Intended for healthcare professionals


Relation between infection with Helicobacter pylori and living conditions in childhood: evidence for person to person transmission in early life

BMJ 1994; 308 doi: (Published 19 March 1994) Cite this as: BMJ 1994;308:750
  1. P M Webb,
  2. T Knight,
  3. S Greaves,
  4. A Wilson,
  5. D G Newell,
  6. J Elder,
  7. D Forman
  1. Imperial Cancer Research Fund, Cancer Epidemiology Unit, Gibson Building, Radcliffe Infirmary, Oxford OX2 6HE
  2. Department of Surgery, Postgraduate Medicine, University of Keele, Stoke on Trent Central Veterinary Laboratories, New Haw, Weybridge KT15 4 NB
  1. Correspondence to: Ms Webb.
  • Accepted 12 November 1993


Abstract Objectives : To relate the prevalence of infection with Helicobacter pylori in adults to their living conditions in childhood to indentify risk factors for infection.

Design : Prevalence study of IgG antibodies to H pylori (>10 μg IgG/ml, determined by enzyme linked immunosorbent assay (ELISA)) and reported living conditions and other socioeconomic factors in childhood.

Setting : Three factories in Stoke on Trent.

Subjects : 471 male volunteers aged 18 to 65 years.

Main outcome measures : Seroprevalence and variables in childhood.

Results : Seroprevalence of H pylori increased with age (22/74 (29.7%) at <30 years v 29/46 (63%) at 55-65 years; P<0.001 for trend) and was related to manual occupation (14/65 (21.5%) for non-manual v 162/406 (39.9%) for manual; P=0.003). After data were adjusted for age and occupation subjects from large families, whose childhood homes were crowded or who regularly shared a bed in childhood, were significantly more likely to be seropositive (adjusted odds ratio (95% confidence interval) 2.15 (1.41 to 3.30) for crowding and 2.13 (1.38 to 3.30) for sharing a bed), but there was no relation with possession of a bathroom, inside toilet, refrigerator, or household pets in childhood.

Conclusions : Close person to person contact in childhood is an important determinant of seroprevalence of H pylori in adulthood, suggesting that the infection is transmitted directly from one person to another and may be commonly acquired in early life.

Epidemiological implications

  • Epidemiological implications

  • Infection with Helicobacter pylori is associated with the development of chronic gastritis, peptic ulcer disease, and, probably,gastric cancer

  • Half of the world population may be infected but little is known about how the organism is transmitted

  • Subjects from larger families, overcrowded homes, or who had regularly shared a bed during childhood were more likely to be seropositive

  • The similar increase in prevalence with age in both manual and non- manual workers was consistent with a cohort effect in which most infections were acquired in childhood

  • Very close person to person contact in childhood - for instance, by sharing a bed - may provide a route for the transmission of H pylori


Infection with Helicobacter pylori has been established as a major cause of chronic gastritis.1 It also plays an important part in the development of peptic ulcer disease2,3 and has been implicated in the process of gastric carcinogenesis.4 Around half of adults may be infected in developed countries, while this figure could be as high as 90% in developing countries,5 yet little is known about the way in which the infection is transmitted. No reservoir of organisms has been identified outside the human stomach, and, although reports have suggested that H pylori can be isolated from faecal material6 and dental plaque,7 these reports have yet to be confirmed.

Infection tends to cluster in families8,9 and in people living in crowded10,11 or closed environments,*RF 12-14* suggesting that person to person contact plays an important part in transmission. Infection has also been associated with poor hygiene,15 and it has been suggested that acquisition may commonly occur in childhood.11

We aimed to relate infection with H pylori in adults to their living conditions in childhood by using the presence of circulating antibodies to H pylori as a marker of infection. Of particular interest was the effect of sharing a bed in childhood, previously reported by Galpin et al,10 and the possibility that this might explain some of the observed relations with overcrowding and lower socioeconomic status.

Data were collected as part of a larger study of gastritis in Stoke on Trent.

Subjects and methods

Male employees at two factories in Stoke on Trent and men attending a blood donation session at a third factory were invited to participate. Volunteers completed a questionnaire that included detailed questions about their living conditions during two periods of childhood (below 5 years and between 5 and 16 years) before attending an appointment when the information was checked for completeness and a blood sample (10 ml) was collected by venepuncture.

