Age-Specific Incidence of Helicobacter pylori

GASTROENTEROLOGY 2006;130:65–72

Age-Specific Incidence of Helicobacter pylori MARION ROWLAND,* LESLIE DALY,‡ MARIAN VAUGHAN,* ANNA HIGGINS,* BILLY BOURKE,*,§,㛳 and BRENDAN DRUMM*,§,㛳 *University College Dublin School of Medicine and Medical Science, The Children’s Research Centre, Dublin, Ireland; ‡University College Dublin School of Public Health Medicine and Population Science, Dublin, Ireland; §University College Dublin Conway Institute of Biomolecular Technology, Dublin, Ireland; and 㛳Our Lady’s Hospital for Sick Children, Crumlin, Dublin, Ireland

Background & Aims: Helicobacter pylori is most likely acquired in childhood, but the incidence of infection has not been determined prospectively by using an appropriate noninvasive test. The aim of this study was to determine the age-specific incidence of Helicobacter pylori infection in children and the risk factors for infection. Methods: Three hundred twenty-seven healthy index children between 24 and 48 months of age were enrolled over 15 months. At baseline, the Helicobacter pylori infection status of each index child and his or her older siblings and parents was assessed by using the carbon 13– urea breath test. All noninfected index children were then followed up with an annual carbon 13– urea breath test for 4 years to determine whether they became infected with Helicobacter pylori and, if so, the age at first infection. Information on potential risk factors was collected at baseline and each subsequent visit. Results: At baseline assessment, 28 of 327 (8.6%) index children were infected with Helicobacter pylori. The mean age of the 28 infected children was 32.78 months (SD, 5.14 months). Over the next 4 years, 279 index children not infected at baseline contributed 970 person-years of follow-up to the study. During this time, 20 children became infected with Helicobacter pylori. The rate of infection per 100 person-years of follow-up was highest in the 2–3-year age group (5.05 per 100 person-years of follow-up (95% confidence interval, 1.64 –11.78) and declined progressively as children aged. Only 1 child became infected after 5 years of age. Having an infected mother, an infected older sibling, and delayed weaning from a feeding bottle (ie, after 24 months of age) were all risk factors for infection. Conclusions: Children who become infected with Helicobacter pylori are infected at a very young age, and the risk of infection declines rapidly after 5 years of age. These findings have important implications for studies on the mode of transmission of infection.

he organism Helicobacter pylori causes antral gastritis and is associated with duodenal ulcer disease in adults1–3 and children.4 Eradication of the organism leads to long-term healing of duodenal ulcer disease.5– 8 The World Health Organization has classified H pylori as a

T

group 1 carcinogen for gastric adenocarcinoma.9 Although it is accepted that H pylori infection is most likely acquired in childhood,10 –12 the age group at greatest risk of infection is unclear, and the mode of transmission is unknown. Studies on the incidence of H pylori infection are important for a number of reasons. Although we know that H pylori infection is clustered in families,13,14 we do not know whether this is because of person-toperson transmission or acquisition from a common environmental source.15 Studies to identify the mode of transmission are more likely to succeed if we can first identify the specific age group or groups at greatest risk of H pylori infection. Identifying the age at which H pylori infection is most likely to occur and the specific risk factors for infection will further the development of appropriately targeted preventative strategies. Enteric infections are most common in children, as evidenced from the burden of morbidity and mortality associated with these infections worldwide. Two early studies suggested that the incidence of H pylori infection in children was low,16,17 whereas more recent studies have suggested widely varying incidence rates from 1.7% to 15%.18 –23 However, most of these studies have been performed by using stored serum samples. The results of such serological studies in children are difficult to interpret because commercial serological enzyme-linked immunosorbent assays fail to diagnose a significant proportion of infected children aged ⬍10 years.24 –27 Because the serum samples were collected for other research protocols, data for specific risk factors for H pylori were not available. Similarly, the use of polyclonal stool antigen is unreliable in children,28 and the newer monoclonal stool antigen tests have not been validated to date for use in young children. Furthermore, the short follow-up in Abbreviations used in this paper: CI, confidence interval; 13C-UBT, carbon 13– urea breath test. © 2006 by the American Gastroenterological Association 0016-5085/06/$32.00 doi:10.1053/j.gastro.2005.11.004

66

ROWLAND ET AL

most studies precluded any determination of the age group at the greatest risk of infection. This is the first study specifically designed to study the acquisition of H pylori infection in children. The prospective nature of this study allowed us to use the carbon 13– urea breath test (13C-UBT) to determine the incidence of H pylori in children. In contrast to serological assays, the 13C-UBT has been shown to have excellent sensitivity and specificity for the noninvasive identification of H pylori infection in children ⬎2 years of age.29 –33 However, the accuracy of the 13C-UBT is reduced in children ⬍2 years of age as a result of the interference of oral urease–producing organisms.34 –36 The aim of this study was to prospectively follow up a group of healthy children in a developed country to determine the agespecific incidence of H pylori infection and the risk factors associated with infection.

