Review article: associations between Helicobacter pylori and obesity - an ecological study
Summary
Background
There is emerging debate over the effect of Helicobacter pylori infection on body mass index (BMI). A recent study demonstrated that individuals who underwent H. pylori eradication developed significant weight gain as compared to subjects with untreated H. pylori colonisation.
Aim
To elucidate the association between H. pylori colonisation and the prevalence of overweight and obesity in developed countries.
Methods
The literature was searched for publications reporting data on H. pylori prevalence rates and obesity prevalence rates. Studies selected reported H. pylori prevalence in random population samples with sample sizes of more than 100 subjects in developed countries (GDP >25 000 US$/person/year). Corresponding BMI distributions for corresponding countries and regions were identified. Nonparametric tests were used to compare the association between H. pylori and overweight and obesity rates.
Results
Forty-nine studies with data from 10 European countries, Japan, the US and Australia were identified. The mean H. pylori rate was 44.1% (range 17–75%), the mean rates for obesity and overweight were 46.6 (±16)% and 14.2 (±8.9)%. The rate of obesity and overweight were inversely and significantly (r = 0.29, P < 0.001) correlated with the prevalence of H. pylori infection.
Conclusions
There is an inverse correlation between H. pylori prevalence and rate of overweight/obesity in countries of the developed world. Thus, the gradual decrease of the H. pylori colonisation that has been observed in recent decades (or factors associated with decrease of) could be causally related to the obesity endemic observed in the Western world.
Introduction
In Western countries, the prevalence of those people classified as overweight (BMI >25 > 30) and obese (BMI >30) has substantially increased.1, 2 The WHO in 2008 estimated that over 500 million adults have obesity, representing 10–14% of the world's population.2 In this context, there is emerging debate over the effect of Helicobacter pylori (H. pylori) infection on body mass index (BMI). While several cross-sectional studies have observed an association of H. pylori infection with BMI,3, 4 the National Health and Nutrition Examination Survey (NHANES III) did not find an association.5 Other studies have evaluated the impact of H. pylori eradication on BMI. Kamada et al. demonstrated that Japanese patients who underwent H. pylori eradication developed significant weight gain as compared to subjects with untreated H. pylori colonisation.6 This has since been confirmed in a larger population-based randomised controlled trial in England where people gained more than 3 kg of weight in the intervention (19%) as compared to placebo (13%), although follow-up of this study was limited to 6 months.7 In addition, animal studies have shown that H. pylori colonisation decreased fasting blood glucose levels, increased levels of leptin, improved glucose tolerance, and suppressed weight gain.8 All this may indeed suggest a role of the reduction of the H. pylori prevalence as a contributing factor for the obesity endemic.
It is widely accepted that environmental factors such as the availability and the characteristics of food and dietary habits9 play a role, while there are also genetic factors10 that influence the manifestation of obesity. On the other hand, it is interesting to note that some cohort studies do not find an association between nutrition and obesity11 suggesting that factors other than caloric intake are important for the obesity endemic. Nevertheless, there are overwhelming data supporting that the endemic of obese patients is linked to an increased incidence of cancer12 and liver disease.13
While the obesity epidemic has been observed for prolonged periods of time in the Western World, there is now emerging evidence that the obesity epidemic is spreading to the developing world.14 Interestingly, there is now a substantial decrease in the H. pylori prevalence. However, it is as yet unknown if on a larger scale (e.g. across different populations and countries) there might be an association between H. pylori and obesitas. In the Western World, the prevalence of H. pylori is decreasing.15 H. pylori infection is prevalent worldwide, with a large disparity between developed and developing countries.2 This decreasing prevalence of H. pylori may represent a risk or contributing factor to the world-wide endemic of obesity with all its complications. However, as yet there are no systematic analyses that aim to assess a potential association between the prevalence of H. pylori and the prevalence of obesity across various geographical regions of the world. Thus, this study aimed to elucidate the potential link between H. pylori and the prevalence of obesity and people who are overweight in developed countries.