Serum samples were stored below 4°C until they could be transferred to a -40°C freezer at the end of each day. Samples were assayed for antibodies to H pylori by using an established enzyme linked immunosorbent assay (ELISA),16 with a cut off point of greater than 10 μg IgG/ml indicating positivity. This method has been shown to give a sensitivity of 96% and specificity of 93%.17

Each subject's occupation and that of their father when the subject was aged about 10 years was classified as non-manual (registrar general's class I, II, or III non-manual) or manual (III manual, IV, or V) according to the classification of occupations (1980) of the Office of Population Censuses and Surveys.18 To assess the effects of crowding an index was calculated for each subject as the ratio of the number of people in the household when they were young to the number of living rooms in the house.

We used the X2 test to compare the percentage prevalence of infection in different categories of the explanatory variables. Percentages were indirectly adjusted for age by using four age groups (<30, 30-39, 40-49, and 50-65 years), and unconditional logistic regression was used to calculate odds ratios and 95% confidence intervals.


A total of 476 men volunteered for the study. This comprised 258/782 (33%), 138/431 (32%), and 80/113 (71%), of the eligible workforce in each factory. Serum samples were not available from five people so data were analysed for the remaining 471 men.

A total of 176 men (37%) were positive for antibody to H pylori. Seroprevalence varied between the three factories (43% (111/257), 34% (47/138), and 24% (18/ 76), X2=10.4, P=0.005), but these differences were of borderline significance after we adjusted for age and manual occupation (X2=6.2, P=0.044). The data were combined for further analyses.

Subjects were aged between 18 and 65 years, and seroprevalence for H pylori increased significantly with age (table I). Seroprevalence was significantly higher in manual workers than non-manual workers and in subjects whose fathers were manual workers compared with those whose fathers were non-manual workers. These effects were independent and remained significant after we adjusted for age (table II). The difference between occupation groups was already present in the youngest age group (table I), and the yearly increase in seroprevalence was similar in both manual and non-manual workers (0.8% and 0.9% per year, respectively).


Numbers (percentages) of subjects seropositive for Helicobacter pylori by age and occupation

View this table:

Table II also shows the association between seroprevalence for H pylori and the other variables studied. Seropositivity was significantly associated with number of siblings, crowding, sharing a bed (but not sharing a bedroom), and with the lack of a bathroom, indoor toilet, or refrigerator in childhood. There was no association with the possession of household pets. After we adjusted for age, the relations with possession of a bathroom, indoor toilet, or refrigerator in childhood became non-significant.


Seroprevalence of Helicobacter pylori and socioeconomic variables

View this table:

The relation with crowding was not linear; subjects from homes with more than one person per room were significantly more likely to be seropositive (67/134=50.0%) than subjects from less crowded homes (92/305=30.2%). This cut off point, which is also used by the Office of Population Censuses and Surveys to classify homes as overcrowded,19 was used for further analyses.

Table III shows adjusted odds ratios for the variables that were significantly associated with seropositivity, after adjustment for age, in table II. The effects of number of siblings, overcrowding, and sharing a bed remained highly significant after we adjusted for age and current occupation, while the effect of father's occupation was reduced. Because of the strong collinearity between these variables associated with childhood they all became non-significant when included in a single, fully adjusted model.


Adjusted odds ratios (95% confidence intervals) for seropositivity for Helicobacter pylori and socioeconomic variables

View this table:

In an attempt to overcome this problem the three variables - number of siblings, overcrowding, and sharing a bed - were included in pairs in a multivariate model that also contained age and current occupation (table III). The effect associated with number of siblings remained significant when included with overcrowding or sharing a bed; this could, however, have arisen simply because this variable could be recalled more accurately. Inclusion of overcrowding and sharing a bed in the same model resulted in effects of similar size and significance.

The above results relate to the period when subjects were aged between 5 and 16 years, but similar effects were observed for the period below 5 years.


The association between infection with H pylori and lower socioeconomic status is well known,20 and recently infection has been related to specific aspects of socioeconomic status, including density of living,11,15 sharing a bed,10 and the lack of a hot water supply15 during childhood. These variables are closely interrelated, and it has not been possible to identify clearly those that might directly affect transmission of the organism.

