|Home | About | Journals | Submit | Contact Us | Français|
Helicobacter pylori seroprevalence levels in US adults participating in the continuous National Health and Nutrition Examination Survey (1999–2000) increased with age in all racial/ethnic groups, with significantly higher age-standardized levels in Mexican Americans (64.0%, 95% confidence interval (CI): 58.8, 69.2) and non-Hispanic blacks (52.0%, 95% CI: 48.3, 55.7) compared with non-Hispanic whites (21.2%, 95% CI: 19.1, 23.2). Although seroprevalence levels remained similar to those found in National Health and Nutrition Examination Surveys from 1988 to 1991 among non-Hispanic blacks and Mexican Americans, they were significantly lower in non-Hispanic whites, especially at older ages. The factors driving the decline in H. pylori seroprevalence appear to be acting preferentially on the non-Hispanic white population.
In the developed world, the prevalence of Helicobacter pylori infection has declined over the past 60 years (1, 2). The drop in seroprevalence in developed countries has been attributed to a combination of factors, including greater antibiotic use and improved sanitation and living conditions (3, 4).
The seroprevalence of H. pylori in the United States has been estimated at 32.5% among adults aged 20 years or older on the basis of analysis of the Third National Health and Nutrition Examination Survey (NHANES III) performed in 1988–1991 (5). Cross-sectional studies using NHANES III have shown that demographic factors such as socioeconomic status and race/ethnicity are significant predictors of seroprevalence to multiple pathogens in the US population, including H. pylori (5, 6). Both non-Hispanic blacks and Mexican Americans carry a significantly higher prevalence of H. pylori than do non-Hispanic whites (5, 7, 8), a finding that persists even after adjustment for country of origin and socioeconomic status (5). It is unclear whether the disparities in H. pylori seroprevalence identified in cross-sectional studies of NHANES III are continuing in the United States or have been reduced as a consequence of increased antibiotic use and treatment and/or further improvements in living conditions over the past decade.
The purpose of this study was to examine whether racial/ethnic disparities in H. pylori seroprevalence have persisted in the United States. To do so, we compared the dynamics of H. pylori infection by age and race/ethnicity in the National Health and Nutrition Examination Survey (1999–2000) (hereafter referred to as “NHANES 1999–2000”) with those in NHANES III from 1988 to 1991.
The analyses used data from the first phase of the National Health and Nutrition Examination Survey performed in 1988–1991 (NHANES III) and the results from NHANES 1999–2000, conducted by the National Center for Health Statistics. Both surveys are cross-sectional, stratified, multistage probability samples of the civilian noninstitutionalized US population, with oversampling of Mexican-American and non-Hispanic black respondents. Information on demographic characteristics was collected by using in-person interviews. All participants or their proxy respondents gave written, informed consent. Data were collected in household interviews, clinical examinations, and laboratory tests. Details of the sampling design and protocol have been described (9, 10). The survey protocol was approved by the Institutional Review Board of the Centers for Disease Control and Prevention (Atlanta, Georgia).
The first phase of NHANES III and NHANES 1999–2000 are the only surveys that include laboratory data on H. pylori seropositivity status. H. pylori seropositivity was determined by using a commercially available enzyme-linked immunosorbent assay (ELISA) (Wampole Laboratories (now Inverness Medical), Princeton, New Jersey). Of the 4,880 participants aged ≥20 years enrolled in NHANES 1999–2000, H. pylori status had been determined in 4,145 (84.9%). For each specimen, an immune status ratio was calculated by dividing the optical density of the specimen by the mean optical density of the cutoff controls. Following past analyses (5), specimens were treated as negative for H. pylori if the immune status ratio was within the range of negative (0–0.90) or equivocal (0.91–1.09) and positive if ≥1.10. Of the 7,465 participants aged ≥20 years for whom H. pyolri status was available in NHANES III, 207 (2.8%) were equivocal and, in NHANES 1999–2000, 113 (2.7%) were equivocal. Because data using the Wampole Laboratories’ ELISA assay were available only for those 20 years of age or older in NHANES III and performance characteristics of the assay have not been established in pediatric populations (11), the analyses presented here focus on adults in the US population.
Race/ethnicity was categorized in NHANES III as non-Hispanic white, non-Hispanic black, Mexican American, and all others. In NHANES 1999–2000, ethnicity was coded in a compatible format, with the addition of an “other Hispanic” category. Individuals identified as members of the “other” and “other Hispanic” ethnicity were included in the estimates for the total population but were not used in other statistical comparisons because few individuals were in these categories. Age in years was grouped by decade as a categorical variable. Education was divided into 4 categories: no high school education, some high school education, completion of high school, and at least some college education. The income/poverty ratio, a ratio of income to the family’s poverty threshold based on US Census data, was divided into 3 categories: below or at the official poverty threshold, from just above the poverty threshold to 3 times the poverty threshold, and above 3 times the poverty threshold.
All statistical analyses were conducted by use of SAS, version 9.2, PROC SURVEY procedures (SAS Institute, Inc., Cary, North Carolina) and accounted for the complex sampling design in NHANES. Sampling errors were estimated by using the primary sampling units and strata provided in the data and calculated through Taylor series linearization. Sampling weights were used to adjust for nonresponse and oversampling in NHANES. The PROC SURVEYFREQ command was used to determine seropositivity by age group. Prevalence estimates (via the PROC SURVEYREG command) that compared population subgroups were standardized by age to the 2000 US Census standard for both NHANES III and NHANES 1999–2000 to adjust for any potential differences in the age structure between these 2 time periods. Logistic regression modeling (via the PROC SURVEYLOGISTIC command) was conducted on the total population and on non-Hispanic white, non-Hispanic black, and Mexican-American racial/ethnic groups to estimate odds ratios comparing demographic factors and corresponding 95% confidence intervals.
In NHANES 1999–2000 data, the age-standardized proportion seropositive for H. pylori among individuals older than 20 years was 30.7% (95% confidence interval (CI): 27.9, 33.6), compared with the value from NHANES III of 34.0% (95% CI: 30.8, 37.1) (5). The difference between these proportions is 3.3% (95% CI: −0.6, 7.2). As in NHANES III, Mexican Americans and non-Hispanic blacks had a higher seroprevalence than did non-Hispanic whites, with an age-standardized prevalence of 21.2% (95% CI: 19.1, 23.2) in non-Hispanic whites, compared with 52.0% (95% CI: 48.3, 55.7) in non-Hispanic blacks and 64.0% (95% CI: 58.8, 69.2) in Mexican Americans. Logistic regression models evaluating seropositivity as a function of the 2 surveys indicated that seroprevalence levels were significantly higher in 1988–1991 than in 1999–2000 in non-Hispanic whites (odds ratio (OR) = 1.38, 95% CI: 1.15, 1.64) and did not change significantly in non-Hispanic blacks (OR = 1.04, 95% CI: 0.85, 1.28) and Mexican Americans (OR = 0.91, 95% CI: 0.70, 1.20).
Comparison of H. pylori immunoglobulin G seroprevalence between the 2 data sets by age group demonstrates declining seroprevalence in older age (Figure 1A; Table 1). Seroprevalence in the total population was significantly higher in 1988–1991 compared with 1999–2000 in the 60–69 and ≥70 year age groups (Table 1). A model to evaluate the change in seroprevalence in the birth cohort from 1930 to 1939 over the nearly 10 years between surveys indicated a nonsignificant trend toward a decline in seroprevalence (OR = 0.79, 95% CI: 0.60, 1.03).
Stratifying by race/ethnicity and age demonstrates that the observed population-wide decline in seropositivity in the ≥60-year age groups is primarily due to a decline in seroprevalence in non-Hispanic whites (Figure 1B; Table 1). In contrast, across all age groups, seroprevalence levels in non-Hispanic blacks and Mexican Americans are not significantly lower (Figure 1, C and andD;D; Table 1), with the exception of the age group 20–29 years in non-Hispanic blacks. Tracking non-Hispanic white cohorts born approximately in 1930–1939 and 1940–1949 also shows that these groups have a trend toward decline in seroprevalence, but that the decline is not statistically significant (OR = 0.74, 95% CI: 0.54, 1.02 and OR = 0.73, 95% CI: 0.51, 1.04, respectively).
We used a multivariate logistic model to assess the impact of race/ethnicity after adjusting for age, educational level, country of birth, and the income/poverty ratio (Table 2). The odds ratios indicate that non-Hispanic black and Mexican-American races/ethnicities have a significantly higher odds of H. pylori infection even after adjustment for age, socioeconomic factors, and country of origin.
The incidence of infection by age can be roughly estimated from cross-sectional surveys with age-stratified seroprevalence data, assuming persistence of immunity and immune markers (12). Under this assumption, the seropositivity by age curve (Figure 1) for non-Hispanic blacks and Mexican Americans suggests a high incidence in youth and early adulthood followed by a significant decline in incidence with age.
H. pylori seroprevalence levels in 2 large surveys of nationally representative populations separated by approximately 10 years remained similar in the overall population but declined significantly in older age groups. This age-related decrease is primarily accounted for by lower levels of seroprevalence in non-Hispanic whites. The seroprevalence levels in non-Hispanic blacks and Mexican Americans showed no decline between the 2 surveys, with the exception of the 20–29 year category in non-Hispanic blacks. In this age group, seroprevalence declined from 37.8% to 23.1%, a decline that may reflect stochastic variation or a decreasing childhood incidence of H. pylori. The seroprevalence levels in 3–9 year and 10–19 year non-Hispanic blacks in the NHANES 1999–2000 data, however, suggest that there has not been a persistent decline in childhood incidence in this population (Figure 1C); we make this interpretation with caution, as we note that performance characteristics of the Wampole H. pylori ELISA assay used in determining seropositivity have not been established in pediatric populations (11). Consistent with a prior evaluation of race/ethnic disparities in H. pylori seroprevalence (5), the disparity between non-Hispanic whites and non-Hispanic blacks and Mexican Americans is not fully explained by the available indicators of socioeconomic status in NHANES. Therefore, other factors, not available in these data, may account for the disparities by race/ethnicity.
The decline in seropositivity between NHANES III and NHANES 1999–2000 in non-Hispanic whites at older ages has several potential explanations. As a formal possibility, the decline could be explained by an increase in mortality rates of individuals with H. pylori seropositivity. However, mortality rates due to diseases in which H. pylori is a known or strongly suspected cause, such as peptic ulcer disease and gastric cancer, declined in the 1990s (13, 14). Moreover, one would expect to see a similar phenomenon in the black non-Hispanic and Mexican-American populations leading to the same or even more profound reduction than in whites because of the high prevalence of this infection in these racial/ethnic groups.
Another explanation for lower age-specific seroprevalence levels at older ages in non-Hispanic whites is increased seroreversion, a possibility directly suggested by the trend toward lower seropositivity within the 1930–1949 birth cohorts in NHANES 1999–2000 compared with NHANES III. Successful H. pylori treatment can result in seroreversion (i.e., antibody levels fall below the clinical cutpoint for seropositivity) in approximately 50% of individuals at 12 months after eradication (15). To determine whether treatment alone could explain the observation, we assume a 50% seroreversion rate after eradication of H. pylori infection. If the incidence of new infection stayed the same in the period between NHANES III and NHANES 1999–2000, the decline from 45.3% to 29% in the 60–69 year age group and from 54.8% to 38.1% in the ≥70 year age group in non-Hispanic whites implies H. pylori diagnosis and treatment in roughly 32% of this population. Because most people infected with H. pylori do not develop symptoms (16), we conclude that it is unlikely that seroreversion from directed treatment alone accounts for the large decline in seroprevalence that we have observed among Non-Hispanic whites. Other potential reasons for greater seroreversion, such as incidental antibiotic-associated clearance, or a decrease in immunoglobulin G “boosting” phenomena, such as repeated exposures, are difficult to estimate without detailed longitudinal cohort studies.
A third possible explanation is a decreasing rate of seroprevalence in childhood, a birth cohort phenomenon that has been invoked to explain similar epidemiologic findings for H. pylori seroprevalence (4). Under this hypothesis, H. pylori acquisition takes place predominantly in childhood; decreasing childhood incidence over time would then account for lower age-specific seroprevalence levels in successive cohorts. However, longitudinal cohort studies demonstrating rates of seroreversion that are higher than seroconversion indicate that birth cohort effects cannot be the complete explanation for the observed pattern (17).
The results presented in these analyses demonstrate persistent racial/ethnic disparity in H. pylori infection prevalence, and they show a declining prevalence in non-Hispanic whites that has not been shared by non-Hispanic blacks and Mexican Americans. The reasons for the declining prevalence are not clear but appear to be a combination of declining incidence during the earlier years of life among non-Hispanic whites and factors that promote seroreversion, such as H. pylori-directed treatment or other incidental antibiotic use that may lead to seroreversion. Conversely, the factors that result in sustained seroprevalence in non-Hispanic blacks and Mexican Americans, which may include acquisition of infection earlier in life, remain uncertain, highlighting the importance of understanding more about how H. pylori infection is transmitted and about lifetime rates of seroconversion and seroreversion in these groups. Further studies to determine whether disparities in H. pylori seroprevalence continue to persist and longitudinal cohort studies that examine specific risk factors for infection and define conditions that favor seroreversion are needed.
Author affiliations: Division of Infectious Diseases, Brigham and Women’s Hospital, Boston, Massachusetts (Yonatan H. Grad); Center for Communicable Disease Dynamics, Department of Epidemiology, Harvard School of Public Health, Harvard University, Boston, Massachusetts (Yonatan H. Grad, Marc Lipsitch); Department of Immunology and Infectious Diseases, Harvard School of Public Health, Harvard University, Boston, Massachusetts (Marc Lipsitch); and Department of Epidemiology, Center for Social Epidemiology and Population Health, University of Michigan School of Public Health, Ann Arbor, Michigan (Allison E. Aiello).
The project was supported by the National Institute of General Medical Sciences (U54GM088558 to M. L.), the National Institute of Allergy and Infectious Diseases (T32 grant AI007061 supporting Y. H .G.), the National Institutes of Health (R21 NR011181–01 to A. E. A.), and Harvard University (Yerby Visiting Faculty Award to A. E. A.).
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of General Medical Sciences or the National Institutes of Health.
Conflict of interest: none declared.