
A total of 209 076 patients were included in the analysis. The median length of observation was 5 months with an interquartile range of 1 to 23. A summary of the demographic and clinical characteristics of patients is presented in Table 1.
Table 1.
Characteristics of Study Population
| Characteristic | All Patients, No. (%) | Patients Tested, No. (%) |
|---|---|---|
| Overall | 209 076 (100) | 17 464 (100) |
| Age, y, median (IQR) | 37 (28–50) | 37 (28–49) |
| 1945–1965 birth cohort | 70 718 (33.8) | 5989 (34.3) |
| Female sexa | 137 695 (65.6) | 9706 (45.6) |
| Race/ethnicitya | ||
| White | 95 992 (45.9) | 6554 (37.5) |
| Black | 58 437 (27.9) | 6757 (38.7) |
| Hispanic | 14 526 (6.9) | 1065 (6.1) |
| Asian | 9492 (4.5) | 909 (5.2) |
| Other | 30 630 (14.7) | 2179 (12.5) |
| Marital statusa | ||
| Married | 94 673 (45.3) | 6434 (36.8) |
| Widowed/divorced | 15 866 (7.6) | 1582 (9.1) |
| Never married | 98 537 (47.1) | 9448 (54.1) |
| Income categorya | ||
| <$30 000 | 34 561 (16.5) | 3671 (21.0) |
| $30 000–$49 999 | 58 227 (27.8) | 5209 (29.8) |
| $50 000–$69 000 | 51 440 (18.7) | 4215 (24.1) |
| $70 000–$99 999 | 39 145 (18.7) | 2800 (16.03) |
| ≥$100 000 | 25 704 (12.3) | 1569 (9.0) |
| Injection drug use | 2992 (1.4) | 1007 (5.8) |
| Hemophilia | 1241 (0.6) | 299 (1.7) |
| Elevated ALT | 12 574 (6.0) | 4618 (26.4) |
| HIV infected | 1240 (0.6) | 767 (4.4) |
Abbreviations: ALT, alanine aminotransferase; HIV, human immunodeficiency virus; IQR, interquartile range.
Missing values for sex, race/ethnicity, marital status, and income were multiply imputed.
Observed and Expected Prevalence Estimates
The overall observed prevalence (ie, the rate of anti-HCV identification) among all patients was 0.53% (n = 1115/209 076). Using multiply imputed anti-HCV values for patients who were not tested, we estimated the corresponding expected prevalence to be 2.87% (n = 6005/209 076), with a range of 0.84% to 4.34% across the 4 sites. We further estimated that 81.5% (n = 4890/6005) of anti-HCV+ persons were not identified using risk-based testing strategy (Table 2). Upon restricting the analytic data to new patients who had been enrolled in the health system for at least 12 months (median follow-up = 30 months), the proportion of unidentified anti-HCV patients was estimated at 80%; a restricted analysis of new patients who were enrolled for at least 24 months (median follow-up = 40 months) yielded a similar result (79%). Among patients born between 1945 and 1965, the estimated proportion of unidentified anti-HCV+ persons was 76% (ie, observed prevalence = 1.2%; expected prevalence = 4.9%).
Table 2.
Observed and Expected Estimates of Hepatitis C Virus Antibody Prevalence Among Previously Undiagnosed Patients (N = 209 076)
| Estimate | All Sites | HFH | MSMC | UAB | UTH |
|---|---|---|---|---|---|
| Observed prevalence estimate, % (No.) | 0.53 (1115) | 0.6 (601) | 0.85 (315) | 0.18 (72) | 0.39 (127) |
| Expected prevalence estimate, % (No.) | 2.87 (6005) | 2.84 (2830) | 3.82 (1423) | 0.83 (328) | 4.33 (1424) |
| Estimated unidentified persons, % (No.) | 81.5 (4890) | 78.7 (2229) | 77.9 (1109) | 78.0 (256) | 91.1 (1297) |
Abbreviations: HFH, Henry Ford Hospital; MSMC, Mt Sinai Medical Center; UAB, University of Alabama at Birmingham; UTH, University of Texas, Houston.
