The Prevalence of Hepatitis C Virus Infection in the United States, 1999 through 2002 FREE

—The Editors

A decade ago, the Third National Health and Nutrition Examination Survey (NHANES III, 1988–1994) showed hepatitis C virus (HCV) to be the most common chronic bloodborne infection in the United States (1). An estimated 3.9 million people (1.8% of the population) tested positive for antibody to HCV (anti-HCV), and 2.7 million had chronic infection. Most (65%) anti-HCV–positive persons were 30 to 49 years of age and had been infected for fewer than 20 years. The genetic diversity of HCV circulating in the United States (2) and the pattern of age-specific prevalence (3 - 4) both suggest that the incidence of infection increased substantially in the 1960s and 1970s and peaked in the 1980s.

Identification of HCV-positive persons for appropriate counseling and management is the major focus of a national prevention program, and routine testing is recommended for persons most likely to have HCV infection (5). To determine the characteristics of HCV-infected persons in the general United States population today and to monitor trends in prevalence, we analyzed data on HCV infection from the most recent NHANES.

The National Center for Health Statistics has conducted NHANES periodically to compile nationally representative statistics on the health of the U.S. population (6). The most recent series was begun in 1999 and is designed to run continuously; data are released every 2 years. Our analysis includes data collected from 1999 through 2002.

Participants were chosen according to a stratified, multistage algorithm to produce a representative sample of the civilian, noninstitutionalized population of all 50 states and the District of Columbia. Extensive efforts were made to ensure high participation rates, and all respondents were reimbursed for time and travel expenses (6).

Initially, a questionnaire covering only nonsensitive topics was used to interview participants in person at home. Information on potentially sensitive subjects, such as sexual practices and illicit drug use, was obtained later at a mobile examination center by means of computer-assisted interviewing technology. The ethnicity of each participant was categorized as non-Hispanic white, non-Hispanic black, and Mexican American. Persons not fitting these categories were classified as “other” and were included in the total population. Blood samples were obtained at the mobile examination center (7). Only participants who were 6 years of age or older were eligible for HCV testing because of low sample volume in younger children.

Serum specimens were sent to the Centers for Disease Control and Prevention, where they were tested for anti-HCV by using Ortho HCV enzyme-linked immunosorbent assay (ELISA), version 3.0 (Ortho-Clinical Diagnostics, Raritan, New Jersey). Supplemental recombinant immunoblot assays (RIBA) (Chiron RIBA HCV Strip Immunoblot Assay, version 3.0, Chiron Corp., Emeryville, California) were performed on all specimens that were repeatedly reactive by ELISA testing. For those specimens classified as positive or indeterminate by RIBA, separate, archived aliquots stored at −70 °C and suitable for nucleic acid amplification testing were submitted for quantitative HCV RNA testing using Cobas Amplicor HCV Monitor Test, version 2.0 (Roche Molecular Diagnostics, Pleasanton, California). If that result was below the level of detection, a qualitative assay (Amplicor HCV Test, version 2.0, Roche Molecular Diagnostics) was performed. Samples found to be reactive by enzyme immunoassay and confirmed by RIBA or Amplicor were considered to be anti-HCV–positive. Alanine aminotransferase (ALT) levels (reference range, 0 to 39 U/L) were measured in specimens that had been stored and shipped under appropriate refrigeration conditions (4 °C to 8 °C).

All statistical analyses were performed with SUDAAN software (RTI International, Research Triangle Park, North Carolina) according to National Center for Health Statistics guidelines. We used appropriate study design variables and published weights that were further adjusted to compensate for missing anti-HCV values (8). These weights accounted for oversampling of certain demographic groups (6) and for nonparticipation such that the sum of the weights for persons with anti-HCV results equaled the U.S. civilian, noninstitutionalized population 6 years of age and older. To estimate the number of HCV RNA–positive persons, these weights were further adjusted to compensate for the RIBA-positive and RIBA-indeterminate specimens that were unavailable for RNA testing because of inadequate specimen volumes.

Proportions from univariable analyses were compared by using chi-square tests (as implemented in SUDAAN). The P values presented were not corrected for multiple comparisons; P values less than 0.05 were considered statistically significant. Two logistic regression models were used for multivariable analysis; 1 model was used for persons 20 to 59 years of age whose drug use and sexual practices data were available, and the other model was used for persons 60 years of age or older. Two variables, history of blood transfusion (both models) and injection drug use (persons 20 to 59 years of age), were forced into the models on the basis of substantial published data that has established them as risk factors for HCV infection. We sought the most parsimonious model by using these and all other variables that were significant at a P value less than 0.20 on univariable analysis. With the resulting model, we then examined the effect of adding other variables of interest, including those variables that had been excluded at earlier steps in the modeling process. In the final models, all first-order interactions were examined for statistical significance, epidemiologic plausibility, and the impact of their inclusion on the other model parameters.