In this study H pylori seropositivity was related to current and childhood socioeconomic status and to measures of overcrowding and close person to person contact. After adjustment for age there was no association with either a lack of refrigerated food storage or, in agreement with Whitaker et al21 but in contrast to Mendall et al,15 with measures of poor hygiene. These latter variables and H pylori infection are all strongly correlated with age. It is therefore impossible to say whether the relations were confounded by age or if the age effect itself could have resulted from changes in living conditions over the past 60 years. In either case the persistence of the relations with measures of crowding after adjustment for age suggests that close person to person contact is more important for transmission of H pylori than poor hygiene or contaminated food. There was no evidence that infection could be acquired from household pets.

Because of the problems of collinearity it was not possible to separate the effects of number of siblings, overcrowding, and sharing a bed in childhood. Socioeconomic status or family size as such, however, cannot be risk factors for infection but are probably markers for some other factor that directly affects transmission. Sharing a bed in childhood could be such a factor as the proximity of two people sharing a bed would allow either oral-oral or faecal-oral transmission of an organism such as H pylori.

In a study of healthy volunteer families with children, 68% of partners of index subjects seropositive for H pylori were also seropositive compared with only 9% of partners of seronegative subjects.22 In contrast, 17.3% of people attending an infertility clinic were seropositive but in only 6.6% of couples were both partners seropositive,23 suggesting that children may be important for the transmission of H pylori in families. It is possible that children are not only more susceptible to infection than adults but are also able to transmit the infection to non-infected adults.

Effect of age

The difference in prevalence between non-manual and manual workers was already apparent in the 20 to 29 year age group, and the change in prevalence above this age was around 1% a year. This is consistent with data from China, where differences were also seen in the youngest age group (under 5 years) with a constant increase in prevalence of about 1% a year above this age.11 The similar rate of change in adults of different social classes suggests two possibilities. Either risk factors for infection in adults are independent of class and therefore differ from those acting during childhood, or most infections may be acquired before adulthood and the observed increase in seroprevalence with age in adults could be predominantly a cohort effect. The latter possibility is supported by a study that compared the prevalence of antibodies to H pylori in stored serum samples collected in 1969, 1979, and 1989.24 Changes in living conditions and trends towards a smaller family size over the past 50 to 60 years could account for the reduction in seroprevalence of H pylori in successive cohorts.

We collected data for two periods during childhood in an attempt to identify the age at which children might be most susceptible to H pylori infection. There was no difference between the effects observed for the two periods, suggesting that infection can be acquired throughout childhood. This is supported by the observation that subjects who had shared a bed both below the age of 5 and between the ages of 5 and 16 were more likely to be seropositive than those who had shared a bed only during one of the two periods (56.2% v 40.0%, P=0.06).

Bias and confounding

In a study reliant on volunteers and the collection of retrospective data, the possibilities of self selection and recall bias cannot be excluded. Subjects were aware that the study would focus on stomach problems but were unaware of whether they were infected with H pylori. There was no association between seropositivity for H pylori and reported stomach problems, and it seems unlikely that, after adjustment for age, there would be any difference in recall between seropositive and seronegative subjects. The absolute prevalence of infection may differ from the general population, but this would not affect the validity of comparisons made within the population. In addition, the effect of bed sharing remained significant when subjects from the first two factories (where only 33% of the eligible workforce volunteered) were considered separately from those from the third factory (where 71% of a group of blood donors participated), suggesting that this effect is not likely to be restricted to this population.

Although number of siblings, overcrowding, and sharing a bed were strong risk factors for infection, a quarter of subjects who had no siblings, whose childhood homes were not crowded, and who did not regularly share a bed were seropositive. This implies that there are other risk factors for infection - for instance, when children come into close contact with other children, perhaps in nurseries or schools.



In conclusion, our results are consistent with the hypothesis that infections with H pylori are commonly acquired in childhood and that the increases in prevalence observed with age may be predominantly a cohort effect. They confirm previous associations between seropositivity and socioeconomic status,20 overcrowding,11,15 and sharing a bed,10 suggesting that close person to person contact, such as in sharing a bed in childhood, may increase the risk of infection.

(c) Crown copyright reserved 1994.