The Evolving Epidemiology of Hepatitis A in the United States: Incidence and Molecular Epidemiology From Population-Based Surveillance, 2005-2007

Background  The incidence of hepatitis A virus (HAV) disease is the lowest ever in the United States. We describe recent incidence and characteristics of cases of HAV disease from 6 US sites conducting hepatitis surveillance in the Emerging Infections Program.

Methods  Health departments conducted enhanced, population-based surveillance for HAV from 2005 through 2007. Demographic and risk factor data were collected on suspected cases (persons with a positive IgM anti-HAV result) using a standard form. Remnant serum specimens from a convenience sample of cases were tested by polymerase chain reaction, followed by sequencing the 315-nucleotide segment of the VP1-P2B junction.

Results  There were 1156 HAV cases reported during 2005 through 2007. The combined population under surveillance was 29.8 million in 2007. The overall annual incidence rate was 1.3 per 100 000 population (range by site, 0.7-2.3). Of reported cases, 53.4% were male, 42.4% were white, 44.7% were aged 15 to 39 years, and 91.4% resided in urban areas. Reported risk factors were international travel (45.8%), contact with a case (14.8%), employee or child in a daycare center (7.6%), exposure during a food or waterborne common-source outbreak (7.2%), illicit drug use (4.3%), and men who had sex with men (3.9%). Genotypes among the 271 case specimens were IA (87.8%), IB (11.4%), and IIIA (0.7%). Of the 271 polymerase chain reaction–positive specimens, 131 (48.3%) were from cases reporting travel or exposure to a traveler; 58 of the 131 cases reported travel to Mexico, and 53 of the 58 were within the US-IA₁ cluster.

Conclusions  International travel was the predominant risk factor for HAV transmission. Health care providers should encourage vaccination of at-risk travelers.

Worldwide, there are an estimated 1.4 million cases of hepatitis A virus (HAV) infection each year.¹ Infection patterns vary geographically. In developing countries with poor sanitary infrastructure and high infection rates, exposed children develop immunity without presenting symptoms of disease and outbreaks are infrequent. In contrast, residents of areas with low levels of HAV infection have fewer opportunities in childhood for developing passive immunity, leading to populations of susceptible adults who are more likely to develop symptomatic disease if infected. Recent outbreaks in the European Union illustrate the pattern of developed countries with record low levels of disease incidence. During 2007 through 2008, outbreaks of HAV disease were reported among injection drug users in Czech Republic,² among restaurant patrons and others in Latvia,³ among persons of low socioeconomic status in Slovakia,⁴ and among travelers in France,⁵ Belgium,⁶ and Germany.⁷ Understanding HAV transmission patterns is critical for targeting vaccination toward susceptible populations.

In the United States, the National Notifiable Diseases Surveillance System (NNDSS) of the Centers for Disease Control and Prevention (CDC) has received voluntary reports of acute, viral hepatitis from all states and the District of Columbia since 1966,⁸ and in 2007, the incidence of HAV disease from NNDSS was the lowest ever reported. The last large outbreak of HAV in the United States occurred in 2003, associated with uncooked, imported green onions served in restaurants.⁹ More than ever, surveillance is needed to identify cases early to limit and prevent disease.

To supplement information from NNDSS and to build a specimen library of hepatitis strains, the CDC recruited states to participate in the Emerging Infections Program (EIP) (http://www.cdc.gov/ncpdcid/deiss/about_eip.html). Funding for molecular testing of HAV specimens in EIP was provided by the Food Safety Initiative of the CDC. The objective of this report was to describe recent incidence and characteristics of cases of HAV disease, including source of infection and molecular characteristics, from the 6 US sites conducting population-based hepatitis surveillance in the EIP.

During 2005 through 2007, the participating EIP sites were Colorado, Connecticut, Minnesota, 34 counties in upstate New York, New York City, and Oregon. The combined population under surveillance in these 6 sites was an estimated 29.8 million in 2007. In all sites, laboratories and health care providers are required to report cases of HAV disease; sites communicate periodically with clinical laboratories to improve compliance. A case of hepatitis A was defined using Council of State and Territorial Epidemiologists criteria as a person with acute illness with discrete onset of symptoms and either jaundice or elevated serum aminotransferase levels and a positive test result for immunoglobulin M to hepatitis A virus (IgM anti-HAV) (http://www.cdc.gov/ncphi/disss/nndss/casedef/case_definitions.htm). Each site evaluated case reports to determine if they met the definition and then investigated cases to determine the source of infection and prevent secondary transmission to exposed contacts, if needed.¹⁰ Investigations of suspected cases were conducted by health department personnel who collected, either through health care provider or patient interview, basic clinical and demographic characteristics, along with risks for transmission during the 2 to 6 weeks prior to onset of symptoms. Sites submitted these data electronically, without personal identifying information, to the CDC every month. We selected cases reported to the participating sites from January 1, 2005, through December 31, 2007, and submitted to the CDC by December 2008. We defined a travel-related case as a case with a yes response to either of these questions: “In the 2-6 weeks before symptom onset, did the patient travel outside the USA or Canada?”; or “In the 3 months prior to symptom onset, did anyone in the patient's household travel outside of the USA or Canada?” Cases with a yes response to either question were also asked to list the country or countries of travel. We categorized reported country of travel by geographic region, consistent with macrogeographic regions used by the United Nations (http://unstats.un.org/unsd/methods/m49/m49regin.htm).

We stratified cases reported from January 2005 through December 2007 by year, site, sex, race, and age as numerators. We obtained corresponding denominators by year, site, sex, race, and age from the US Bureau of the Census. We classified counties into urban and nonurban using criteria for metropolitan statistical areas and divisions from the Office of Management and Budget (see http://www.census.gov/population/estimates/metro-city/List4.txt).

Sites contacted clinical laboratories to collect remnant serum specimens obtained within 6 weeks of the onset of symptoms from confirmed cases, if available. Serum specimens were shipped frozen to the CDC for genetic characterization and storage in a library of viral strains. If the volume of serum was 0.2 mL or greater, the Division of Viral Hepatitis laboratory at the CDC conducted nucleic acid detection and amplification using reverse transcription–polymerase chain reaction (RT-PCR). Specimens that were PCR positive were then sequenced for the 315-nucleotide segment of the VP1-P2B junction using methods described previously.¹¹ The Accelrys GCG Package (Accelrys Inc, San Diego, California) was used for sequence analyses. Sequences obtained from EIP specimens were compared with a CDC reference library containing more than 3000 HAV isolates and described in more detail elsewhere.¹² Briefly, the CDC reference library of HAV isolates is a convenience sample and includes isolates from multiple sources including a special CDC collaborative project from the interior of Mexico and the US-Mexican border, the CDC Sentinel Counties project, and select sequences from GenBank.¹² The reference library isolates are mostly from North America and have been helpful in understanding transmission networks of HAV in the United States.¹³ Within the genotypes and subgenotypes identified, 3 major clusters have been described: US-IA₁, US-IA₂, and US-IA₃. Typically, isolates from Mexico have been closely related to isolates in the US-IA₁ cluster, isolates from injecting drug users (mostly US-IA₂), and isolates from men who have sex with men (mostly US-IA₃).¹² Final phylogenetic tree construction was based on an unweighted pair-group method with arithmetic mean algorithms.¹²

There were 1156 cases of acute HAV disease reported during 2005 through 2007; of these, 566 (49.0%) were reported from New York City. In most sites, at least half of the cases were male (range, 49.1%-54.6%) and white (range, 20.1%-76.7%) (Table 1); in New York City, 38.7% of cases were Hispanic. The greatest proportion of cases were aged 15 to 39 years but varied by site, from 36.1% in New York State to 49.2% in Colorado. Most cases were residents of urban areas (range, 77.4%-100%).

In New York City and Minnesota, there were select peaks in the rate of case reports, consistent with outbreaks (Figure 1). In New York City, there were 3 peaks in the rate observed during quarter 3 in 2005, 2006, and 2007. An investigation of 75 confirmed cases during 2006 through 2007 revealed that those cases were mostly Hispanic (43%) and reporting travel (48%) to an endemic region and not part of an outbreak. In Minnesota, the rate peaked during quarters 2 and 3 of 2007, reflecting 2 independent outbreaks. On further investigation, these outbreaks were found to be associated with restaurants; however, no individual contaminated food item or infected food handler was identified. The remaining sites were characterized by relatively constant case reporting over time.

For all sites combined, the incidence rate of HAV disease was 1.3 per 100 000 population (range by site, 0.7-2.3 per 100 000) during 2005 through 2007 (Table 1). The rate was 1.5 in 2005, 1.1 in 2006, and 1.2 in 2007. Annual incidence was somewhat higher among men (1.4 per 100 000) than women (1.2 per 100 000) and among persons aged 15 to 39 years (1.7 per 100 000), Hispanics (2.7 per 100 000), and persons reporting multiple race/ethnicity (5.6 per 100 000). Within states, county-specific annual incidence per 100 000 population ranged by site as follows: 0.0 to 4.3 in Colorado, 0.4 to 1.1 in Connecticut, 0.0 to 26.5 in Minnesota, 0.0 to 2.0 in participating counties in New York State, 0.5 to 2.7 in New York City, and 0.0 to 5.3 in Oregon.

Cases could report more than 1 behavior or exposure during the 2- to 6-week period before onset of symptoms as potential risk factors for infection (Figure 2). Overall, 529 (45.8%; range by site, 32.2%-50.8%) of the 1156 cases reported travel as a risk factor, including 476 (41.2%) cases reporting personal international travel and 217 (18.4%) cases reporting exposure to an international traveler. Of the 529 travel-related cases, 164 (14.2%) both traveled and were exposed to a traveler, and 52 (4.5%) reported being exposed to a traveler without having traveled personally. In 395 cases (34.2%), traveling or being exposed to a traveler was reported as their only risk factor. The destination most frequently reported among travelers (30.3%) and household contacts (26.3%) was Mexico. However, the frequency of travel destinations reported varied widely by site: 69.8% of cases from Oregon reported travel to Mexico, in contrast to only 6.9% of cases from New York State.

Other risk factors reported among cases, during the 2- to 6-week period before onset of symptoms, included contact with a case (14.8%), being an employee or child in a daycare center (7.6%), exposure during a common-source outbreak (7.2%), using illicit drugs (4.3%), or men who have sex with men (3.9%). In 421 cases (36.4%), the case denied all the above risk factors.

Overall, specimens were available for 350 of the 1156 cases (30.3%), but availability varied by site (range, 16.6%-63.0%; Table 1). Of the 350 specimens, 271 (77.4%) were PCR positive and could be sequenced. There were differences in PCR positivity by age and risk factor but not by site, sex, or race. Genotypes among the 271 case specimens were IA (87.8%), IB (11.4%), and IIIA (0.7%) (Figure 3).

Travel-related cases were largely similar to strains from other countries or cases traveling to other countries in the historic reference library. A total of 131 PCR-positive specimens were observed among travel-related cases (ie, cases reported travel or were exposed to a traveler); 58 of these cases involved travel to Mexico (Table 2 and Figure 3, shown in red). Of the 58, 53 (91.4%) were within the US-IA₁ cluster. Similarly, 9 of 10 specimens (90%) from cases involving travel to Central America were within the US-IA₁ cluster. In contrast, 19 of 22 specimens (86.4%) from travel-related cases to South America were genotype IA but not IA_1-3 clusters. A total of 17 (13.0%) of all PCR-positive specimens from travel-related cases were genotype IB, 2 (1.5%) were genotype IIIA, and 30 (22.9%) were IA but were not within the IA_1-3 cluster.

Of 68 PCR-positive specimens with no known epidemiologic risk factor, 37 (54.4%) were similar to historical reference outbreak strains of genotype IA, cluster US-IA₁. All HAV strains from men who have sex with men were IA, almost exclusively found in the US-IA₃ cluster (Figure 3, shown in blue); in addition, 11 of 68 specimens with no reported risk factors matched these strains. There were 37 specimens from cases with more than 1 risk. These are illustrated in Figure 3 by a partitioned strip with more than 1 color.

The overall incidence of HAV disease during 2005 through 2007 in these sites was low (1.3 per 100 000 population), consistent with the rate in NNDSS for the same period (1.0-1.5 per 100 000).⁸ Rates varied moderately by site, from 0.7 to 2.3 per 100 000 population. In contrast, geographic variability was more extreme in the United States before the availability of the vaccine. For example, in 1995, the incidence of HAV disease ranged from 1.1 per 100 000 in Kentucky and New Hampshire to 86.7 per 100 000 in Oregon.⁸ One extreme observed in our data was the rate of 26.5 per 100 000 observed in a small county in Minnesota (population, approximately 9000 residents), where a foodborne outbreak of HAV was investigated. This illustrates how the epidemiologic pattern of HAV is in transition from outbreak-driven cases to only sporadic outbreaks in the United States. Continued disease surveillance is essential to controlling outbreaks and represents an important system needed to inform vaccination programs and target groups at highest risk of infection.¹³

The HAV vaccination strategy in the United States was initiated in 1996 with the Advisory Committee on Immunization Practices (ACIP) recommendation to vaccinate travelers among other groups at increased risk of disease.¹⁴ By 1999, there was epidemiologic evidence that the impact on national disease incidence was limited; thus, the vaccination strategy was expanded to include routine vaccination of children 2 years or older residing in areas where incidence was at least twice the US average.¹⁵ The impact of this expansion was dramatic; by 2003, the incidence rate of HAV disease had declined by 76% overall and by 87% among children aged 2 to 18 years.¹⁶ In 2006, the ACIP recommended routine, universal vaccination of children 12 months or older in the United States.¹⁰

Before routine vaccination, the epidemiologic patterns of HAV disease were well documented in 4 US counties participating in the Sentinel Counties Study of Acute Viral Hepatitis of the CDC. This study conducted enhanced case surveillance including patient interview and serologic testing using a rigorous standardized protocol from 1982 through 2006.¹⁷ Analyses of data from 1983 through 1995 demonstrated that HAV disease was transmitted largely by local, sustained outbreaks.¹⁷ Disease occurred primarily among persons aged 15 to 29 years. The most frequently reported risk factors were history of injecting drug use (14%) and sexual or household contact with an individual with HAV (12%). Risk factor information remained unknown in 52% of cases, and travel was identified as a risk factor in only 4% of cases.¹⁷

We document major, but not unexpected, differences in the epidemiologic pattern of HAV disease during 2005 through 2007 in participating sites. The most frequent potential source of infection among persons reported with HAV disease was travel. Our epidemiologic findings used patient self-report but were confirmed by genetic patterns indicating similarities to HAV strains available in the reference library from international travelers. While a departure from the 4% of travel-associated cases in the prevaccination era, our finding of 45.8% of travel-related HAV disease is consistent with surveillance in Switzerland, where between 2000 and 2004, 41.2% of cases reported a history of travel.¹⁸

This finding points to a prevention opportunity in the hands of health care providers to recommend pre-exposure vaccination to their patients planning travel to HAV endemic areas.

In the prevaccination era (before 1995), asymptomatic children were thought to be the source of infection among cases with unknown risk factors. For example, in the Sentinel Counties study, 33% of cases with unknown risk were exposed to children younger than 5 years.¹⁷ In NNDSS, risk factor information was not available for 50% of reported cases of HAV disease during 2007.⁸ In EIP sites, 36% of cases had no risk factor identified. Future studies should evaluate how to improve descriptions of risk patterns among HAV cases.

