Hepatitis B Prevalence in Young Women Living in Low-Income Areas: The Population-Based San Francisco Bay Area's Young Women's Survey

HEPATITIS B VIRUS (HBV) infection is a major public health concern in the United States. Even with the availability of an effective vaccine, 200,000 to 300,000 new infections occur every year,1 of which 6% to 10% develop into chronic HBV infection.2,3 Acute HBV infection is mostly asymptomatic; therefore, serologic studies are necessary to understand more fully the epidemiology of HBV infection.4,5

Defining sexual and other risk factors of HBV infection in young women, particularly those of low socioeconomic status, is of interest for several reasons. Young adults comprise the majority of new HBV infections in the United States, and sexual transmission is the most commonly reported HBV risk factor in women.6,7 Immunization programs for adults at risk for sexually acquired HBV infection have been underused. The Centers for Disease Control and Prevention did not establish HBV immunization guidelines for newborns until 1991; therefore, many persons born before 1991 remain unimmunized.8,9 In addition, 30% to 40% of all reported acute HBV infections among minority groups and persons of low socioeconomic status are classified as "no identified risk."6

In the United States, Alaskan Natives, Pacific Islanders, and children of immigrant mothers from countries with endemic HBV infection (many Asian and Pacific Island countries) have high rates of infection due to perinatal or household transmission.8 Likewise, intravenous drug users (IDUs) have high HBV infection rates due to parenteral transmission.10,11 By excluding these two risk groups, sexual transmission accounts for most HBV infections in persons younger than 30 years.

Transfusion and hemodialysis have been rare sources of HBV infection in the last 30 years, mostly because of HBV screening and the exclusion of paid donors in blood banks.6,12 Occupational exposure has also been uncommon, particularly since the introduction of preexposure immunization and postexposure prophylaxis in the last 20 years.6,13 Even before the 1991 perinatal vaccination strategies were implemented, HBV infections among infants and children outside of the highly endemic immigrant or Alaskan Native populations contributed relatively little to the total HBV infection burden.5,7

Studies identifying correlates of sexual behavior and HBV infection in general populations have been rare. Increased risk of HBV infection among heterosexual men and women has been associated with number of sexual partners, number of years of sexual activity, history of sexually transmitted diseases (STDs), and serologic history of syphilis infection.7,14-17 We used data from a random population-based serosurvey of young women in the San Francisco Bay Area to assess sexual risk behavior and other potential correlates of HBV infection.

The San Francisco Bay Area Young Women's Survey (YWS) was a one-stage probability cluster sample of low-income census tract households in which the target population of women age 18 to 29 years resided. This population-based, door-to-door, cross-sectional survey determined the prevalence of HIV, STDs, hepatitis infection, and related drug-using and sexual risk behaviors. The methodologies and principal outcomes are described elsewhere.18 The survey was conducted in five northern California counties: Alameda, Contra Costa, San Francisco, San Joaquin, and San Mateo. Within each county we identified census-block groups with median household income below the 10th percentile, as determined from the 1990 census data. Outreach teams identified eligible participants by enumerating household occupants in randomly selected street blocks during door-to-door encounters. Households were defined as any sheltered dwelling and could include abandoned buildings with squatting residents. Participants first provided informed consent and then were administered a 45-minute, face-to-face, structured interview. The survey collected information regarding sociodemographic characteristics, medical history, drug use history, sexual behavior, and knowledge, attitudes, and beliefs related to HIV transmission and prevention. After counseling and referral information, serum samples were collected from participants in all five counties and urine samples were collected in three counties. The target enrollment was 500 women per county and 2,500 women overall. Sera from the counties of Alameda, San Mateo, and San Francisco were tested for HBV.

The study was approved by the State of California, Health and Welfare Agency, Committee for the Protection of Human Subjects.

Hepatitis B virus core antibody in serum was detected by Abbott Corzyme EIA (Abbott Laboratories, Abbott Park, IL). Positive specimens were tested for the HBV surface antigen (HBsAg) by Abbott Auszyme Microparticle EIA (Abbott Laboratories). Blood samples were tested for syphilis by rapid plasma reagin or VDRL, and reactive specimens were confirmed by microhemagglutination assay for antibodies to Treponema pallidum.

Hepatitis C virus (HCV) antibody was detected by HCV-encoded antigen (Recombinant c22-3, c200, NS5I, Ortho HCV EIA 3.0) and positive sera were confirmed by recombinant immunoblot assay (Chiron, Emeryville, CA; Ortho Diagnostic Systems, Raritan, NJ). Sera were also tested for herpes simplex virus (HSV). A strip-recombinant type-specific immunoblot assay (RIBA HSV type 1 and 2 SIA; Chiron) was used to differentiate HSV type 1-specific and type 2-specific antibodies on recombinant antigen bands for gG1, gB1, gG2, gD2.19,20 Urine samples were tested for Chlamydia trachomatis and Neisseria gonorrhoeae using ligase chain reaction (LCx, Abbott Laboratories).

