Background
Helicobacter pylori (H. pylori) infection is chronic and common throughout the world, with a higher prevalence in developing than in developed countries [
1].
H. pylori infection causes gastritis and is the most important risk factor for peptic ulcer disease (gastric and duodenal)[
2,
3]. It also contributes to the onset of gastric cancer and primary gastric B-cell lymphoma [
2,
4] and more recently has been investigated as a risk factor for ischemic heart disease[
5] and intrauterine growth restriction[
6,
7].
In Brazil, in 1998, among the malignancies, gastric cancer was second only to lung cancer as the cause of mortality in men[
8]. Despite of this, local epidemiological data on
H. pylori infection are scarce. Through a non-restrictive search at Medline and SciElo databases using "
Helicobacter pylori" and "prevalence or seropositivity" and "Brazil or Brazilian" as title/abstract descriptor terms, only seven studies primarily planned to measure prevalence of the infection among adult non-patient individuals were found [
9‐
15]. Five of these studies had been conducted among selective populations (abattoir workers[
9], Japanese Brazilians and Japanese residents in Brazil[
11,
12], blood donors[
14], and native populations from Brazilian Western Amazon)[
15]. Overall prevalence rate in those studies varied from 48% among Japanese Brazilians living in four different cities in Brazil[
12] to 84.7% among the adult residents of a rural area of the state of Mato Grosso[
10].
This paper reports the prevalence rate of the H. pylori infection and the factors that showed to be associated with the infection in a study conducted in Pelotas, a city of 320,000 inhabitants, located in the state of Rio Grande do Sul (southern Brazil).
Methods
This was a population-based cross-sectional study. A multi-stage sampling method, taking the structure of the population as a basis, was used for sample selection. Fifty eight of the 404 census tracts of the city were randomly chosen and, subsequently, a random sample of five households per census tract was selected to the study. All adults (20 years and over) living in the selected households were invited to participate. Unoccupied houses were excluded and the house next-door was selected. In the case of temporary absence of one or more adults from the household (students, workers) the interviewer returned later, or during the next day, to perform the interview. After three unsuccessful attempts of contact, at different days, the adult/household was considered a loss. Refusals were also considered losses.
Subjects were interviewed at home by trained field workers using a standard questionnaire. Information on current (family monthly income in minimum wages, educational level, marital status, and size of the family) and past socio-economic characteristics (history of having lived only in urban or in urban and rural areas, level of formal education of the parents of the interviewee, number of siblings, and attendance to day-care centres in childhood) was collected.
Age was gathered in complete years. Gender and ethnicity (as defined according to skin colour) were observed by the interviewer. History of upper abdominal complaints in the last year: dyspeptic symptoms (pain or discomfort), substernal burning pain, decreased appetite, abdominal fullness, abdominal bloating, early satiety, and vomiting were investigated. History of peptic ulcer disease and gastric cancer in first degree relatives was explored.
Frequency of consumption of raw vegetables, coffee, alcoholic beverages and mate (Ilex paraguayensis) were collected. Smoking history was also provided by the interview. Changes in nutritional habits, alcoholic beverages ingestion and in smoking pattern due to the presence of upper abdominal symptoms were also investigated.
H. pylori infection was assessed by the isotope technique, the
13C-Urea Breath Test (
13C-UBT) after at least six hours fastening. Each dose consisted into about 50 gram portion of
13C-urea (Cambridge Isotope Laboratories Inc., Massachusetts, USA). The samples were measured in a mass spectrometer coupled to a gas chromatographer (FinniganMAT GmbH, ThermoQuest Corp., Bremen, Germany). Subjects with an excess δ
13CO
2 value of > 3.5 per ml were defined as
H. pylori positive[
16].