No external funding was received for this study.

Of 21 509 participants 6 years of age or older, 17 548 were interviewed and 15 079 gave a blood sample suitable for anti-HCV testing (final response rate for testing, 70.1%). Among those who completed home interviews, participation rates did not differ significantly between those with and without risk factors for HCV infection.

The weighted prevalence of anti-HCV in the United States was 1.6% (95% CI, 1.3% to 1.9%), corresponding to 4.1 million (CI, 3.4 million to 4.9 million) anti-HCV–positive persons (Table 1). Of anti-HCV–positive participants, 78.8% had specimens suitable for HCV RNA testing; 79.7% (CI, 70.4% to 86.6%) of these tested positive for HCV RNA. After we accounted for untested specimens, the nationwide prevalence of HCV RNA among all participants was 1.3% (CI, 1.0% to 1.5%), equating to 3.2 million (CI, 2.7 million to 3.9 million) HCV RNA–positive persons.

Anti-HCV prevalence was significantly higher in men than in women (Table 1). Prevalence was also higher in non-Hispanic black participants than in either of the other 2 ethnic groups. Among persons younger than 50 years of age, prevalence of anti-HCV increased with age from 1.0% in those 20 to 29 years of age to a peak of 4.3% in those 40 to 49 years of age (Figure 1). Among older persons, anti-HCV prevalence decreased to 1.6% in persons 50 to 59 years of age and to 0.9% in persons 60 years of age and older. Prevalence was higher in men than in women in most age groups (Figure 1). The higher overall prevalence among non-Hispanic black persons compared with non-Hispanic white persons was almost entirely attributable to differences among older participants. Among participants 40 to 49 years of age, 9.4% of non-Hispanic black persons had positive results for anti-HCV compared with 3.8% of non-Hispanic white persons (P < 0.001); of participants 50 years of age or older, 3.3% of non-Hispanic black persons had positive results compared with 0.9% of non-Hispanic white persons (P = 0.002). The demographic group with the highest prevalence was non-Hispanic black men between 40 and 49 years of age (13.6% [CI, 10.0% to 18.2%]). Prevalence was not significantly different between non-Hispanic black and non-Hispanic white persons who were younger than 40 years of age (1.2% vs. 1.1%; P = 0.73). Participants who were born in the United States had a higher prevalence of anti-HCV than those who were not, and prevalence increased with decreasing family income and level of education (Table 1). Among men, prevalence did not vary according to service in the military (Table 1). The sample of women who had served in the military was too small to analyze.

The overall prevalence of anti-HCV in the current survey was similar to that observed in NHANES III, but the peak in age-specific prevalence moved from persons 30 to 39 years of age to those 40 to 49 years of age (Figure 2). When participants in both surveys were categorized according to approximate birth year, there was little difference in prevalence (Figure 2). Most anti-HCV–positive participants identified in both surveys (68.7% in NHANES III and 65.6% in the current survey) were born between 1945 and 1964.

Risk factor analysis was restricted to adults 20 years of age or older because of the limited risk factor data and the small number of anti-HCV–positive persons (n = 3) among younger participants. Among persons 20 to 59 years of age, 21.3% (CI, 19.4% to 23.4%) had ever used illicit drugs (injection or noninjection drugs, excluding marijuana) and 2.0% (CI, 1.6% to 2.7%) had ever injected illicit drugs. Among persons 20 to 39 years of age, lifetime history of injection drug use was more common in non-Hispanic white persons (2.4%) than in non-Hispanic black persons (0.6%) (P = 0.024 for the difference). However, among persons 40 to 59 years of age, history of injection drug use was more common among non-Hispanic black persons (5.3%) than among non-Hispanic white persons (2.0%) (P = 0.016 for the difference). Of all those who acknowledged ever injecting drugs, regardless of age, most (83.3%) had not done so for at least 1 year before the survey. Injection drug use correlated with such demographic factors as low family income and little education and with such risk factors as high numbers of sex partners and noninjection drug use.