Positivity Among Patients Tested
Among patients who were tested for anti-HCV, 6.4% (n = 1115/17 464) were positive (Table 3). About 75% of all anti-HCV+ persons were born during 1945–1965. Anti-HCV positivity was significantly higher in patients born from 1945 to 1965 (13.8%) compared with the referent group of those born before 1945 or after 1965 (2.5%); blacks (12.1%) or Hispanics (8.5%) relative to whites (5.0%); widowed/divorced/separated (10.4%) or never married (6.7%) compared with married (5.0%); and males (8.8%) vs females (4.4%) (Table 3). Anti-HCV positivity was also greater in patients with a history of elevated ALT (14.9%), IDU (35.8%), hemophilia (10.7%), or HIV (17.3%).
Table 3.
Probability of Testing Hepatitis C Virus Antibody Positive, by Demographic and Clinical Characteristics
| Characteristic | Patients Tested, No. | Patients Positive, No. (%) | P Value* |
|---|---|---|---|
| Overall | 17 464 | 1115 (6.4) | … |
| 1945–1965 birth cohort | 286 | ||
| No | 11 475 | 286 (2.5) | … |
| Yes | 5989 | 829 (13.8) | <.0001 |
| Sexa | |||
| Female | 9706 | 429 (4.4) | … |
| Male | 7758 | 686 (8.8) | <.0001 |
| Race/ethnicitya | |||
| White | 6554 | 330 (5.0) | … |
| Black | 6757 | 574 (8.5) | <.0001 |
| Hispanic | 1065 | 129 (12.1) | .0001 |
| Asian | 909 | 16 (1.7) | <.0001 |
| Other | 2179 | 67 (3.1) | .0119 |
| Marital statusa | |||
| Married | 6434 | 319 (5.0) | … |
| Widowed/divorced | 1582 | 164 (10.4) | <.0001 |
| Never married | 9448 | 632 (6.7) | .0178 |
| Income categorya | |||
| <$30 000 | 3671 | 366 (10.0) | <.0001 |
| $30 000–$49 999 | 5209 | 361 (6.9) | <.0001 |
| $50 000–$69 000 | 4215 | 225 (5.3) | .0014 |
| $70 000–$99 999 | 2800 | 106 (3.8) | .5743 |
| ≥$100 000 | 1569 | 57 (3.6) | … |
| Elevated ALT | |||
| No/unknown | 12 846 | 425 (3.3) | … |
| Yes | 4618 | 691 (14.9) | .0001 |
| Injection drug use | |||
| No/unknown | 16 457 | 754 (4.6) | … |
| Yes | 1007 | 365 (35.8) | <.0001 |
| Hemophilia | |||
| No/unknown | 17 165 | 1083 (6.3) | … |
| Yes | 299 | 32 (10.7) | .0318 |
| HIV infected | |||
| No/unknown | 16 697 | 982 (5.9) | … |
| Yes | 767 | 133 (17.3) | .0045 |
Abbreviations: ALT, alanine aminotransferase; HIV, human immunodeficiency virus.
Missing values for sex, race/ethnicity, marital status, and income were multiply imputed.
P values obtained from univariate generalized linear mixed model incorporating random effect of site.
Predictors of Anti-HCV Positivity Among Patients Tested
Following multivariate adjustment in a multilevel logistic regression model, we identified IDU (adjusted odds ratio [AOR], 6.3 [95% confidence interval {CI}, 5.2–7.6]), 1945–1965 birth cohort (AOR, 4.4 [95% CI, 3.8–5.1]), elevated ALT (AOR, 4.8 [95% CI, 4.2–5.6]), black race (AOR, 1.9 [95% CI, 1.6–2.2]), Hispanic ethnicity (AOR, 1.5 [95% CI, 1.2–2.0]), widowed/divorced/separated (AOR, 1.5 [95% CI, 1.2–2.0]), never married (AOR, 1.4 [95% CI, 1.2–1.6]), and male sex (AOR, 1.3 [95% CI, 1.2–1.6]) as significant correlates of anti-HCV positivity (Table 4). Hemophilia and HIV-positive status were not significantly associated with anti-HCV positivity. Sensitivity analysis using complete cases indicated that point estimates were comparable to results from the imputed data, imputation restored statistical efficiency and produced estimates with better precision (Table 5).
Table 4.