Molecular epidemiologic methods have been useful in understanding HAV transmission within networks of persons with similar risk factors.¹¹ When applied in combination with conventional epidemiologic methods, HAV sequencing has also been useful in the investigation and control of outbreaks, where it clearly augments and complements the epidemiologic information.¹⁹ However, for routine surveillance purposes, collecting specimens and conducting molecular characterization of HAV strains does not provide actionable information. The sequencing of specimens from EIP HAV cases during 2005 through 2007 confirmed existing data on risk factors but did not provide insight into transmission routes among cases with unknown risk factors. The intent of molecular characterization of specimens collected during routine surveillance may provide useful background data to be used in the event of an outbreak to discriminate outbreak-related strains from sporadic strains.⁹

Specimens from case patients in a variety of countries have revealed some consistency in the worldwide distribution of HAV genotypes. Among the specimens sequenced from the EIP sites, almost 90% were genotype IA, the most common genotype in the United States and worldwide; the remainder were largely genotype IB (less than 1% were genotype IIIB). Whereas the distribution of genotype IA includes Mexico and many Central and South American countries, the distribution of genotype IB has been limited to Brazil.¹² Consistent with our findings for travelers to Mexico, viral sequences from patients with a history of travel are closely related to the sequence patterns of isolates from countries where the travel occurred.¹²

In the United States, pre-exposure vaccination is recommended for travelers to areas of high or intermediate HAV endemicity. The risk of hepatitis A among susceptible travelers to developing countries has been estimated to range from 3 to 20 per 1000 travelers²⁰ depending on length of stay and region visited. Unfortunately, awareness of the risk among travelers appears to be low: in an airport survey, only 17% of US travelers to mostly developing countries identified hepatitis A as a risk at their destination.²¹ Furthermore, travelers' risk behaviors may vary by whether they are familiar with the area vs not. In an effort to measure the frequency of travel “home” (as opposed to tourism), we found that 86 of 108 cases reporting direct travel and known to have been born outside the United States had traveled to their country of birth. Overall, based on our finding that 45.8% of HAV cases may be travel related, enhanced implementation of existing recommendations for the prevention of HAV among travelers pre-exposure and postexposure could further reduce HAV disease in the United States.²² Identifying reasons for travel (eg, business, visiting family and relatives, tourism) among persons with HAV may be helpful in directing future vaccination initiatives.

We also found that 24% of cases reporting exposure to a traveler did not travel themselves. Investigations of HAV disease among contacts of international adoptees²³ prompted the ACIP to expand the recommendation for use of hepatitis A vaccine in February 2009 to include household members and other close personal contacts (eg, regular babysitters) of internationally adopted children from countries of high or intermediate HAV endemicity (http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5901a5.htm). We did not ascertain the number of cases from EIP sites that were contacts of adoptees; however, implementation of this expanded recommendation should further reduce the incidence of HAV disease in the United States.

This report is subject to several limitations. Incidence values represent a minimum estimate because the current surveillance case definition captures only persons with symptomatic illness and even symptomatic cases are underreported to local and state public health authorities.²⁴ We may have overestimated travel as a risk factor because it is easier for investigators to elicit travel history during interviews in contrast to, for example, drug use history. Furthermore, while cases may have reported travel, it is impossible to know whether they acquired their infection during travel. We do know that, compared with prevaccination dates, our finding of frequent reported travel among HAV cases does not reflect an underlying increase in travel in the population. In 1990, an estimated 19.7 per 100 000 US residents traveled internationally, whereas in 2007, 21.2 per 100 000 did so.²⁵ Specifically, the proportion of US residents traveling to Mexico decreased, from 36.6% of international travelers in 1990 to 30.4% in 2007.²⁵