Methods
Helicobacter pylori prevalence
The literature was searched for publications reporting H. pylori prevalence rates between 1990 and 2012. Two researchers (NL & GH) conducted searches independently. Literature keyword searches included ‘Helicobacter pylori’ and ‘prevalence’ and retrieved 6945 articles. This was narrowed down on the basis of the outlined inclusion and exclusion criteria including adult population, random (population) samples of both gender, H. pylori infection assessed via serology, urea breath test or stool samples and sample size >100 subjects. Studies focussing on nonrandom samples or patient cohorts with H. pylori associated disease were excluded. In addition, only studies from countries with a GDP >$ 25 000 were included. Based upon this, 196 publications were identified. Full text articles were reviewed to ensure that all inclusion and exclusion criteria were met. Finally, 50 studies with data from 10 European countries, Japan, USA, Canada and Australia were identified that were suitable for further analysis (Table 1). WHO and other databases were then searched for overweight and obesity prevalence rates. Data from the relevant countries and years corresponding (±5 years) with the H. pylori data were selected (Table 1).
Study | Sample size | Methods | Country | Year | Prevalence H. pylori | Prevalence overweight | Prevalence obesity |
---|---|---|---|---|---|---|---|
Shapira et al.31 | 100 | Serology | Italy | 2011 | 0.3 | 0.44 | 0.098 |
Peleteiro et al34 | 649 | Serology | Portugal | 2011 | 0.85 | 0.535 | 0.142 |
Lane et al.7 | 10 537 | UBT | UK | 2011 | 0.155 | 0.61 | 0.227 |
Adamu et al.33 | 5229 | Serology | Germany | 2011 | 0.4984 | 0.665 | 0.129 |
Whiteman et al.35 | 1346 | Serology | Australia | 2010 | 0.23 | 0.49 | 0.164 |
Telaranta-Keerie et al.36 | 4256 | Serology, ELISA | Finland | 2010 | 0.19 | 0.488 | 0.157 |
Sasazuki et al.37 | 494 | Serology | Japan | 2010 | 0.751 | 0.232 | 0.031 |
Salomaa-Rasanen et al.38 | 504 | Serology | Finland | 1994 | 0.36 | 0.42 | 0.106 |
Risch et al.32 | 690 | Serology; ELISA | USA | 2010 | 0.17391 | 0.669 | 0.339 |
Kawai et al.39 | 418 | Serology, IgG | Japan | 2010 | 0.337 | 0.232 | 0.031 |
Fullerton et al.40 | 2437 | Serology | England | 2009 | 0.264 | 0.61 | 0.227 |
Moujaber et al.41 | 2413 | Serology | Australia | 2002 | 0.151 | 0.462 | 0.151 |
De Bastiani et al.43 | 104 | Urea breath test | Italy | 2008 | 0.54 | 0.44 | 0.098 |
Stettin et al.44 | 563 | Stool samples | Germany | 2007 | 0.21 | 0.665 | 0.129 |
Sanchez Ceballos et al.45 | 481 | Urea breath test | Spain | 2007 | 0.603 | 0.534 | 0.156 |
Naja et al.46 | 1306 | Serology | Canada | 2007 | 0.231 | 0.591 | 0.231 |
Sharara et al.47 | 104 (males only) | Serology | Italy | 2006 | 0.683 | 0.44 | 0.098 |
Macenlle Garcia et al.48 | 383 | Urea breath test | Spain | 2006 | 0.691 | 0.534 | 0.156 |
Cardenas et al.49 | 7462 | Serology | USA | 2000 | 0.271 | 0.583 | 0.307 |
Kuepper-Nybelen et al.49 | 6545 | Serology | Germany | 1999 | 0.407 | 0.60 | 0.203 |
Ioannou et al.26 | 3130 | Serology | USA | 2005 | 0.35 | 0.669 | 0.339 |
Ang et al.50 | 595 | Serology | Japan | 2005 | 0.463 | 0.325 | 0.069 |
Cho et al.5 | 7003 | Serology | USA | 2005 | 0.38 | 0.669 | 0.339 |