The survey methodology was devised to yield population-based estimates by adjusting for the population sample size of each county and for the design effect introduced by the cluster-sample method. The analysis included measuring the variability among and within each of the randomized block clusters of each county. Adjusted population-based estimates and correlates of HBV infection (defined as HB core antibody positivity, which included HbsAg-positive specimens) were calculated for all participants with available data in the YWS sample and also for a YWS subsample. We sought to identify correlates of sexually acquired HBV infection in a YWS subsample whose risk for HBV had been through sexual transmission by excluding IDUs, transfusion recipients, and persons of Asian/Pacific Islander (A/PI) ethnicity. By excluding these groups, there is the possibility that we excluded HBV infections due to sexual transmission. We excluded these groups from the analysis because these HBV transmission routes account for a markedly high HBV prevalence (both in the literature and in our analysis) not associated with sexual transmission.

The questionnaire collected information regarding only transfusions undergone after 1980. Transfusion history was also an exclusion criteria of the subsample despite current literature suggesting that the likelihood of HBV infection due to blood products in persons younger than 30 years is negligible.6,12 We found transfusion history to be a significant correlate of HBV infection in bivariate analysis (Table 1), but it was correlated with sexual risk behavior and was ultimately not significant in multivariate analysis (data not shown).

TABLE 1. Characteristics and Hepatitis B Virus Infection Prevalence*

Our base model controlled for demographics and included sexual risk behavior variables associated with HBV infection (e.g., number of sexual partners, age at first vaginal intercourse, history of STD).14-18 Demographic variables such as age, race, and income have also been associated with HBV infection.5,21 We then tested other sexual risk behavior variables as significant correlates of HBV infection by including them in the base model and retaining those with statistical significance (P < 0.05). HSV type 2 (HSV-2) infection, a viral STD that is associated with lifetime sexual behavior risk,22-24 was also tested.

There was no correlation analysis of chronic HBV (defined as HBsAg positivity) because there were few infections found (four A/PI, five black, one IDU). All statistical procedures were done using STATA 6.0 software (College Station, TX).

From April 1996 to January 1998, 14,223 inhabited dwellings were enumerated in 448 randomly selected street blocks within three California counties. Contact was made with a resident in 76% of these dwellings. Of 2,148 eligible women identified, 1,439 (67%) agreed to participate in the study. Of these participants, 1,337 (92.9%) had HBV test results that were available for analysis; missing results were due to refusal or insufficient amount for testing. There was no significant difference in demographics, risk behaviors, and prevalence of other STDs between those with and without HBV serologic test results. The population-based estimate of HBV infection (HBV core antibody) in young women of low socioeconomic status in three San Francisco Bay Area counties was 8.9% (95% CI, 7.2-10.5%), and the estimate of of chronic HBV infection (HBsAg) was 0.75% (95% CI, 0.27-1.22).

Table 1 shows the characteristics and population-based estimates of HBV infection among the entire YWS sample. Alameda and San Francisco counties had a significantly higher HBV prevalence than San Mateo county (11.3% and 10.0% versus 5.2%, respectively). Persons of different race/ethnicity or country of birth also had significantly different HBV prevalence. The highest prevalence was among persons of A/PI ethnicity (26.8%) and A/PI country of birth (30.2%). The risk for HBV infection increased with age (P = 0.03, Mantel-Haenzel chi-square test) and lower education level (odds ratio [OR], 1.6; P = 0.05). IDUs had an HBV prevalence of 22%. Transfusion was a significant risk factor of HBV infection (OR 2.7, P = 0.02). Of those who reported a history of HBV infection, 62.4% were HBcAb positive; however, only 16.5% of HBcAb-positive persons reported a history of HBV infection. Persons with HCV infection had an HBV prevalence of 27.0%. Most of the HCV infections (70.6%) were among IDUs.

Figure 1 depicts the distribution of HBV infections among YWS participants by the putative source of infection. Cause of infection (in order of precedence) was perinatal transmission (A/PI ethnicity), parenteral transmission (intravenous drug use, transfusion), and sexual transmission. Sexual transmission accounted for 58% of HBV infection.

Fig. 1. Putative sources of hepatitis B virus infection among San Francisco Bay Area's Young Women's Survey participants (n = 117). *History of intravenous drug use or transfusion. **Asian/Pacific Islander ethnicity.

A total of 1,126 participants met the inclusion criteria of the YWS subsample to identify correlates of sexually acquired HBV infection. The population-based estimate of HBV infection among these women was 6.4% (95% CI, 5.7-7.6%), whereas the population estimate for chronic HBV was 0.44% (95% CI, 0.01-0.87%).