Other limitations include a lack of complete geographic diversity and an overrepresentation of urban areas among the EIP sites. Also, the New York City data for the first 6 months of this report may have included cases that did not meet the CDC case definition; this is because, prior to July 2005, New York City classified all hepatitis A IgM reports as confirmed cases. Finally, specimens were a convenience sample of remnant serum specimens and collection rates varied by site, thus molecular data are not representative of the US population. However, even with these limitations, findings from hepatitis surveillance in the EIP are valuable to describe the epidemiologic pattern of viral hepatitis in the United States. For example, the frequency of unknown risk factors for HAV disease was lower in EIP cases (34%) than in NNDSS (50%) cases. In addition, EIP sites are representative of a larger US population (29.8 million) compared with the Sentinel Counties project (estimated 2008 population, 2.9 million).

Vaccination of children and high-risk groups in the US has reduced the incidence of HAV disease to record lows. Further reductions will depend on the continuation of routine, universal vaccination as well as health care providers encouraging vaccination of susceptible travelers.

Correspondence: R. Monina Klevens, DDS, Division of Viral Hepatitis, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention, 1600 Clifton Rd, MS G-37, Atlanta GA 30333 (rmk2@cdc.gov).

Accepted for Publication: April 6, 2010.

Author Contributions: Dr Klevens and Messrs Miller and Iqbal had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Klevens and Khudyakov. Acquisition of data: Klevens, Iqbal, Thomas, Rizzo, Hanson, Sweet, Phan, Cronquist, Khudyakov, and Xia. Analysis and interpretation of data: Klevens, Miller, Iqbal, Thomas, Hanson, Cronquist, Khudyakov, Xia, and Spradling. Drafting of the manuscript: Klevens, Iqbal, Thomas, Xia, and Spradling. Critical revision of the manuscript for important intellectual content: Klevens, Miller, Thomas, Rizzo, Hanson, Sweet, Phan, Cronquist, Khudyakov, and Spradling. Statistical analysis: Miller and Xia. Obtained funding: Klevens and Xia. Administrative, technical, and material support: Klevens, Iqbal, and Phan. Study supervision: Klevens, Thomas, Hanson, Khudyakov, and Spradling.

Financial Disclosure: None reported. All authors were employees of local or state government health departments or the CDC.

Funding/Support: Funding for this study was provided by the CDC to these health departments through a cooperative agreement in the Emerging Infections Program to conduct enhanced hepatitis surveillance.

Role of the Sponsor: The CDC funded the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript.

Previous Presentation: This study was presented in part as a poster at the 49th Interscience Conference on Antimicrobial Agents and Chemotherapy; September 12, 2009; San Francisco, California.

Additional Contributions: The following staff contributed to this report: Gary Budnick, MS, technical and material support, including submission of specimens; Terry Rabatsky-Ehr, MPH, Katherine Purviance, MPH, Jaime Krasnitski, MPH, and Jessica Brockmeyer, MPH, data collection, and Patricia Mshar, MPH, supervision, at the Connecticut Dept of health; Cynthia Kenyon, MPH, data collection at the Minnesota Department of Health; Kari E. Kufel, MPH, Tracey Knickerbocker, MA, and Kristen E. Mulhern, BA, collection of epidemiologic data and submission of serum specimens to the CDC Virology Laboratory at the New York State Department of Health; Susan Anderson, RN, data collection; Sharon Balter, MD, supervision; Katherine Bornschlegel, MPH, supervision; Paula Del Rosso, RN, data collection; Catherine M. Dentinger, FNP, MS, data collection; AnaMaria Fireteanu, MPH, data collection; Lara Kidoguchi, MPH, data analysis; Kristen Moore, MPH, data collection; Vasudha Reddy, MPH, data analysis; Kathryn Turner, RN, data collection; Eliza Wilson, BSMT, technical support; Melissa Wong, MPH, data collection; David Y. Yang, BS, technical support at the New York City Department of Health and Mental Hygiene; and Tasha Poissant, MPH, coordinating data collection at the Oregon Public Health Division. All acknowledged persons were employees of the health departments and did not have other sources of compensation.