Model-Adjusted Odds Ratios for Hepatitis C Virus Antibody Positivity
| Characteristic (n = 17 464) | Adjusteda OR (95% CI) | P Value |
|---|---|---|
| 1945–1965 birth cohort | ||
| No | 1.0 | … |
| Yes | 4.4 (3.8–5.1) | <.0001 |
| Sexb | ||
| Female | 1.0 | … |
| Male | 1.4 (1.2–1.6) | <.0001 |
| Race/ethnicityb | ||
| White | 1.0 | … |
| Black | 1.9 (1.6–2.2) | <.0001 |
| Hispanic | 1.5 (1.2–2.0) | .0022 |
| Asian | 0.6 (.4–1.1) | .0874 |
| Other | 1.1 (.8–1.5) | .4142 |
| Marital statusb | ||
| Married | 1.0 | … |
| Widowed/divorced | 1.5 (1.2–2.0) | .0019 |
| Never married | 1.4 (1.2–1.6) | .0003 |
| Elevated ALT | ||
| No/unknown | 1.0 | … |
| Yes | 4.8 (4.2–5.6) | <.0001 |
| Injection drug use | ||
| No/unknown | 1.0 | … |
| Yes | 6.3 (5.2–7.6) | <.0001 |
| Hemophilia | ||
| No/unknown | 1.0 | … |
| Yes | 1.1 (.7–1.7) | .6207 |
| HIV infected | ||
| No/unknown | 1.0 | … |
| Yes | 1.0 (.8–1.3) | .7571 |
Abbreviations: ALT, alanine aminotransferase; CI, confidence interval; HIV, human immunodeficiency virus; OR, odds ratio.
Multilevel model adjusted for random effect of site and the following individual-level fixed effects: birth year, sex, race, marital status, injection drug use, elevated ALT, hemophilia, HIV status, and total number of encounters with healthcare system.
Missing values for sex, race/ethnicity, and marital status were multiply imputed.
Table 5.
Comparison Between Estimates From Multiply Imputed and Complete Case Analytic Models for Hepatitis C Virus Antibody Positivity
| Characteristic | Adjusteda OR (95% CI) | |
|---|---|---|
| Multiply Imputedb (n = 17 464) |
Complete Case (n = 13 689) |
|
| 1945–1965 birth cohort | ||
| No | 1.0 | 1.0 |
| Yes | 4.4 (3.8–5.1)* | 4.2 (3.2–5.5)* |
| Sex | ||
| Female | 1.0 | 1.0 |
| Male | 1.4 (1.2–1.6)* | 1.3 (1.0–1.6) |
| Race/ethnicity | ||
| White | 1.0 | 1.0 |
| Black | 1.9 (1.6–2.2)* | 1.8 (1.5–2.2)* |
| Hispanic | 1.5 (1.2–2.0)** | 1.5 (1.1 –2.2)*** |
| Asian | 0.6 (.4–1.1) | 0.6 (.3–1.2) |
| Other | 1.1 (.8–1.5) | 1.2 (.8–1.7) |
| Marital status | ||
| Married | 1.0 | 1.0 |
| Divorced/widowed | 1.5 (1.2–2.0)** | 1.5 (1.1 –2.0)*** |
| Never married | 1.4 (1.2–1.6)* | 1.4(1.1–1.7)*** |
| Elevated ALT | ||
| No/unknown | 1.0 | 1.0 |
| Yes | 4.8 (4.2–5.6)* | 4.8 (3.7–6.2)* |
| Injection drug use | ||
| No/unknown | 1.0 | 1.0 |
| Yes | 6.3 (5.2–7.6)* | 6.4 (4.6–8.9)* |
| Hemophilia | ||
| No/unknown | 1.0 | 1.0 |
| Yes | 1.1 (.7–1.7) | 1.0 (.5–2.2) |
| HIV infected | ||
| No/unknown | 1.0 | 1.0 |
| Yes | 1.0 (.8–1.3) | 1.0 (.6–1.6) |
Abbreviations: ALT, alanine aminotransferase; CI, confidence interval; HIV, human immunodeficiency virus; OR, odds ratio.
Adjusted for random effect of site and the following fixed effects: birth year, sex, race, marital status, injection drug use, elevated ALT, hemophilia, HIV status, and total number of encounters with healthcare system.
Missing values for sex, race/ethnicity, and marital status were multiply imputed.
P < .001.
P values were between <.001 and <.01.
P values were between <.01 and <.05.
DISCUSSION
This is a multisite study in the United States examining anti-HCV positivity in a large cohort of newly enrolled, previously undiagnosed adult primary care outpatients from different regions of the country. We found that over a 6-year period, 6.4% of tested patients were anti-HCV+. However, as the majority of patients were not tested, the observed prevalence was just above 0.5%. We estimated that the expected anti-HCV prevalence in this population could be as high as 2.9%, indicating that during the study period, >81% of anti-HCV+ persons may have been undetected.
The findings from this study are supported by previous research. Anti-HCV positivity among patients tested and the observed prevalence are comparable to results from Spradling et al, who used a similarly diverse population of patients with access to care: estimates for anti-HCV prevalence among patients tested and observed are 5.5% and 0.7%, respectively, when the data are restricted to patients with no history of HCV diagnosis [13]. Roblin et al also reported an anti-HCV positivity of 5.1% among patients tested and an overall observed prevalence of 0.2%, although the proportion of patients tested was considerably lower [12]. Southern et al [14] found a much higher prevalence of anti-HCV among patients tested (11.5%) and overall observed prevalence (4.6%), likely reflecting a higher local prevalence of HCV and HCV risk factors [26, 27] and a higher-than-average proportion of patients tested for anti-HCV of nearly 40% [14]. Our expected prevalence estimate can be reasonably compared to an estimate of 2.0% anti-HCV prevalence among US adults aged 20–70 years [18], accounting for the fact that African Americans, a high-prevalence group, are substantially overrepresented in the outpatient primary care cohort used for the current study. Moreover, our finding that testing based on risks or medical indications in the clinical setting underestimates anti-HCV prevalence by 81% is within the range of earlier findings [12–14]. Notably, our data further suggest that anti-HCV prevalence may vary regionally in the United States (Table 2). This observation is supported by prevalence estimates from prior studies [14, 26, 28, 29].
Our analysis also indicated that patients’ birth year, history of IDU, and elevated ALT were significant correlates of anti-HCV positivity, as were black race, Hispanic ethnicity, single marital status, and male sex. History of IDU and elevated ALT as predictors of anti-HCV positivity are consistent with the data supporting risk-based testing recommendations [12, 14, 15, 30]. Race, ethnicity, sex, and marital status have also previously been associated with anti-HCV positivity or current HCV infection [1, 13, 14]. Nevertheless, the finding that 74% of the identified anti-HCV+ patients were born from 1945 to 1965 provides additional data in support of recommendations by the CDC and the US Preventive Task Force [17, 20] for HCV testing in this birth cohort and confirms previous findings of the burden of prevalence in this cohort [13, 14, 19, 21].
This study has several strengths. The analysis of a large, demographically diverse, multisite, multiyear cohort of outpatient primary care patients representing different geographic regions of the country increases the potential generalizability of the findings. Also, the use of multilevel modeling to adjust for correlation between patients within sites (healthcare centers) ensures that standard error estimates are conservative. Moreover, the use of multiple imputation to replace missing data restored statistical power and improved precision of estimates.
However, there are also some important limitations. First, blood transfusion before 1992, an important determinant of anti-HCV positivity among older adults, was not included in this analysis due to a lack of consistent data collection across sites. Second, it is likely that risk factors such as IDU were underreported [31–35]; nondisclosure of relevant patient risk factors could lead to confounding of estimates [36]. Nonetheless, we determined that the proportions of patients identified as having history of IDU, hemophilia, HIV positivity, or elevated ALT were comparable to estimates reported in previous studies [12, 14, 16]. Third, cross-sectional analysis of this dataset limits our ability to assess temporal associations between independent risk factors and anti-HCV positivity; however, cross-sectional analysis was appropriate for other important objectives in this study such as estimates of anti-HCV prevalence and proportion of unidentified anti-HCV+ persons. Fourth, it is likely that retrospective review of EMR did not adequately identify patients diagnosed with HCV outside participating health systems prior to their first encounter; patients with a previous HCV diagnosis who were misclassified at the first encounter would have been included in the study and may not have received additional HCV testing during the study. In addition, we were unable to identify patients diagnosed outside participating health systems during the study, and our estimate of unidentified anti-HCV+ persons may be inflated as a result. Finally, in using multiple imputation to assign plausible values to missing data (sex, race/ethnicity, marital status, expected anti-HCV prevalence), we assumed that missing values were missing at random (eg, missing data values for sex or anti-HCV status are related conditionally to observed patient characteristics) [22, 23]. Despite our efforts to make this assumption more plausible [25, 37] by including a large number of relevant predictors in the imputation models, the assumption may not be fully met because there is a good likelihood that not all patient variables were observed. Still, others have noted that missing data originating from medical research are often related to observed patient data and that the missing-at-random assumption may be appropriate [22,38].
Despite these limitations, our findings demonstrate that testing based on risk and medical indications alone failed to identify four-fifths of previously undiagnosed adults with past exposure to HCV. This may be due in part to the difficulty in capturing complete patient risk history (eg, IDU) in EMRs to support the implementation of comprehensive risk–based HCV testing algorithms. HCV-infected persons who are not aware of their status cannot receive further clinical evaluation or antiviral treatment, and are unlikely to benefit from preventive services or secondary prevention recommendations (eg, reduction in alcohol use and other lifestyle changes) aimed at limiting disease progression and reducing liver-related morbidity and mortality [39, 40]. In the routine clinical environment, recent CDC and US Preventive Task Force recommendations to test patients born during 1945–1965 for HCV without the need for prior ascertainment of risk factors should be implemented [17, 20]. With the alignment of the risk-based and birth cohort testing recommendations, it is expected that identification of infected persons who were previously undiagnosed will increase, leading to higher rates of linkage to care and treatment as appropriate, or to programs that support linkage to care and treatment adherence, which would further result in reduced morbidity and mortality associated with HCV.
Supplementary Material
Acknowledgments
Financial support. This study was supported by the CDC.
Footnotes
Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online (http://cid.oxfordjournals.org). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.
Notes
Disclaimer. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC).
Potential conflicts of interest. K. A. B. reports receiving funds from Gilead, Merck, Janssen, BMS, Novartis, Vertex, Medscape, Hepatitis C Virus Viewpoint, Hyperion, CDC Foundation, and the Chronic Liver Disease Foundation. D. B. R. reports receiving funds from the CDC Foundation, Gilead Sciences and Centers for Disease Control and Prevention, Division of Viral Hepatitis. All other authors report no potential conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1.Denniston MM, Jiles RB, Drobeniuc J, et al. Chronic hepatitis C virus infection in the United States, National Health and Nutrition Examination Survey 2003 to 2010. Ann Intern Med. 2014;160:293–300. doi: 10.7326/M13-1133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chak E, Talal AH, Sherman KE, Schiff ER, Saab S. Hepatitis C virus infection in USA: an estimate of true prevalence. Liver Int. 2011;31:1090–101. doi: 10.1111/j.1478-3231.2011.02494.x. [DOI] [PubMed] [Google Scholar]
- 3.Rein DB, Wittenborn JS, Weinbaum CM, Sabin M, Smith BD, Lesesne SB. Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis. 2011;43:66–72. doi: 10.1016/j.dld.2010.05.006. [DOI] [PubMed] [Google Scholar]
- 4.Davis GL, Alter MJ, El-Serag H, Poynard T, Jennings LW. Aging of hepatitis C virus (HCV)-infected persons in the United States: a multiple cohort model of HCV prevalence and disease progression. Gastroenterology. 2010;138:513–21. 521.e1–6. doi: 10.1053/j.gastro.2009.09.067. [DOI] [PubMed] [Google Scholar]
- 5.Centers for Disease Control and Prevention. Viral hepatitis surveillance, United States. 2011 Available at: http://www.cdc.gov/hepatitis/Statistics/2011Surveillance/PDFs/2011HepSurveillanceRpt.pdf. Accessed 15 September 2014.