Methods Three hundred twenty-seven healthy index children between 24 and 48 months of age were enrolled. Index children were enrolled over a 15-month period (October 1997 to December 1998). At baseline, the H pylori infection status of each index child was assessed by using the 13C-UBT. All noninfected index children were then followed up with annual 13C-UBT for 4 years to determine whether they became infected with H pylori and, if so, the age at first infection. Information on potential risk factors for H pylori infection was collected at baseline and each subsequent visit. The infection status of parents and siblings (nonindex children) was determined at baseline.

Source of Participants Nineteen family doctors were approached to provide patients for the study, and 17 participated. Nine of 15 practices were located in city areas with a high density of government housing and young families. Parents of eligible children were invited by letter from their family doctor to participate in this study. Opportunistic recruitment of frequent attendees was not possible because the 13C-UBT protocol required a 2-hour fast. Children with conditions such as cystic fibrosis and cerebral palsy were not included in the database of eligible children, but common childhood illnesses were not exclusion criteria. In 6 families in which there were 2 children in the target age group at enrollment, the younger child was enrolled as the index child. In the case of twins (six pairs), the second twin was considered the index child, and the first twin was considered an older sibling. None of these 12 index children, their twins, or their siblings became infected with H pylori during the study.

Annual Breath Tests The 13C-UBT was performed after a 2-hour fast on all index children and siblings ⬍18 years of age.29 Breath samples

GASTROENTEROLOGY Vol. 130, No. 1

were collected at baseline and 30 minutes after ingestion of 50 mg of 13C-urea (75 mg in children that weighed ⬎50 kg). A cutoff of 5.0 ␦ 13CO2‰ was used to classify participants as infected with H pylori. Breath tests for parents and siblings were performed as described by Braden et al.37

Criteria for the Diagnosis of Helicobacter pylori Infection To ensure that index children who were infected with H pylori were classified correctly, the following protocol was adopted. All index children who had a positive or borderline (2.5–5.0 ␦ 13CO2‰) 13C-UBT had a second confirmatory test. The second test was performed after an overnight fast, because the discriminatory power of a 13C-UBT is greater after an overnight fast.29 Only index children in whom a second confirmatory 13C-UBT was positive were considered infected with H pylori. Children who were classified as infected with H pylori did not have any further breath tests until the final year of the study. Children with 1 positive or borderline test followed by a negative fasting confirmatory test were classified as noninfected and continued to have annual breath tests. This strict criterion for the classification of an index child as infected was adopted to ensure that no index child was misclassified because of a false-positive13C-UBT. Index children with initial negative tests were classified as noninfected and continued to have annual breath tests.

Risk Factor Assessment Family members. The H pylori status of family members—parents and siblings—was established to determine whether having an infected family member was a risk factor for infection. Only older siblings had breath tests, because the index child was the youngest child in each family eligible for breath testing. Other risk factors. Information on socioeconomic status, including income based on entitlement to free health care services38; child-rearing practices, including breast-feeding, bottle feeding, and pacifier use; usual sleeping arrangements; and crèche and preschool use, was collected on each index child at baseline and at annual follow-up visits.

Statistical Analysis A sample size of 200 children was calculated based on a requirement of 180, allowing for a 10% rate of loss to follow-up. With 180 subjects, the cumulative incidence of infection at age 5 years (judged to be of the order of 30%39) would be detected within approximately ⫾7% on the basis of a 95% confidence interval (CI). The average annual incidence would be detected to within ⫾2%. In addition, for elucidation of risk factors for infection, the sample size had an 80% power to detect, at a 2-sided 5% level of significance, risk factors for infection that had a relative risk of ⱖ1.9 between groups exposed or not exposed to a putative risk factor. This is on the assumption of an overall 5-year incidence of 30% and a prevalence of 15%– 60% for the risk factor. Because of a substantial decline in the cost of the 13C-UBT, it was possible to enroll

January 2006

Figure 1. Flow-chart of children enrolled in the study.

327 children in the study, and this increased the precision of the sample from ⫾7% to ⫾5%. Index children who had a positive and a confirmatory 13CUBT at baseline were categorized as infected with H pylori. These children (prevalent cases) were not included in the calculation of the age-specific incidence. Index children not infected at baseline had their H pylori status assessed annually over 4 years to give the age-specific incidence of infection. In the analysis of age-specific infection rates, person-years of follow-up were calculated as described by Breslow and Day.40 The 4-year observation period was divided into 1-year age intervals determined by chronological age. The number of children not infected at the start of each interval, the number who became infected, and the number who were lost to follow-up were used to estimate the rate of infection for each age interval. The age-specific rate of H pylori infection was calculated as the number of new cases of infection in index children in the specific age interval divided by the number of person-years of follow-up contributed by all children at risk of infection in this interval. Confidence intervals (95%) for the incidence rate per 100 person-years of follow-up were calculated by using a Poisson distribution for the observed number of incident cases.41 Index children who had a positive 13C-UBT (and a confirmatory test) at any time were deemed to have become infected 6 months before the date of the first positive test, and person-years of follow-up were calculated accordingly. Person-years of follow-up were counted up to the date of the test for negative tests. Because the incidence of new infection was low, we analyzed risk factors for H pylori infection in all 290 index children by combining prevalent cases and incident cases and comparing them with children who did not become infected with H pylori over the study period (Figure 1). Risk factor analysis was performed on the basis of a nested case-control approach, because 58% of infected children were infected at baseline, and

AGE–SPECIFIC INCIDENCE OF HELICOBACTER PYLORI 67

a further 25% of infected children were infected in the first year of follow-up. Index children who had no eligible sibling tested for H pylori were excluded from analysis of siblings as a risk factor for H pylori infection. Inclusion or exclusion of index children for whom no eligible sibling in the family was tested did not materially alter the risk associated with other factors in the multivariate analysis. Only index children in 2-parent families were included in the univariate analysis of father as a risk factor for infection. Because ⬍60% of fathers had breath tests performed, this variable was not analyzed in any logistic regression model. Categorical risk factors were examined by using odds ratio with 95% CIs, and continuous variables were compared by using t tests. Stratified Mantel–Haenzel analysis and multiple logistic regression were used to examine the simultaneous effect of several different explanatory variables on the risk of H pylori infection. All explanatory variables in the logistic regressions were binary.41 In stratified analysis, there was an interaction between having an infected mother and using 2 or more bottles per day; therefore, an interaction term was included in the final logistic regression model. Adjusted odds ratios, together with their 95% CIs, were derived from the logistic regression coefficients. The overall significance level was set at 5%.

Ethical Approval Ethical approval was obtained from the Ethics Committee of Our Lady’s Hospital for Sick Children, Dublin. Signed informed consent was obtained from the mother or father of all index children enrolled in the study. Separate consent was also obtained from the mother or father for his or her own 13C-UBT and for siblings ⬍18 years of age. Informed signed consent was obtained from siblings ⬎18 years of age. An appropriate verbal assent was obtained from index children and their siblings.

Results Three hundred twenty-seven index children were enrolled in the study from 15 family practices. Seventyfive percent of children resided in the greater Dublin city area, and the remainder lived in 2 rural towns (Mallow, Co Cork, and Kingscourt, Co Cavan, Ireland). After baseline assessment, 37 (11.3%) noninfected index children were lost to follow-up over the subsequent 4 years (Figure 1). Twenty of these 37 were lost between baseline and the date they were due to have their first annual follow-up. The remaining 17 contributed several person-years to the follow-up but did not complete the study. Index children were lost to follow-up because they had moved and could not be contacted, changed doctors, or did not wish to participate further in the study. The mean age of the 327 index children enrolled in the study was 33.25 months (SD, 7.24 months; range, 23– 48 months), and 45% were female.

68

ROWLAND ET AL

GASTROENTEROLOGY Vol. 130, No. 1

Table 1. Age-Specific Rate of Infection for a Cohort of 279 Children Not Infected With Helicobacter pylori at Baseline Age interval (y) Variable

2–3

3–4

4–5

5–6

6–7

7–8

Children at risk Person-years of follow-up New H pylori infection Rate/100 person-years 95% CI

181 99 5 5.05 1.64–11.78

256 214 9 4.20 1.92–7.98

256 241 5 2.07 0.67–4.84

230 224 0 0 0–1.64

200 146 1 0.68 0.02–3.81

85 46 0 0 0–8.02

NOTE. Only 1 child became infected after the age of 5 years, during 416 person-years of follow-up (0.2 per 100 person-years), compared with 19 younger than 5 years, with 554 person-years of follow-up (3.4 per 100 person-years; relative risk, 17.0).

Prevalence of Helicobacter pylori at Baseline Assessment At baseline assessment, 28 of 327 (8.6%) index children were infected with H pylori. The mean age of the 28 infected children was 32.78 months (SD, 5.14 months). Twenty-three of those infected were ⬍3 years of age, and the remaining 5 were ⬍4 years of age. Twenty-six of the 28 (92.8%) children (2 lost to followup) were retested after 4 years, and all children were still categorized as infected with H pylori. Age-Specific Rate of Helicobacter pylori Infection Over 4 years of follow-up, 279 index children not infected at baseline contributed 970 person-years of follow-up to the study. During this time, 20 children became infected with H pylori. The number of children who became infected for each 1-year age interval is presented in Table 1. The rate of infection per 100 person-years of follow-up was highest among 2–3-year-olds (5.05 per 100 person-years of follow-up; 95% CI, 1.64 – 11.78) and declined progressively as children got older. Risk Factor Analysis Only information on risk factors recorded at baseline were analyzed, because 40 of 48 children were infected within the first year of follow-up (prevalent and incident cases). Therefore, only factors present at baseline could be causally related to infection. The variables assessed as risk factors for H pylori infection are outlined in Table 2. As has been shown previously, all measures of socioeconomic status, including income, maternal education (⬍11 years), overcrowding, and bed sharing with the mother, were risk factors for infection. The overall rate of breast-feeding, although low by international standards, at 35%, was consistent with the rates in the Irish population. There was no difference in the rate or duration of breast-feeding between infected and noninfected index children. At baseline, 171 (58.9%) index children had completely finished

using a feeding bottle, 29 (10%) used ⱕ1 bottle per 24 hours (usually at night), and 90 (31.0%) continued to have ⱖ2 bottles per day. Having ⱖ2 bottles per day was a statistically significant risk factor for infection (Table 2). Pacifier use was not a risk factor for infection (Table 2), and there was no difference in the mean duration of pacifier use at baseline between the 2 groups. Parents There were 234 (80.6%) 2-parent families and 39 (13.4%) single-parent families, and in 17 (5.86%) families the mother was separated, divorced, or widowed. Ninety-eight percent of mothers had breath tests performed, and 45.1% (129/286) were infected with H pylori. The odds ratio for having an infected mother was 5.26 (95% CI, 2.39 –12.45; P ⬍ .001; Table 2). Ten of 48 mothers of infected index children were not infected with H pylori, and none of these reported having had H pylori eradication therapy in the past. The 4 mothers who were not tested had H pylori–infected index children. One hundred thirty-six fathers (58.1%), in 2-parent families, were tested for H pylori, and 69 were infected with H pylori. The odds ratio for having an infected father was 5.39 (95% CI, 1.50 –19.35; P ⫽ .004; Table 2). Siblings At baseline, the mean number of siblings per family was 1.87 (SD, 2.39) in infected index cases and 1.54 (SD, 1.27) in noninfected index cases (mean difference, 0.33; 95% CI, 0.12– 0.78). At baseline, 52 index children had no siblings (only children), 26 had only younger siblings (who could not be reliably assessed for H pylori by using the 13C-UBT because they were ⬍2 years of age), and 190 index children had at least 1 older sibling tested. Twenty-two (7.6%) index children had no eligible sibling tested. Having an infected older sibling at baseline increased the risk of H pylori infection (odds ratio, 7.30; 95% CI, 3.50 –15.23; P ⬍ .001; Table 2). Thirty-one percent of infected index children were first-born, whereas 27.3% of noninfected index children

January 2006

AGE–SPECIFIC INCIDENCE OF HELICOBACTER PYLORI 69

Table 2. Univariate Analysis of Risk Factors for Helicobacter pylori in 48 Infected (Index Hp⫹) and 242 Noninfected (Index Hp⫺) Index Children Factor Breast-fed ⱖ2 bottles/day Pacifier use 3 Portions fruit and vegetables/day Share bedroom Share bed Alone Mother ⬎2 nights/week Mother always Socioeconomic group 1 2 3 4 5 6 Income ⬍£8320/c Urban dwelling Crowding index ⬎1 Mother ⬍11 y education Marital status; single parent Day care Pets Mother infected Father infected Any infected siblingd

Index Hp⫹, n (%)a

Index Hp⫺, n (%)

Odds ratiob

95% CI

12 (25) 21 (43.75) 32 (66.6) 29 (60.4) 32 (66.66)

92 (38.0) 69 (28.5) 160 (66.1) 101 (41.7) 131 (54.13)

0.54 1.95 0.93 2.13 1.69

0.26–1.09 1.03–3.6 0.46–1.87 1.08–4.24 0.88–3.25

.086 .037 .834 .017 .115

21 (43.7) 14 (29.16) 13 (27.08)

143 (59.09) 62 (25.62) 37 (15.28)

1.00 1.19 2.05

0.60–2.37 0.99–4.25

.048

2 (4.1) 5 (10.4) 9 (18.75) 10 (20.8) 9 (18.75) 13 (27.08) 27 (56.2) 41 (85.4) 13 (27.7) 25 (52.1) 14 (29.2) 16 (33.3) 14 (29.16) 34 (77.3) 15 (75.0) 20 (47.6)

23 (9.5)} 46 (19.0) 69 (28.5) 34 (14.0) 34 (14.0)} 24 (9.9) 76 (31.4) 165 (68.18) 30 (12.4) 63 (26.2) 25 (13.4) 72 (29.75) 90 (37.19 95 (39.2) 54 (46.5) 25 (11.1)

1.0

2.41 2.80 2.73 2.70 3.05 3.57 1.18 0.69 5.26 5.39 7.30

1.27–4.58 1.49–5.28 1.10–7.07 1.28–5.69 1.61–5.76 1.69–7.54 0.57–2.41 0.33–1.42 2.39–12.45 1.50–19.35 3.50–15.23

P value

.006 .001 .016 .007 ⬍.001 ⬍.001 .623 .661 ⬍.001 ⬍.004 ⬍.001

are out of the total in Hp⫹ and Hp⫺. binary variables, odds ratios are relative to those in the other category, which has an odds ratio of 1.0. cBased on income guidelines in Irish pounds for free health care from the Department of Health under the General Medical Scheme in 1997 for a 2-parent family with 1 child (equivalent US$11,373 in 1997).38 dTwenty-two (7.6%) index children were excluded because no eligible sibling was tested for H pylori. aPercentages bFor

were first-born. Increasing birth order was not a risk factor for H pylori infection (␹21 for trend ⫽ 2.68; P ⫽ .101; data not shown). Multivariate Logistic Regression Joint effects of risk factors were examined initially by using a stratified approach, which was then followed by logistic regression analysis that included interaction terms. Table 3 shows the multivariate results that best explain the interrelationships.

In the model, all 5 factors were included simultaneously. The multivariate odds ratios are those for each factor in the presence of the other 4. Univariate odds ratios are given for comparison but differ from those in Table 2 because of a decrease in numbers: 22 (7.6%) index children were excluded because they had no eligible sibling tested for H pylori. Because there was an interaction between having ⱖ2 bottles per day and an infected mother in the stratified analysis, an interaction term was included (infected mother with ⬎2 bottles per

Table 3. Univariate and Multivariate Analysis of Risk Factors for 268 Index Children: An Interaction Term for Infected Mother and Using 2 or More Bottles per Day Is Included Factor present

Univariate odds ratioa

Multivariate odds ratio

P value

Infected mother Infected sibling ⱖ2 bottles/day Always sleep with mother Income ⬍£8320/y Interaction term: mother infected/2 bottles/day

5.40 (2.45–11.88) 7.30 (3.50–15.23) 1.89 (0.96–3.73) 2.07 (0.95–4.53) 2.50 (1.28–4.88)

7.20 (1.97–26.37) 4.90 (2.15–11.55) 5.01 (1.15–21.8) 1.00 (0.38–2.59) 1.3 (0.59–2.6) 0.29 (0.05–1.66)

.002 .002 .03 .99 .49 .165

aUnivariate odds ratios differ from those in Table 2 because 22 (7.6%) index children who had no eligible sibling tested at baseline were excluded from the multivariate analysis.

70

ROWLAND ET AL

day). Maternal education and overcrowding were examined but were nonsignificant risk factors and, therefore, were not included in the final model. Low socioeconomic status, as measured by eligibility for free health care (income ⬍£8320/y), was included in the final model because it was the most objective measure of poor socioeconomic status. Neither annual income ⬍£8320 or bed sharing was a risk factor for H pylori infection when the other factors were accounted for. Having an infected mother and an infected sibling and having ⱖ2 bottles per day at baseline were statistically significant risk factors for infection in this model.

Discussion This is the first prospective study to examine the age-specific incidence of H pylori in children. This study was specifically designed to examine the incidence of H pylori in children ⬎24 months of age by using a valid diagnostic test with strict a priori criteria for the classification of H pylori infection status. Information on risk factors was collected when or before infection occurred. The results show that children in a developed country who become infected with H pylori are infected at a very young age. Overall, 47 of 48 H pylori–infected children became infected before the age of 5 years. In the agespecific analysis, only 1 child became infected after the age of 5 years during 416 person-years of follow-up (0.2 per 100 person-years of follow-up), compared with 19 under the age of 5 years with 554 person-years of follow-up (3.4 per 100 person-years; relative risk, 17.0). The highest incidence of H pylori infection in this study was in the youngest children (2–3-year age group). Because of the limitations of the 13C-UBT in very young children, the incidence of infection could not be evaluated in children ⬍2 years old. The rigorous classification of H pylori infection used in this study is important. Because of concerns previously reported by us and others about the validity of the 13C-UBT in children ⬍24 months old,34 –36 only children ⬎24 months of age were studied. Furthermore, index children were categorized as having H pylori infection only after they had a positive 13C-UBT followed by a second confirmatory test. This strict criterion meant that only children with established infection were categorized as prevalent or incident cases. Forty-six of the 48 infected index children were retested for H pylori at the end of the study, and all 46 were still categorized as infected with H pylori, thus further confirming that all children who were categorized as infected had established infection. All noninfected index children had 5 negative 13C-UBT results over the duration of the study.

GASTROENTEROLOGY Vol. 130, No. 1

We therefore found no evidence that spontaneous clearance of H pylori infection occurs or that such clearance is followed by a subsequent infection, as has been reported.22,23,42,43 In this study, person-years of follow-up were allocated to age-dependent exposure categories (based on chronological age from 2 to 8 years). This allowed the examination of the risk of infection at different ages among those children still at risk of infection. It also provided a more complete picture of the risk of infection at each age interval. A fixed time analysis would have shown a similar overall incidence of infection (4.13 per 100 person-years). However, it would have missed the important finding that children are at the greatest risk of H pylori infection when they are ⬍3 years old. The results presented differ significantly from the findings of Malaty et al,20 who examined the incidence of infection in a group of ⬎200 children by using a commercial serological assay on stored serum samples from the Bogalussa Heart Study. They classified any positive result as evidence of infection, even if it was followed subsequently by all negative results. They found that the incidence of infection was highest in the 4 –5-year age group (2.1% per year) and that the rate of acquisition of infection remained high in the 7–9-year age group.20 In contrast, the results presented here suggest that infection with H pylori occurs much earlier than 5 years of age and is uncommon after this age. This difference can be explained by Malaty and colleagues’ use of a commercial serological assay. These assays have been shown to underestimate the true rate of infection, particularly in children younger than 10 years of age,24 –27 and therefore it is likely that the peak incidence after the age of 5 years described by Malaty et al represents delayed diagnosis with such an assay. Malaty et al also stated that older children were far more likely to have persistence of antibody response than children younger than 3 years of age, thus confirming the unsuitability of the assay for identifying H pylori infection in younger children. Our findings indicate that the prevalence of infection is decreasing rapidly in developed countries, in keeping with the findings of Sipponen et al.44 Whereas 48% of the mothers in our study were infected with H pylori, the prevalence of infection in their offspring declined to 16.5% (48/290). The precision of the 13C-UBT and the strict diagnostic criteria are unlikely to explain the low prevalence rate in children at the end of this study. The decision not to perform a second confirmatory test on noninfected index children would not have affected the prevalence of infection, because each noninfected index child subsequently had 4 13C-UBTs during the study. Furthermore, all index children with borderline results

January 2006

had the 13C-UBT repeated fasting, so it is unlikely that they were misclassified as noninfected. Although this study did show that H pylori was clustered in families and that infected family members were risk factors for infection, the risk of transmission occurring within families in developed countries has decreased markedly, as evidenced by the rapid decline in prevalence in index children compared with their mothers in this study. The prevalence of infection in mothers in this study was higher than reported previously from developed countries.2,3,11,12 This may be due to some differences in the socioeconomic profile of the populations studied. Furthermore, on the basis of the findings of this study, it may reflect the childhood socioeconomic status of the mothers in this study rather than their current socioeconomic status. Normal child-rearing guidelines for Western countries recommend that children should be weaned from their feeding bottle toward the end of the first year.45 In this study, using ⱖ2 bottles per day after 24 months of age was an independent and statistically significant risk factor for infection. A feeding bottle could potentially provide a vehicle for transmission of infection from mother to child or from sibling to sibling. Bottles could also come into contact with emesis—reflux and regurgitation are common in young children—and promote transmission from this source. In contrast, the use of a pacifier was not a risk factor for infection. It was expected that both the use of a pacifier and the use of a bottle by toddlers would carry a similar risk of infection. Both pacifiers and bottles are likely to be in contact with the same surroundings as the child becomes more mobile and to be subject to similar cleaning practices after the age of 1 year. A possible explanation for the different risks associated with prolonged pacifier or bottle use may be that the bottle provides a moist surface at all times, whereas pacifiers can dry out, and H pylori is very sensitive to drying out. H pylori has been shown to survive in milk for up to 10 days.46,47

Conclusion This is the first prospective study on the incidence of H pylori infection. This study shows that most young children with H pylori are infected before the age of 3 years and that the risk of infection is very low after 5 years of age. Having an infected mother, an infected older sibling, or not being weaned from a bottle by 2 years of age are all independent and statistically significant risk factors for infection. This finding that children become infected at a very young age has important implications for studies on the mode of transmission of H

AGE–SPECIFIC INCIDENCE OF HELICOBACTER PYLORI 71

pylori and the development of vaccine strategies to prevent infection.

References 1. Warren JR. Unidentified curved bacilli on gastric epithelium in active chronic gastritis. Lancet 1983;1:1273–1275. 2. Peterson WL. Helicobacter pylori and peptic ulcer disease. N Engl J Med 1991;324:1043–1048. 3. Dooley CP, Cohen H, Fitzgibbons PL, et al. Prevalence of Helicobacter pylori infection and histologic gastritis in asymptomatic persons. N Engl J Med 1989;321:1562–1566. 4. Drumm B, Sherman P, Cutz E, Karmali M. Association of Campylobacter pylori on the gastric mucosa with antral gastritis in children. N Engl J Med 1987;316:1557–1561. 5. Walsh JH, Peterson WL. The treatment of Helicobacter pylori infection in the management of peptic ulcer disease. N Engl J Med 1995;333:984 –991. 6. Hentschel E, Brandstatter G, Dragosics B, et al. Effect of ranitidine and amoxicillin plus metronidazole on the eradication of Helicobacter pylori and the recurrence of duodenal ulcer. N Engl J Med 1993;328:308 –312. 7. Coughlan JG, Gilligan D, Humphries H, et al. Campylobacter pylori and recurrence of duodenal ulcers—a 12 month follow up study. Lancet 1987;2:1109 –1111. 8. Goggin N, Rowland M, Imrie C, Walsh D, Clyne M, Drumm B. Effect of Helicobacter pylori eradication on the natural history of duodenal ulcer disease. Arch Dis Child 1998;79:502–505. 9. International Agency for Research on Cancer Working Group on the Evaluation of Carcinogenic Risks to Humans. Schistosomes, liver flukes and Helicobacter pylori. Vol 61. Lyon, France: International Agency for Research on Cancer, 1994:1–241. 10. Parsonnet J. Helicobacter pylori: the size of the problem. Gut 1998;43(Suppl 1):S6 –S9. 11. Banatvala N, Mayo K, Megraud F, Jennings R, Deeks JJ, Feldman RA. The cohort effect and Helicobacter pylori. J Infect Dis 1993; 168:219 –221. 12. Suerbaum S, Michetti P. Helicobacter pylori infection. N Engl J Med 2002;347:1175–1186. 13. Drumm B, Perez-Perez GI, Blaser MJ, Sherman PM. Intrafamilial clustering of Helicobacter pylori infection. N Engl J Med 1990; 322:359 –363. 14. Malaty H, Graham DY, Klein P, Evans DG, Adam E, Evans DJ. Transmission of Helicobacter pylori infection. Studies in families of healthy individuals. Scand J Gastroenterol 1991;26:927–932. 15. Lu Y, Redlinger TE, Avitia R, Galindo A, Goodman K. Isolation and genotyping of Helicobacter pylori from untreated municipal wastewater. Appl Environ Microbiol 2002;68:1436 –1439. 16. Ashorn M, Maki M, Hallstrom M, et al. Helicobacter pylori infection in Finnish children and adolescents. A serologic cross-sectional and follow-up study. Scand J Gastroenterol 1995;30:876 – 879. 17. Mitchell HM, Li YY, Hu PJ, et al. Epidemiology of Helicobacter pylori in Southern China: identification of early childhood as the critical period for acquisition. J Infect Dis 1992;166:149 –153. 18. Granstrom M, Tindberg Y, Blennow M. Seroepidemiology of Helicobacter pylori infection in a cohort of children monitored from 6 months to 11 years of age. J Clin Microbiol 1997;35:468 – 470. 19. Glynn MK, Friedman CR, Gold BD, et al. Seroincidence of Helicobacter pylori infection in a cohort of rural Bolivian children: acquisition and analysis of possible risk factors. Clin Infect Dis 2002;35:1059 –1065. 20. Malaty HM, El-Kasabany A, Graham DY, et al. Age at acquisition of Helicobacter pylori infection: a follow-up study from infancy to adulthood. Lancet 2002;359:931–935. 21. Isenbarger DW, Bodhidatta L, Hoge CW, et al. Prospective study of the incidence of diarrheal disease and Helicobacter pylori

72

22.

23.

24.

25. 26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

ROWLAND ET AL

infection among children in an orphanage in Thailand. Am J Trop Med Hyg 1998;59:796 – 800. Perez-Perez GI, Sack RB, Reid R, Santosham M, Croll J, Blaser MJ. Transient and persistent Helicobacter pylori colonization in Native American children. J Clin Microbiol 2003;41:2401–2407. Naficy AB, Frenck RW, Abu-Elyazeed R, et al. Seroepidemiology of Helicobacter pylori infection in a population of Egyptian children. Int J Epidemiol 2000;29:928 –932. Raymond J, Kalach N, Bergeret M, et al. Evaluation of a serological test for diagnosis of Helicobacter pylori infection in children. Eur J Clin Microbiol Infect Dis 1996;15:415– 417. Czinn SJ. Serodiagnosis of Helicobacter pylori in pediatric patients. J Pediatr Gastroenterol Nutr 1999;28:132–134. de Oliveira A M, Rocha GA, Queiroz DM, et al. Evaluation of enzyme-linked immunosorbent assay for the diagnosis of Helicobacter pylori infection in children from different age groups with and without duodenal ulcer. J Pediatr Gastroenterol Nutr 1999; 28:157–161. Khanna B, Cutler A, Israel NR, et al. Use caution with serologic testing for Helicobacter pylori infection in children. J Infect Dis 1998;178:460 – 465. Rothenbacher D, Bode G, Brenner H. Dynamics of Helicobacter pylori infection in early childhood in a high-risk group living in Germany: loss of infection higher than acquisition. Aliment Pharmacol Ther 2002;16:1663–1668. Rowland M, Lambert I, Gormally S, et al. Carbon 13-labeled urea breath test for the diagnosis of Helicobacter pylori infection in children. J Pediatr 1997;131:815– 820. Casswall TH, Nilsson HO, Bergstrom M, et al. Evaluation of serology, 13C-urea breath test, and polymerase chain reaction of stool samples to detect Helicobacter pylori in Bangladeshi children. J Pediatr Gastroenterol Nutr 1999;28:31–36. Kato S, Ozawa K, Konno M, et al. Diagnostic accuracy of the 13 C-urea breath test for childhood Helicobacter pylori infection: a multicenter Japanese study. Am J Gastroenterol 2002;97:1668 – 1673. Delvin EE, Brazier JL, Deslandres C, Alvarez F, Russo P, Seidman E. Accuracy of the [13C]-urea breath test in diagnosing Helicobacter pylori gastritis in pediatric patients. J Pediatr Gastroenterol Nutr 1999;28:59 – 62. Kalach N, Briet F, Raymond J, et al. The 13carbon urea breath test for the noninvasive detection of Helicobacter pylori in children: comparison with culture and determination of minimum analysis requirements. J Pediatr Gastroenterol Nutr 1998;26: 291–296. Imrie C, Rowland M, Bourke B, Drumm B. Limitations to carbon 13-labeled urea breath testing for Helicobacter pylori in infants. J Pediatr 2001;139:734 –737. Kindermann A, Demmelmair H, Koletzko B, Krauss-Etschmann S, Wiebecke B, Koletzko S. Influence of age on 13C-urea breath test results in children. J Pediatr Gastroenterol Nutr 2000;30:85–91. Koletzko S, Feydt-Schmidt A. Infants differ from teenagers: use of non-invasive tests for detection of Helicobacter pylori infection in children. Eur J Gastroenterol Hepatol 2001;13:1047–1052. Braden B, Duan LP, Caspary WF, Lembcke B. More convenient 13 C-urea breath test modifications still meet the criteria for valid diagnosis of Helicobacter pylori infection. Z Gastroenterol 1994; 32:198 –202. Ireland Department of Health and Children. Income guidelines for general medical services, 1997.