Among adults who had ever injected illicit drugs, prevalence of anti-HCV was 48.4% in those between 20 and 59 years of age. Overall prevalence of anti-HCV was 57.5% and was higher among non-Hispanic black persons (88.5%) than among non-Hispanic white persons (51.7%) (P = 0.034). Anti-HCV prevalence was significantly higher among those who had ever injected drugs than among those who had ever used noninjection illicit drugs (other than marijuana); both of these groups had a significantly higher prevalence than persons who had never used illicit drugs or had only used marijuana (Table 1). Anti-HCV prevalence was also significantly correlated with a history of receiving a blood transfusion before 1992, increasing lifetime number of sexual partners, decreasing age at first sexual intercourse, and positive results on herpes simplex virus type 2 antibody testing (Table 1). Among men, anti-HCV was not associated with ever having had sex with another man (data not shown).

In a logistic regression model that included data from participants between 20 and 59 years of age and controlled for age and sex, anti-HCV was most strongly associated with injection drug use. Anti-HCV was also associated with Mexican-American ethnicity, birth in the United States, low family income, noninjection illicit drug use, and a lifetime total of 20 or more sexual partners (Table 2). Blood transfusion before 1992 was not significantly associated with anti-HCV after adjustment for other variables but was kept in the model because it is an accepted risk factor for HCV infection. In a separate logistic regression model for participants 60 years of age or older whose drug use and sexual histories had not been collected, anti-HCV was independently associated with non-Hispanic black ethnicity and a history of blood transfusion before 1992 (Table 3).

Among anti-HCV–positive participants whose serum specimens were sent for HCV RNA determination, more men (89.0%) than women (63.4%) had positive results (P = 0.017), and more persons who were 40 years of age or older (89.6%) had positive results than did persons who were 6 to 39 years of age (60.2%) (P = 0.023). Rates of HCV RNA positivity were similar among non-Hispanic white persons (77.3%) and non-Hispanic black persons (79.7%).

Participants who had positive results for HCV RNA were significantly more likely to have abnormal serum ALT levels (58.7%) than were anti-HCV–positive, HCV RNA–negative participants (10.3%) or participants with no HCV markers (8.8%). The prevalences in the latter 2 groups were not significantly different from each other.

Adults who had positive results for HCV RNA reported heavier alcohol intake during the previous year than other adults; this group was almost 3 times more likely to consume an average of more than 1 drink per day (35.3% vs. 13.5%; P = 0.003) and almost 8 times more likely to consume more than 3 drinks per day (19.2% vs. 2.4%; P = 0.010). Adults who were HCV RNA positive were also more likely to have had 5 or more drinks in a single day during the previous year than other adults (47.8% vs. 27.7% per year; P = 0.002), and 33.2% had done so on at least 50 days during the previous year. Regardless of HCV status, the proportion of adults with abnormal serum ALT levels was greater among those who reported more than 1 drink per day than in those who reported 1 or fewer drinks per day (Table 4). This difference was statistically significant only among anti-HCV–negative persons.

Of HCV RNA–positive adults, 9.8% had received at least 1 dose of hepatitis A vaccine and 14.7% had received at least 1 dose of hepatitis B vaccine. These rates were not significantly different from those of other adults (P = 0.66 and P = 0.132, respectively).

Among persons 20 to 59 years of age, 7.3% had a history of injection drug use or had received a blood transfusion before 1992; anti-HCV testing of this group would identify 53.1% of HCV RNA–positive persons (Table 5). If testing also included those with abnormal serum ALT levels (a common “incidental finding” in medical practice), 85.1% of HCV-infected persons could be identified on the basis of testing 18.1% of the population. Among persons 60 years of age and older, 21.1% had received a blood transfusion before 1992 or had an abnormal serum ALT level; anti-HCV testing of this population segment would identify 87.4% of HCV RNA–positive persons.

These results highlight the substantial burden of infection attributable to HCV in the United States. Approximately 3.2 million persons in the general population have chronic infection; however, this figure is probably an underestimate because the sampling frame of NHANES did not include incarcerated or homeless persons, who are known to have high prevalences of HCV infection (5,9 - 10). If the estimated number of HCV infections among incarcerated persons (9) were added to those from NHANES, the total number of persons with chronic infection would increase to an estimated 3.5 million.

It is unknown what proportion of HCV-positive persons in the United States are aware of their infection; however, the high rates of alcohol use (a co-factor known to accelerate HCV-related chronic liver disease) and low rates of hepatitis A and hepatitis B immunization noted in our study suggest that many of the HCV-positive persons we identified have not been tested, have not received appropriate counseling, or have disregarded that counseling. Most HCV-infected persons are only now reaching an age when complications of liver disease may start to develop, and multiple studies have predicted a rise in future HCV-related morbidity and mortality rates (3 - 4,11 - 12). To characterize the type of care and counseling that HCV-infected persons receive in the United States, researchers are performing a follow-up survey of anti-HCV–positive participants who were identified in NHANES.