- 6.Poynard T. Hepatitis C: natural history, biology, treatment monitoring [in French] Pathol Biol (Paris) 1999;47:911–6. [PubMed] [Google Scholar]
- 7.van der Meer AJ, Veldt BJ, Feld JJ, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA. 2012;308:2584–93. doi: 10.1001/jama.2012.144878. [DOI] [PubMed] [Google Scholar]
- 8.Lawitz E, Mangia A, Wyles D, et al. Sofosbuvir for previously untreated chronic hepatitis C infection. N Engl J Med. 2013;368:1878–87. doi: 10.1056/NEJMoa1214853. [DOI] [PubMed] [Google Scholar]
- 9.Vaidya A, Perry CM. Simeprevir: first global approval. Drugs. 2013;73:2093–106. doi: 10.1007/s40265-013-0153-9. [DOI] [PubMed] [Google Scholar]
- 10.Centers for Disease Control and Prevention. Recommendations for prevention and control of hepatitis C virus (HCV) infection and HCV-related chronic disease. MMWR Recomm Rep. 1998;47(RR-19):1–39. [PubMed] [Google Scholar]
- 11.1999 USPHS/IDSA guidelines for the prevention of opportunistic infections in persons infected with human immunodeficiency virus. U.S. Public Health Service (USPHS) and Infectious Diseases Society of America (IDSA) MMWR Recomm Rep. 1999;48(RR-10):1–59. 61–6. [PubMed] [Google Scholar]
- 12.Roblin DW, Smith BD, Weinbaum CM, Sabin ME. HCV screening practices and prevalence in an MCO, 2000–2007. Am J Manag Care. 2011;17:548–55. [PubMed] [Google Scholar]
- 13.Spradling PR, Rupp L, Moorman AC, et al. Hepatitis B and C virus infection among 1.2 million persons with access to care: factors associated with testing and infection prevalence. Clin Infect Dis. 2012;55:1047–55. doi: 10.1093/cid/cis616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Southern WN, Drainoni ML, Smith BD, et al. Hepatitis C testing practices and prevalence in a high-risk urban ambulatory care setting. J Viral Hepat. 2011;18:474–81. doi: 10.1111/j.1365-2893.2010.01327.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Trooskin SB, Navarro VJ, Winn RJ, et al. Hepatitis C risk assessment, testing and referral for treatment in urban primary care: role of race and ethnicity. World J Gastroenterol. 2007;13:1074–8. doi: 10.3748/wjg.v13.i7.1074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Almario CV, Vega M, Trooskin SB, Navarro VJ. Examining hepatitis C virus testing practices in primary care clinics. J Viral Hepat. 2012;19:e163–9. doi: 10.1111/j.1365-2893.2011.01539.x. [DOI] [PubMed] [Google Scholar]
- 17.Smith BD, Morgan RL, Beckett GA, et al. Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945–1965. MMWR Recomm Rep. 2012;61(RR-4):1–32. [PubMed] [Google Scholar]
- 18.Smith BD, Yartel AK. Comparison of hepatitis C virus testing strategies: birth cohort versus elevated alanine aminotransferase levels. Am J Prev Med. 2014;47:233–41. doi: 10.1016/j.amepre.2014.05.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Moorman AC, Gordon SC, Rupp LB, et al. Baseline characteristics and mortality among people in care for chronic viral hepatitis: the Chronic Hepatitis Cohort Study. Clin Infect Dis. 2013;56:40–50. doi: 10.1093/cid/cis815. [DOI] [PubMed] [Google Scholar]
- 20.Moyer VA, Force USPST Screening for hepatitis C virus infection in adults: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2013;159:349–57. doi: 10.7326/0003-4819-159-5-201309030-00672. [DOI] [PubMed] [Google Scholar]
- 21.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 2002. Ann Intern Med. 2006;144:705–14. doi: 10.7326/0003-4819-144-10-200605160-00004. [DOI] [PubMed] [Google Scholar]
- 22.Janssen KJM, Donders ART, Harrell FE, et al. Missing covariate data in medical research: to impute is better than to ignore. J Clin Epidemiol. 2010;63:721–7. doi: 10.1016/j.jclinepi.2009.12.008. [DOI] [PubMed] [Google Scholar]
- 23.Raghunathan T. A multivariate technique for multiply imputing missing values using a sequence of regression models. Surv Methodol. 2001;27:79–90. [Google Scholar]
- 24.Barnard J, Rubin DB. Small-sample degrees of freedom with multiple imputation. Biometrika. 1999;86:948–55. [Google Scholar]
- 25.Gelman A, Hill J. Data analysis using regression and multilevel/hierarchical models. New York: Cambridge University Press; 2007. [Google Scholar]
- 26.Bornschlegel K, Berger M, Garg RK, et al. Prevalence of hepatitis C infection in New York City, 2004. J Urban Health. 2009;86:909–17. doi: 10.1007/s11524-009-9396-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.New York City Department of Health and Mental Hygiene. Hepatitis A, B, and C surveillance reports. Available at: http://www.nyc.gov/html/doh/downloads/pdf/cd/cd-hepabc-surveillance-report-08-09.pdf. Accessed 29 April 2014.
- 28.Houston Department of Health and Human Services. Study examines local hepatitis C prevalence. Texas Medical Center News. Available at: http://www.tmc.edu/tmcnews/01_15_02/page_21.html. Accessed 29 April 2014.
- 29.Hall MR, Ray D, Payne JA. Prevalence of hepatitis C, hepatitis B, and human immunodeficiency virus in a Grand Rapids, Michigan emergency department. J Emerg Med. 2010;38:401–5. doi: 10.1016/j.jemermed.2008.03.036. [DOI] [PubMed] [Google Scholar]
- 30.Rein DB, Smith BD, Wittenborn JS, et al. The cost-effectiveness of birth-cohort screening for hepatitis C antibody in U.S. primary care settings. Ann Intern Med. 2012;156:263–70. doi: 10.7326/0003-4819-156-4-201202210-00378. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Torrone EA, Thomas JC, Maman S, et al. Risk behavior disclosure during HIV test counseling. AIDS Patient Care STDS. 2010;24:551–61. doi: 10.1089/apc.2010.0087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Edlin BR, Kresina TF, Raymond DB, et al. Overcoming barriers to prevention, care, and treatment of hepatitis C in illicit drug users. Clin Infect Dis. 2005;40(suppl 5):S276–85. doi: 10.1086/427441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Swan D, Long J, Carr O, et al. Barriers to and facilitators of hepatitis C testing, management, and treatment among current and former injecting drug users: a qualitative exploration. AIDS Patient Care STDS. 2010;24:753–62. doi: 10.1089/apc.2010.0142. [DOI] [PubMed] [Google Scholar]
- 34.Batki S, Sorensen J. Care of injection drug users with HIV. In: Cohen PT, Sande MA, Volberding P, editors. The AIDS knowledge base: a textbook on HIV disease from the University of California, San Francisco and San Francisco General Hospital. 3rd. Philadelphia, PA: Lippincott, Williams and Wilkins; 1999. [Google Scholar]
- 35.O’Brien SF, Xi G, Yi QL, Goldman M. Understanding non-disclosure of deferrable risk: a study of blood donors with a history of intravenous drug use. Transfus Med. 2010;20:15–21. doi: 10.1111/j.1365-3148.2009.00969.x. [DOI] [PubMed] [Google Scholar]
- 36.Shapiro S, Castellana JV, Sprafka JM. Alcohol-containing mouthwashes and oropharyngeal cancer: a spurious association due to underascertainment of confounders? Am J Epidemiol. 1996;144:1091–5. doi: 10.1093/oxfordjournals.aje.a008886. [DOI] [PubMed] [Google Scholar]
- 37.Horton NJ, Kleinman KP. Much ado about nothing: a comparison of missing data methods and software to fit incomplete data regression models. Am Stat. 2007;61:79–90. doi: 10.1198/000313007X172556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Schafer JL, Graham JW. Missing data: our view of the state of the art. Psychol Methods. 2002;7:147–77. [PubMed] [Google Scholar]
- 39.Younossi ZM, Zheng L, Stepanova M, Venkatesan C, Mir HM. Moderate, excessive or heavy alcohol consumption: each is significantly associated with increased mortality in patients with chronic hepatitis C. Aliment Pharmacol Ther. 2013;37:703–9. doi: 10.1111/apt.12265. [DOI] [PubMed] [Google Scholar]
- 40.McMahon BJ, Bruden D, Bruce MG, et al. Adverse outcomes in Alaska natives who recovered from or have chronic hepatitis C infection. Gastroenterology. 2010;138:922–31.e1. doi: 10.1053/j.gastro.2009.10.056. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.