GASTROENTEROLOGY Vol. 130, No. 1

39. Fiedorek SC, Malaty HM, Evans DL, et al. Factors influencing the epidemiology of Helicobacter pylori infection in children. Pediatrics 1991;88:578 –582. 40. Breslow NE, Day NE. The design and analysis of cohort studies. In: Statistical methods in cancer research. Vol 2. Lyon, France: International Agency for Research on Cancer Scientific Publications, 1987:82–118. 41. Daly LE, Bourke GJ. Interpretation and uses of medical statistics. London: Blackwell Scientific, 2000. 42. Casswall TH, Sarker SA, Albert MJ, et al. Treatment of Helicobacter pylori infection in infants in rural Bangladesh with oral immunoglobulins from hyperimmune bovine colostrum. Aliment Pharmacol Ther 1998;12:563–568. 43. Thomas JE, Dale A, Harding M, Coward WA, Cole TJ, Weaver LT. Helicobacter pylori colonization in early life. Pediatr Res 1999; 45:218 –223. 44. Sipponen P, Helske T, Jarvinen P, Hyvarinen H, Seppala K, Siurala M. Fall in the prevalence of chronic gastritis over 15 years: analysis of outpatient series in Finland from 1977, 1985, and 1992. Gut 1994;35:1167–1171. 45. Findberg L. Feeding the healthy child. In: McMillan JA, DeAngelis CD, Feigin RD, Warshaw JB, eds. Oski’s pediatrics: principles and practice. Philadelphia: Lippincott Williams & Wilkins, 1999:470 – 479. 46. Fan XG, Chua A, Li TG, Zeng QS. Survival of Helicobacter pylori in milk and tap water. J Gastroenterol Hepatol 1998;13:1096 – 1098. 47. Dore MP, Sepulveda AR, Osato MS, Realdi G, Graham DY. Helicobacter pylori in sheep milk. Lancet 1999;354:132. Received June 10, 2005. Accepted September 28, 2005. Address requests for reprints to: Marion Rowland, MB, PhD, UCD School of Medicine and Medical Science, The Children’s Research Centre, Our Lady’s Hospital for Sick Children, Crumlin, Dublin 12. e-mail: [email protected]; fax: (353) 1-455-5307. This study was funded by a Wellcome Trust Project grant (051393/ Z/97/Z). We would like to acknowledge the contribution of the general practitioners and their staff who provided us with such generous access to their patients and facilities during the study. A special word of thanks goes to their secretaries who contributed in so many ways to this study and to Professor Gerard Bury, Coombe Family Practice, Dublin 8; Dr Fergus O’Kelly and Dr Kevin O’Doherty, Rialto Medical Centre, Dublin 8; Dr Philip O’Connell, James Street, Dublin 8; Dr Mary Slevin and Dr Claire McNicholas, Marian Road, Dublin 14; Dr Kieran Harkin and Dr Emer Loughrey, Inchicore Medical Centre, Dublin 8; Dr Michael Costello, Dr Philip Wiehe, and Dr Imelda Sliney, Sundrive Medical Centre, Dublin 12; Dr Liam Quigley and Dr Lindy Barnes, Glenfield Family Practice, Dublin 22; Dr Liam Lynch and Dr Lynda Hamilton, Neilstown Health Centre, Dublin 22; Dr John Carty, North Clondalkin Health Project, Dublin 22; Dr Peter Keogh, Dr Aodhan O’ Reilly, Dr Deborah McGrane, Dr Patrick McGrath, and Dr Alan Byrne, Parkhouse Family Practice, Dublin 24; Dr Andrew Jordan, Dr William Twomey, and Dr Elanor Johnson, Aylesbury Clinic, Dublin 24; Dr Robert Quigley, Dr William Kavanagh, and Professor Tom O’Dowd, Springfield Medical Centre, Tallaght; Dr Tony Feeney and Dr Mary Walsh, Lucan, Co Dublin; Dr Mary Carmel Bourke, Dr Catherine Hayes, and Dr Jacqueline McKee, Glasnevin Family Practice, Dublin 9; Dr David Moloney and Dr Kevin Brennan, The Red House Family Practice, Mallow, Co Cork; Dr Billy Christopher, West End, Mallow, Co Cork; and Dr John P. McMahon and Dr Shiela Casey, Kingscourt, Co Cavan.