Most HCV-infected Americans were born between 1945 and 1964, and most have engaged in high-risk drug and sexual behaviors at some point in their lives. Many with such histories may not recognize that transient behavior decades earlier has put them at risk for a potentially lifelong infection. Of those who ever injected drugs, for example, more than 8 out of 10 had not done so in the past year and may not have injected drugs for many years.

High-risk behavior does not account for all HCV infections in the United States. Among HCV-positive survey participants 60 years of age or older, more than half reported receiving blood transfusions before 1992. Among younger participants, blood transfusion probably accounted for a relatively smaller proportion of HCV infections. This trend reflects public health interventions during the 1980s and 1990s that led to a progressive reduction in the risk for post-transfusion HCV infection (5,13 - 14). Other well-known risk factors, such as long-term hemodialysis and health care work involving frequent exposure to blood (5), could not be evaluated in NHANES because of their low frequency or lack of availability in the NHANES data set. These risk factors generally account for fewer than 10% of infections (15 - 16).

The lower prevalence of anti-HCV among adolescents and young adults (<30 years of age) compared with that among older adults is encouraging. The similarity of the prevalences among young non-Hispanic black and white adults, as well as the lower prevalence of injection drug use among young non-Hispanic black persons, is consistent with that of another national survey (17); this finding suggests that younger non-Hispanic black persons may not be subject to the disproportionately high burden of disease that was seen in the previous generation. However, the small number of younger anti-HCV–positive participants limits definitive conclusions.

The prevalence of chronic infection did not vary significantly by race. In the previous NHANES, non-Hispanic black persons were more likely to have chronic infection than non-Hispanic white persons (1). This difference could represent a loss of HCV markers as the cohort aged; a real change in prevalence of chronic infection; or a chance observation, particularly because of the smaller numbers of HCV-positive participants in the current survey.

There are limitations to NHANES, a survey with a broad scope of which HCV is but one of many components. In addition to excluding certain high-risk populations, the survey allowed only self-reporting of such risk factors as injection drug use; the limitation of this format may result in an underestimation of the prevalence of these factors and the strength of association with HCV. In 1 longitudinal study, 7.4% of persons who had acknowledged a history of cocaine use on at least 2 previous surveys did not do so when responding to subsequent surveys (18). Studies of volunteer blood donors found that 30% to 50% of those whose donations were rejected because of a positive HCV test result admitted to a history of injection drug use after being presented with their result (19 - 21). None of these donors had reported this risk factor at the time of donation. Because injection drug use is an overwhelmingly strong risk factor for HCV infection, small amounts of unacknowledged injection drug use can confound any cross-sectional survey (22). In our study, for example, injection drug use correlated with such variables as low family income, high numbers of sex partners, and noninjection drug use. Unacknowledged injection drug use may account for at least part of the association observed between these variables and HCV infection. Finally, sources of HCV infection for individual participants cannot be definitively determined in a cross-sectional study.

The lack of change in HCV prevalence between the 1988 to 1994 NHANES and the 1999 to 2002 NHANES is consistent with the 5- to 10-fold decrease in incidence of acute hepatitis C that was observed in the early 1990s (3,15). However, there is no guarantee that the incidence of acute infection will remain low. Among young injection drug users, the annual incidence of HCV infection ranges from 10% to 36% (23 - 27). Furthermore, the stabilization of HCV prevalence will not prevent the increase in cirrhosis and liver cancer that is projected to result from HCV infections acquired in previous decades (11 - 12,28). Therefore, prevention of HCV-related liver disease should remain a key focus of clinical and public health interventions, which rely on the identification of persons with chronic HCV infection (5). Although there has been insufficient time to demonstrate that antiviral therapies and other interventions directly reduce HCV-related morbidity and mortality rates, we cannot ignore the potential benefits of reducing liver injury by eradicating the virus or by eliminating other hepatotoxic agents (for example, infections and alcohol) in the interim (29 - 30).

There may be concern regarding the feasibility of ascertaining risk factor histories during routine patient visits, but the data in this report demonstrate that about 85% of HCV-positive persons can be identified on the basis of 1 of 2 risk factors, history of injection drug use or receipt of a blood transfusion before 1992, or 1 laboratory result showing an abnormal serum ALT level. Of particular importance to clinicians is that most HCV-infected persons who acknowledge a history of injection drug use have not used these drugs recently and may not fit conventional stereotypes of injection drug users. All patients should be asked about their history of injection drug use, and those who report such a history, no matter how transient or remote, should be tested for HCV.