Simon et al.51 | 6746 | Serology | USA | 2003 | 0.32 | 0.545 | 0.307 |
Robertson et al.52 | 500 | Serology | Australia | 2003 | 0.32 | 0.49 | 0.164 |
Nishise et al.53 | 695 | Serology | Japan | 2003 | 0.6 | 0.232 | 0.031 |
Iszlai et al.54 | 756 | ELISA | Hungary | 2000 | 0.586 | 0.532 | 0.177 |
Moayyedi et al.55 | 8429 | 13C urea breath test | England | 2002 | 0.276 | 0.61 | 0.227 |
Bode et al.56 | 474 | Serology | Germany | 2001 | 0.308 | 0.606 | 0.194 |
Bazzoli et al.57 | 1533 | 13C Urea breath test | Italy | 2001 | 0.679 | 0.42 | 0.085 |
Rosenstock et al.58 | 2527 | Serology | Denmark | 1983–1994 | 0.247 | 0.417 | 0.094 |
Everhart et al.59 | 5465 | Serology | USA | 1988–1991 | 0.325 | 0.55 | 0.227 |
Russo et al.60 | 2598 | Serology | Italy | 1995–1997 | 0.45 | 0.385 | 0.064 |
Peach et al.61 | 324 | Serology | Australia | 1999 | 0.3 | 0.573 | 0.193 |
Collett et al.62 | 1060 | Serology | Australia | 1999 | 0.24 | 0.573 | 0.193 |
Senra-Varel et al.63 | 332 | Serology | Spain | 1998 | 0.43 | 0.36 | 0.121 |
Martin de Argila et al.64 | 301 | Serology | UK | 1998 | 0.522 | 0.669 | 0.11 |
Lin et al.65 | 273 | Serology | Australia | 1998 | 0.38 | 0.598 | 0.193 |
Dite et al.66 | 309 | Serology | Czech republic | 1998 | 0.588 | 0.46 | 0.114 |
Babus et al.67 | 456 | Serology | Croatia | 1998 | 0.5105 | 0.644 | 0.231 |
Rodrigo Saez et al.68 | 480 | Serology | Spain | 1997 | 0.492 | 0.488 | 0.121 |
Peach et al.61 | 217 | Serology | Australia | 1995 | 0.306 | 0.573 | 0.193 |
Murray et al.69 | 4742 | Serology | Northern Ireland | 1997 | 0.505 | 0.669 | 0.11 |
Babus et al.70 | 3082 | Serology | Croatia | 1997 | 0.604 | 0.644 | 0.231 |
Martin-de-Argila et al.71 | 381 | Serology | Spain | 1996 | 0.53 | 0.488 | 0.121 |
Breuer et al.72 | 260 | Serology | Germany | 1996 | 0.392 | 0.606 | 0.194 |
Asaka et al.73 | 109 | Serology | Japan | 1995 | 0.743 | 0.224 | 0.022 |
Holtmann et al.74 | 180 | Serology | Germany | 1994 | 0.317 | 0.606 | 0.186 |
Graham et al.75 | 485 | 13C Urea breath test | USA | 1991 | 0.502 | 0.55 | 0.227 |
- The table includes only data from countries with a gross domestic product >US$ 17 000 (purchasing power parity (PPP) adjusted).
Overweight and obesity prevalence
Overweight and obesity are defined as abnormal or excessive fat accumulation that may impair health. Based upon the WHO definitions, we used the body mass index (BMI). It is defined as a person's weight in kilograms divided by the square of his height in metres (kg/m2). A BMI greater than or equal to 25 is overweight and a BMI greater than or equal to 30 defines obesity.16
Data analysis
Mean prevalence rates for the different countries and different methods for the assessment were calculated. In a second step, the correlations between the H. pylori prevalence and overweight and obesity rates in the respective geographical regions were calculated. To ensure that the association between H. pylori and obesity/overweight prevalence rates were not due to an effect of GDP per capita an additional partial correlation analysis that removed the effect of the purchase power parity (PPP) adjusted GDP per capita was performed. All analysis statistical analysis were done utilising spss version 22.
Results
Forty-nine studies with a total of 99 463 subjects were identified (Table 1). There were significant differences in H. pylori prevalence rates for different countries. The lowest H. pylori rates were reported from Australia (15.1%), the highest for Portugal (85%) with a mean H. pylori rate of 41.9%. Similarly, there was considerable variability in prevalence rates for overweight and obesity ranging from 22.4 to 66.9% (mean 51.9) and 2.2 to 33.9% (mean 16.2), respectively. As shown in Figure 1 the prevalence of obesity and overweight were inversely and significantly (r = −0.43, P < 0.01 and r = −0.292, P < 0.05) correlated with the prevalence of H. pylori infection (Figure 1). The association between H. pylori prevalence and overweight or obesity remained significant after adjusting for GDP per capita (r = −0.352, P < 0.02 and r = −0.291, P < 0.05). Interestingly, GDP per capita and H. pylori prevalence were inversely correlated (r = −0.387, P < 0.01).

Discussion
The key finding of this study is a striking inverse association between the H. pylori prevalence in various countries and the prevalence of overweight or obesity in these countries. This finding is consistent with previous observations in controlled trials that after successful H. pylori eradication patients experience a significant increase in weight that was not observed in control subjects who had placebo instead of H. pylori eradication.7 Along the same lines, a cohort study observed that children who were never infected with H. pylori or cleared the infection grew significantly faster (gained weight) than those with persistent H. pylori infection.18 Another intervention study demonstrated in H. pylori-infected children growth retardation and low serum acylated ghrelin.16 In these children, H. pylori eradication restored ghrelin levels and increased body weight gains and growth.
Weight gain following H. pylori eradication could be attributed to improvement of postprandial symptoms such as early satiety that may affect some people. However, the efficiency of H. pylori eradication with regard to symptoms is very small17 and other large well controlled studies were not able to demonstrate any effect.19 More importantly, a positive trial6 conducted in a geographical region that had a high peptic ulcer prevalence and long waiting times for endoscopic procedures suggests that the beneficial effect might be due to undiagnosed peptic ulcer disease. There is also some evidence that H. pylori may protect against symptoms of gastro-oesophageal reflux while obesity is a risk factor for the manifestation of GERD.20 Thus, it might be speculated that the manifestation of GERD symptoms affects behaviour and thus may have an impact on weight gain.
Among other factors, the hormone ghrelin is relevant for the regulation of appetite and food intake.21 Interestingly, in one study, plasma ghrelin increased profoundly in asymptomatic subjects after H. pylori cure.22 This could lead to increased appetite and consequently weight gain, and contribute to the increasing obesity seen in Western populations where H. pylori prevalence is low. Interestingly, other studies have shown that circulating meal-associated leptin and ghrelin levels change significantly after H. pylori eradication, providing further direct evidence that H. pylori colonisation is involved in ghrelin and leptin regulation, with consequent effects on body morphometry.23 In another study of 156 patients, who were undergoing laparoscopic vertical-banded gastroplasty for obesity, the density of ghrelin-positive cells was significantly lower for H. pylori-infected patients. Moreover, there was a significant stepwise decrease in density of ghrelin-positive cells, with progression of histological severity of chronic inflammation, neutrophil activity and glandular atrophy in the corpus.24 H. pylori infection was also associated with a reduction in circulating ghrelin levels independent of sex and BMI.25 In contrast, in a cohort study with nearly 7000 subjects, H. pylori seropositivity and CagA antibody status were not associated with body mass index or fasting serum leptin level.26
It might be argued that the inverse relation between H. pylori and obesity that was found in our study is mediated by the development status of the various countries which is reflected by the GDP. In our study, we have only included countries with a (PPP adjusted) GDP per capita >US$ 25 000 to ensure that only developed countries were included. In addition, in a separate analysis on the association between H. pylori and the prevalence of overweight and obesity, we even adjusted for the GDP per capita in a partial correlation analysis. Even after this adjustment, the association between H. pylori and overweight/obesity remained significant across these developed countries. Thus, our data clearly support the assumption that the association is not simply mediated by the income or the development status of the populations included in this analysis.