Due to the high prevalence of the
H. pylori infection, prevalence rates, prevalence ratios, and the respective 95% confidence intervals were calculated through robust Poisson regression[
17], taking the family clustering of the data into account. The Stata 8.0 package was used for these analyses. A hierarchical model of analysis[
18] was applied according to a conceptual framework established
a priori, through which caudal variables were adjusted for cranial ones. At the first most cranial level were the socio-economic characteristics in childhood. At the second level, the current socio-economic variables, followed at the third level by the demographic factors. In the fourth and fifth levels, lifestyle characteristics and complaints of upper gastrointestinal symptoms, respectively, were entered to the model. Variables significant at their level were kept in the model even whether the inclusion of caudal variables turned their association with the outcome statistically non significant. At every level, a backward selection of variables was performed. Those presenting a p value ≤ 0.20 were kept in the model for adjustment purpose. Only variables with p value < 0.05 were considered significantly associated with
H. pylori infection.
The study protocol was cleared by the Ethical Committee of the Federal University of Pelotas. A written informed consent was obtained from participants before enrolment in the study.
Discussion
The principal limitation of this study is the high refusal rate of the eligible individuals to realization of the
13C-UBT. Despite the simplicity and the non-invasiveness character of the
13C-UBT, which could make it suitable for use in field-based population studies, the refusal rate was very high (36.2%) what, as will be considered in detail below, can compromise the validity of some of the observed results. The need of 6-hour fastening is probably one of the main constraints for its use in field-based surveys. Nevertheless, similar or even higher refusal rates were reported by other authors using different methods in population-based studies collecting primary data: 32% in Australia[
19], 41.6% in Northern Ireland[
20] and 74% in England[
21]. Comprehensively, normal individuals are less compelled to adhere to a test than patients attending consultation for abdominal symptoms, the target population more frequently found in the published literature on
H. pylori prevalence. Studies conducted among symptomatic individuals, however, suffer from a different kind of selection bias since findings obtained from samples of gastroenterological patients are not necessarily representative of the whole population and can not be extrapolated with confidence to non-patient individuals. It is not surprising that prevalence rates available in the literature come mainly from the baseline phase of large community interventions focusing on modification of cardio-vascular risk factors and conducted in developed countries [
19‐
21].
Despite the refusals and considering that the prevalence rate of H. pylori infection was over 50% among individuals non-exposed to most of the exposures investigated, the study had a power of 80% to detect relative risks ≥ 2.0, at the significance level of 5%. Lack of power may have impaired the detection of association between the variables abdominal fullness, early satiety, vomiting, and abdominal bloating, because their prevalence was under 30% in the study population.
Another limitation of this study is the impossibility of declaring causality between associated factors and the outcome. The temporality between exposure and outcome can not be ascertained with precision in cross-sectional studies.
To the authors' knowledge, this is the first population-based study of
H. pylori infection targeting urban adult population in Brazil. Overall 64.3% of the population ≥ 20 years old was infected with
H. pylori. This figure is higher than the prevalence of infection detected in population-based studies conducted in developed countries like Australia (30.6%)[
19], England (27.6%)[
21] and United States (32.7%)[
22], and similar to prevalence rates detected in developing settings in South America, Africa and parts of Asia[
1]. Considering that the occurrence of infection was significantly associated to dyspepsia even after allowing for confounding and that refusals were more prevalent among non-symptomatic individuals, it is possible that the self-selective inclusion of symptomatic individuals may have in part overestimated the true prevalence of the infection. Projecting the infection rate observed among symptomatic and non-symptomatic complier subjects to symptomatic and non-symptomatic non-complier participants, however, the prevalence rate would be 64.0% (95% CI 60.0%–68.0%).
In the present study past socio-economic variables in childhood presented the strongest association with the occurrence of the infection in adult life (an adjusted increase of approximately 50% in the probability of infection). The age-dependent increase of
H. pylori prevalence which is observed in most of the studies conducted worldwide lost the significance in the present study when adjusted by socio-economic conditions in childhood. This finding is in agreement with the "birth cohort phenomenon" largely described in the literature. Despite the cross-sectional design and the adverse effect of recall bias, at this and at the study conducted by Mendall
et al[
23] in London, it was possible to explore the association of past socio-economic indicators in childhood with the infection rate in adulthood, a feature generally disregarded in prevalence studies conducted among adult subjects.
The high prevalence rate observed in Pelotas probably reflects infection acquired in childhood and carried on throughout life. In fact, in a population-based study conducted among children from an urban community in north-east Brazil[
24], 75.4% were
H. pylori positive by the age of 12–14 years. Another study conducted among low socio-economic children attending an outpatient clinic in Belo Horizonte, Brazil, showed that infection occurs early and increases with age[
25].