Table 2 shows the demographic and sexual risk characteristics of the survey subsample. There was increased risk for sexually acquired HBV infection among black women and among those with a monthly household income of less than $500. The prevalence of HBV infection among blacks was eight times that of whites (OR, 8.1; P < 0.001). HBV infection increased with age among blacks, though the trend was not statistically significant (P = 0.22, Mantel-Haenzel chi-square test, data not shown). Being born in Mexico versus the United States was associated with a decreased risk for sexually acquired HBV infection (P = 0.03).

TABLE 2. Characteristics, Sexual Risk History, and Hepatitis B Virus Infection Prevalence*

Among the lifetime sexual behavior risk variables, only number of lifetime sexual partners and history of STD were significantly associated with sexually acquired HBV infection. HSV-2 infection was also a correlate of HBV infection (P < 0.001). Other sexual behavior risk variables that were not significantly associated with increased risk for HBV infection were having male sex partners that have sex with men, were IDUs, being high on drugs during sex, history of sex work, and age at first vaginal intercourse. In addition, more recent markers of sexual risk behavior including number of partners in the last 6 months, recent condom use, and current chlamydial, gonococcal, or syphilis infection were not significant correlates of sexually acquired HBV infection.

Significant correlates of sexually acquired HBV infection in multivariate analysis were black race (OR, 3.9; 95% CI, 1.2-11.9) versus white race, and HSV-2 infection (OR, 2.0; 1.0-3.9). The multivariate model was adjusted by county, age group, income, education, age at first intercourse, lifetime number of sexual partners, and history of STD.

The enumeration and participation rates met the goals originally set for the YWS of 80% enumeration and 70% participation and were comparable to a similar study.25 Regardless, the nonparticipating section of the target population may encompass small groups that contribute significantly to risk behavior and disease prevalence such as IDUs, sex workers, and homeless women not found living in a sheltered dwelling. Information regarding nonparticipants was unavailable. The inclusion criteria of the sexually acquired HBV subsample may have included infections due to perinatal transmission (despite the low rates in non-A/PI) and parenteral transmission because of unavailable transfusion information before 1980.

The San Francisco Bay Area YWS was the first population-based study to assess behavior risk and sexually transmitted disease prevalence among women living in low-income areas. The population-based estimate of HBV for the target group was 8.9%. The most recent national estimate of HBV prevalence comes from analysis of The National Health and Nutrition Examination Surveys (1976-1994).7 The HBV infection prevalence for women in a similar age group as the YWS sample was 6.6% (95% CI, 4.4-10.1%) for non-Hispanic blacks, 2.9% (95%CI, 1.8-4.6) for Mexican Americans, and 1.5% (95% CI, 0.7-3.2) for non-Hispanic whites (McQuillan GM, NHANES III unpublished data, 1999). Although the YWS HBV prevalence among black women is almost double that of the national estimates, the confidence intervals overlap and are therefore not statistically different.

Black race was a correlate of HBV infection even while controlling for socioeconomic status and sexual behavior, which suggests that it is the prevalence within the sexual network rather than the type of high-risk behavior that is most important for transmission. Black race has been identified as an independent risk factor for HBV in the United States1,7 and for other STDs in the San Francisco Bay Area.26,27

The YWS was designed to study a population within the narrow range of young age, low income, and low levels education. Therefore, because of the homogeneity of this target population, the contribution of other demographic variables to HBV infection was more difficult to detect.

HSV-2 was also a correlate of HBV infection that was independent of sexual behavior risk variables. HSV-2 infection may increase the likelihood of HBV infection for the same reasons that it has been hypothesized to facilitate or cause HIV transmission (i.e., ulceration, disruption of the mucosal membrane, increased viral shedding).28,29 Further research should examine whether HSV-2 infection predisposes to sexually acquired HBV or vice-versa, or whether HSV-2 is a better marker for lifetime sexual risk compared with reported sexual risk behavior.

Prevention efforts need to focus on enhanced immunization of young women in low-income groups and in the San Francisco Bay Area, specifically in low-income black communities. Because there were no independent sexual behavior correlates of HBV infection, our analysis suggests that universal immunization of young adults would be a better strategy to curtail transmission than the present guidelines, which identify and target high-risk persons.8 The susceptible persons at risk for sexually acquired HBV infection are not characterized by high-risk sexual behavior.

Population-based studies are expensive and labor intensive, but are critical in monitoring patterns of disease and behavior in general populations and communities associated with the highest burden of disease; in the case of hepatitis B where asymptomatic infection and competing health priorities cause the disease to be relatively underappreciated, our study provides timely and relevant data to enhance prevention efforts.