While our data may suggest that the decrease in H. pylori prevalence observed in many countries in recent decades could be a contributing factor to the obesity endemic of the Western world, our study cannot rule out that other factors that are correlated with the risk of a H. pylori infection are causal for the observed association. Indeed this ecological study does not provide individual-level analysis. This can be seen as an advantage since an ecological study might not be affected by some of the biases that may affect individual-level analysis. However, there are other potential limitations such as ecological fallacy of the ecological approach that require a careful interpretation of the data. In an ecologic study, the correlation between individual variables is deduced from the correlation of the variables collected for the group to which those individuals belong. As a consequence, the correlation of aggregate quantities may not be equal to the correlation of individual quantities. On the other hand, the ecological correlation (correlation of aggregate quantities) may avoid specific selection biases (e.g. the patient cohorts may have received antibiotic treatment for the management of obesity related diseases that may have affected the H. pylroi status). It also needs to be considered that there might be a temporal association between H. pylori and obesity. i.e. while H. pylori decreases over time, there is a parallel increase of obesity. However, it is unlikely that this effect is the explanation for our results since during the time period between 1991 (oldest study) and 2011 (most recent study included) there was no significant reduction of the H. pylori prevalence over time (P > 0.20). Thus, a temporal association is unlikely to explain the link between H. pylori and obesity. Another potential error could be some kind of selection bias. However, a very comprehensive review of the published literature plus strict inclusion criteria should help to avoid this error. It is important in this context to note that independent of our study other case control studies found very similar results.3, 4
However, other potential factors need to be considered. e.g. the gastrointestinal microbiome might be different in subjects with or without H. pylori infection. Subjects with an H. pylori infection might have been exposed to substantially more bacteria that ultimately may colonise the gastrointestinal mucosa. Thus subjects with H. pylori infection may have been exposed to more bacteria as compared to pylori-negative subject and thus may have a different gastrointestinal microbiome. In addition, the patient's gastrointestinal microbiome may respond to antibiotic therapy.27 Thus while this ecological study clearly supports a link between H. pylori and obesity it cannot be excluded that this is an indirect relationship. It might not be the H. pylori infection but other factors linked to H. pylori that are causal for the observed association between H. pylori and obesitas. Thus the role that the gastrointestinal microbiome may play needs to be properly addressed. It is well established that there is an interplay between diet, microbiome and health.28 Recent evidence suggests that the gut microbiome may be related to the development of obesity. This includes evidence of associations between obesity phenotypes and microbial class representation in the gut29 although further studies are needed. Studies looking at the effects of eradication treatment for H. pylori infections on the intestinal microflora27, 30 have been plagued by methodological weakness and more research is needed in order to fully understand the role that the microbiome may play in the observed relationship between decreasing H. pylori prevalence and the increasing rate of overweight/obesity.
While this study has several strengths including the focus on developed countries to avoid potential confounders, it should be noted that the same obesity epidemic is taking place in developing countries14 and not necessarily driven by changes of H. pylori prevalence. As we conducted an ecological study, we were not able to provide individual-level analysis and it is therefore critical to carefully interpret the data. However, with regard to the association between H. pylori and obesity besides direct effects of the H. pylori infection also potential confounders such as the influence of antibiotic use and the gastrointestinal microbiome should be examined in future studies in this area.
In summary, while previous data already have suggested that weight gain may occur after antibiotic therapy targeting H. pylori eradication, our data demonstrate that the prevalence of gastric H. pylori colonisation in various countries is inversely related to the prevalence of obesity. The obesity endemic observed in the Western world thus may partly be linked to a reduction of the prevalence of H. pylori. Alternatively, it might be speculated that hygiene factors that favour a high H. pylori prevalence have a protective role with regard to the manifestation of obesity e.g. via effects on the gastrointestinal microbiome. Thus the H. pylori prevalence could be a marker for these protective factors. While these observations require further studies, they may provide important novel insights into the mechanisms that are relevant for the obesity endemic in the developed world.
Authorship
Guarantor of the article: None.
Author contributions: Nele Lender: study concept, data extraction and interpretation of data, drafting of the manuscript. Nicholas J. Talley: study concept, interpretation of data and critical revision of manuscript. Paul Enck, Sebastian Haag, Stephan Zipfel, Mark Morrsion: interpretation of data and critical revision of manuscript. Sebastian Haag- interpretation of data and critical revision of manuscript. Gerald Holtmann: study idea, concept and design, analysis and interpretation of data, drafting of the manuscript and critical revision of the manuscript. All authors approved the final version of the manuscript.
Acknowledgement
Declaration of personal and funding interests: None.