Without a known significant animal or environmental reservoir for human strains of
H. pylori, person-to-person contact appears to be the most likely mode of transmission. As a consequence and as others have shown, the number of siblings, particularly the number of older siblings, domestic crowding and living in orphanages are important determinants of the prevalence of H pylori infection[
26,
27]. The findings of this study confirmed that larger families and higher exposure as in day-care centres was associated with increased prevalence of the infection.
The study of Replogle
et al[
28] showed a sex difference, higher in men, in the prevalence of infection. The higher prevalence of
H. pylori-associated diseases like peptic ulcer and gastric cancer in males supports the hypotheses of a real association. In the present study no statistical association was observed between sex and dyspeptic symptoms in crude or in adjusted analysis. However, since the refusal rate was higher among men, it is not possible to exclude with certainty that men with increased risk of infection had not been lost preferably to others. This bias may have masked a real association between sex and
H. pylori infection, if it actually exists.
Among the demographic variables explored, only ethnicity remained independently associated with the infection. After allowing for present and past socio-economic conditions, non-white individuals presented a probability of infection 32% higher than the observed among the whites. In 1992, Malaty
et al[
29] in a study conducted in the United States with Hispanics matched with blacks and whites for age and socio-economic status, found that the risk of infection was almost identical in Hispanics and blacks and significantly higher than in whites. As suggested by the authors it was probably a reflection of a generation cohort phenomenon related to the generational distance from very low socio-economic status, i.e., the prevalence of
H. pylori in Hispanics and blacks is currently lower than that of their parents but higher than that of the white population, which has experienced higher socio-economic status for several generations. Lack of information in the present study regarding generational socio-economic conditions impaired to test this hypothesis.
In the study of McQuillan
et al [
22] also conducted in the United States showed that race remained statistically associated with the infection after adjustment for socio-economic factors only in low risk groups, what suggests that ethnicity can be a surrogate for other non-explored factors.
Regarding behavioural factors, the majority of recent studies have not found tobacco use or alcohol consumption to be risk factors for
H. pylori infection[
30]. A study specifically planned to measure whether smoking or consumption of alcohol or coffee was associated with active
H. pylori in southwest England concluded that smoking or coffee consumption were not related to active
H. pylori infection and that total alcohol consumption was associated with a small, but not statistically significant, decrease in the odds of infection[
31]. In the present study no association was found between these exposures and
H. pylori infection. These findings however must be seen with caution because changes in lifestyle due to the development of upper gastrointestinal symptoms may mislead the results of cross-sectional studies. In fact, in the present study, changes in alcohol, coffee and mate intakes were reported, respectively, by 35%, 61% and 35% of the participants with dyspeptic symptoms, most of them having reduced the amount or quitted those intakes as an attempt to deal with the symptoms. At the context of the cross-sectional studies, these variables are generally analyzed taking the current status instead of the status before the onset of upper gastrointestinal symptoms as the exposure. A misclassification error moving the association toward the unit is a possible consequence of such approach.
Regarding gastrointestinal complaints only dyspeptic symptoms were significantly associated with the infection. Meta-analyses of trials which have been done in patients with functional (that is, investigated) dyspepsia have shown no benefit from eradication of
H. pylori[
32]. Since individuals classified as presenting dyspeptic complaints in the current study were not investigated for identifying the cause of their symptoms, they may have functional dyspepsia or diseases such as peptic ulcer or gastro-oesophageal reflux disease. At this particular, a randomised placebo controlled trial conducted in 36 family practices in Canada to determine whether a "test for
Helicobacter pylori and treat" strategy improved symptoms in patients with uninvestigated dyspepsia showed a significant symptomatic benefit at 12 months of follow-up[
33].
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
ISS conceived the study, participated in its design, analysis and coordination, and drafted the manuscript. JB and ASS participated in the design of the study and carried out the isotope analyses. NCJV participated in the design of the study and performed the statistical analysis. CSH, MCB and RDL participated in the design and coordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript.