https://orcid.org/0000-0002-6783-4670
Figures
Abstract
Associated with gastritis, peptic-ulcer disease, and gastric carcinoma, Helicobacter pylori (H. pylori) also has been associated with decreased cognitive function and dementia. In this study, we used data from the UK Biobank to further examine associations between H. pylori seropositivity and serointensity and performance on several cognitive tasks in adults 40 to 70 years of age (M = 55.3, SD = 8.1). In these analyses, H. pylori seropositivity (i.e., either positive or negative for H. pylori) and serointensity (concentration of antibodies against H. pylori antigens) in adjusted models were associated with worse function on tasks of Numeric memory, Reasoning, and errors on the Pairs matching test but better function on the Tower rearrangement task. Together, these findings suggest that H. pylori seropositivity and serointensity might be associated with worse cognitive function in this age group.
Citation: Erickson LD, White DS, Bassett P, Gale SD, Brown BL, Hedges D (2023) Cognitive function in UK adults seropositive for Helicobacter pylori. PLoS ONE 18(6): e0286731. https://doi.org/10.1371/journal.pone.0286731
Editor: Calogero Edoardo Cicero, University of Catania Department of Surgical and Medical Sciences Advanced Technologies GF Ingrassia: Universita degli Studi di Catania Dipartimento di Scienze Mediche Chirurgiche e Tecnologie Avanzate GF Ingrassia, ITALY
Received: December 16, 2022; Accepted: May 22, 2023; Published: June 7, 2023
Copyright: © 2023 Erickson et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data are available from the UK Biobank Institutional Data Access / Ethics Committee (contact via http://www.ukbiobank.ac.uk) for researchers who meet the criteria for access to confidential data.
Funding: The authors received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Helicobacter pylori (H. Pylori) is a gram-negative bacterium that infects the stomach and duodenum in humans [1]. Most commonly, infection with H. pylori occurs during childhood, with characteristics of infection ranging from asymptomatic to mild (e.g., gastric distress, abdominal pain) to severe (e.g., peptic-ulcer disease, atrophic gastritis, gastric carcinoma) [2–5]. H. pylori infection occurs most commonly through ingestion, although transmission can also occur from mother to fetus [2]. H. pylori is estimated to infect more than one-half of the human population [6]. Despite host-mediated immune response to H. Pylori, infection can persist for the lifetime of the host in the absence of antibacterial treatment [7,8].
In addition to its associations with gastric disease, H. pylori seropositivity has been associated with decreased cognitive function in some, but not all, studies. In a study of US adults using data from the US Centers for Disease Control and Prevention’s National Health and Examination Surveys (NHANES), H. pylori was adversely associated with cognitive function [9]. In contrast, a study of young to middle-aged adults (20 to 59 years of age) that also used data from the NHANES found no main effects between H. pylori and cognitive function [10], although statistically significant interactions between H. pylori seropositivity and both race-ethnicity and educational attainment predicted cognitive function. Furthermore, the study found an interaction between H. pylori seropositivity and a common parasitic infection, Toxoplasma gondii, such that the presence of both was associated with decreased cognitive function. Using data from the NHANES, Cárdenas et al. [11], however, found that older adults seropositive for H. Pylori had depressed scores on the digit-symbol substitution test. In addition, H. pylori has been associated with dementia [12,13].
Infection with H. pylori is common, and an association between H. pylori and cognitive function would represent both personal and public-health problems. However, H. pylori is potentially treatable. In that some early reports show possible associations between H. pylori and decreased cognitive function [9–11] and because of possible associations between H. pylori and dementia [12,13], we sought to further characterize the relationship between H. pylori seropositivity and serointensity and cognitive function.
Materials and methods
Study sample
We obtained data from the UK Biobank Resource, a community-based sample of adults for which the National Research Ethics Service Committee North West-Haydock provided approval (reference 11/NW/0382). All participants provided informed consent (http://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=200). Data can be obtained through application to the UK Biobank http://www.ukbiobank.ac.uk. We received approval to use deidentified data under application number 41535. Although the UK Biobank data are not representative of the UK population and the UK Biobank had a low initial response rate, research findings suggest that data from the UK Biobank can be used to investigate exposure and outcome associations [14]. The UK Biobank enrolled approximately 500,000 adults from ages 40 to 70 years from population-based registries (http://www.ukbiobank.ac.uk) and obtained clinical and demographic data from biological samples, nurse interviews, physical examinations, and questionnaires (http://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=200).
The analytic sample we used included all participants who had seropositivity and serointensity data for H. pylori, had undergone testing of cognitive function, and who had the preidentified covariates. Sample sizes varied across the individual tests of cognitive function based on availability of the relevant data (https://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/infdisease.pdf), ranging from 379 to 6,785.
Measures
H. pylori.
In the UK Biobank, H. pylori seropositivity was defined based on concentrations of antibodies against CagA, VacA, OMP, GroEL, Catalase, and UreA antigens. Because of a lab-handling error, there are two definitions of H. pylori seropositivity in these data. In the first definition, participants were considered seropositive if they were positive for two of the six antibodies measured using median fluorescence intensity (MFI). Cutoffs were as follows: 400 for CagA, 100 for VacA, 170 for OMP, 80 for GroEL, 180 for Catalase, and 130 for UreA. The second definition also required positive tests for two of the antigens using these cutoffs but excluded the CagA antigen because it was the object of the lab error. In addition to using seropositivity, we also used natural-log (ln) transformed concentrations of the CagA, VacA, OMP, GroEL, Catalase, and UreA antibodies as measures of serointensity. Finally, we standardized the logged versions of the antigens and included their mean. For participants who did not have a measure of the CagA antigen, we excluded that from this measure (https://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=23062).
Cognitive function.
The UK Biobank assessed cognitive function with a battery of neuropsychological tasks. Further descriptions of these cognitive tasks can be found elsewhere: https://biobank.ctsu.ox.ac.uk/crystal/label.cgi?id=100026. Briefly, Numeric memory, a measure of working memory, requires participants to view an increasing longer string of numbers and then after a brief delay to write what they saw (higher score is better). Reasoning consists of questions that require logic, executive function, and problem-solving skills (higher score is better). Pairs matching, a memory task, requires memorization and recall of the location of similar cards. There were six pairs and participants were allowed to continue until they matched all six pairs correctly, which nearly all did. Thus, due to a ceiling effect, we analyzed the number of errors (Pairs matching incorrect; lower score is better). Matrix pattern completion, a visual pattern reasoning task, requires participants to view a pattern with a missing piece and select one that would complete the pattern (higher score is better). Tower rearrangement, an executive functioning task, includes planning a sequence of moves and rule-following to rearrange the stimuli into directed configurations (higher score is better). Symbol-digit substitution, which includes elements of processing speed, psychomotor function, and memory, consists of matching symbols and digits per a guide as quickly as possible during a time limit (correct matches; higher score is better). Reaction time measures average response time (in milliseconds) to correctly identify matching versus non-matching image pairs (less total time is better). Give the skewness of the data from the Reaction time test, we analyzed the natural log of these scores on this test. Trails: numeric assesses total time to click sequentially on set of numbers on the screen (less total time is better). Trails: numeric measures psychomotor speed and visual attention. Trails: alphanumeric is similar except there are both numbers and letters on the screen and the participant must click on them sequentially alternating between numbers and letters (i.e., 1, A, 2, B etc.). Performance is based on total time to complete the task (less total time is better). Trails: alphanumeric requires skills similar to those required on Trails: numeric but additionally requires an element of executive function. Information regarding reliability and validity of the UK Biobank cognitive tests can be found elsewhere [15].
Covariates.
To control for variables that potentially could confound associations between H. pylori infection and cognitive function, we adjusted all statistical models for age in years, sex (female, male), race-ethnicity (White, non-White), income (the midpoint of reported categories in 10,000 pounds/year: less than 18,000; 18,000 to 30,999; 31,000 to 51,999; 52,000 to 99,999; and 100,000£ and above), educational attainment (college degree, less than college degree), self-rated health (four-point scale from poor to excellent), body-mass index (body weight in kg/height in m2), smoking history (non-smoker, past, current), and alcohol use (six categories ranging from never to daily or almost daily).
Statistical analysis
Using Stata 17.0 (StataCorp, Stata Statistical Software, Release 17.0. College Station, Texas) for all statistical analyses, we first estimated a series of adjusted multivariable linear regression models using the nine measures of H. pylori as focal independent variables in separate models of each of the nine cognitive tests for a total of 72 models. We also estimated adjusted interaction models of age, sex, educational attainment, and income with each measure of H. pylori for each of the nine cognitive functioning measures. This is an additional 72 models for each of the four interaction models, or an additional 288 models.
With such a large number of hypothesis tests, we estimated multivariate tests using Stata’s suest command to decrease the risk of false-positive results from multiple comparisons. The suest command combined model estimates of a single predictor (e.g., H. pylori seropositivity) and the nine cognitive functioning dependent variables into a single parameter vector. The resulting vector accounts for the joint covariance structure of the outcome variables. Then, rather than traditional estimates of multivariate significance, such as Wilk’s Lambda, Pillai’s Trace, Hotelling-Lawley Tract, and Roy’s Greatest Root [16], a single multivariate test is estimated with the null hypothesis that the joint relationship between the predictor and each of the outcomes is zero. If the p value from the multivariate test was not statistically significant (that is, if the multivariate p value was greater than .05), we did not consider any associations between the focal predictor variable and an outcome variable to be statistically significant in the individual models. This is in accordance with the recommendations from Rencher and Scott’s multivariate simulation study [16,17]. Like the four traditional multivariate tests of significance, the suest command method accounts for the covariance structure of the multivariate space. This strikes a balance between protecting against alpha inflation while maintaining statistical power [16,18].
Results
Table 1 shows the demographic characteristics of the sample. The analytic sample size varied depending on the data available for the cognitive functioning variable, ranging from 379 to 6,785. In the largest analytic sample, 30.1 percent were H. pylori seropositive. The average age was 55.3 years, and women comprised 54.6% of the sample. The sample was predominantly white (95.1%), and 40.1% had obtained a college degree.
Adjusted models of the various H. pylori measures and their relationship with cognitive functioning are presented in Table 2. H. pylori seropositivity was associated with worse performance on the Reasoning (b = -.386, p < .001) and Pairs matching incorrect (b = .238, p < .01) tasks but with better performance on Tower rearrangement (b = 1.346, p < .001) task; the associated p value from the multivariate test was < .001. lnCagA was associated with worse performance on Numeric memory (b = -.087, p < .01) and Reasoning (b = -073, p < .05) tasks; the p value from the multivariate test was .002. lnVacA was associated with worse performance on the Reasoning task (b = -.079, p < .01); the p value from the multivariate test was .048. lnOMP was associated with worse Reasoning (b = -.080, p < .001); the p value from multivariate test was < .001). lnGroEL was associated with worse Reasoning (b = -.066, p < .001) but with better performance on Tower rearrangement (b = .135, p < .05); the p value from the multivariable test was .001). lnCatalase was associated with worse Reasoning (b = -.081, p < .001), but the associated p value from the multivariate test was only marginally significant at .051. lnUreA was associated with worse Reasoning (b = -.059, p < .01); the p value from the associated multivariate test was .049. The mean of the standardized antigens was associated with worse Reasoning (b = -.296, p < .001) and with better performance on the Tower rearrangement task (b = .516, p < .05); the p value from the associated multivariate test was < .001 (Table 2).
Of 32 multivariate models of interactions between H. pylori and age, sex, educational attainment, and income in predicting cognitive function, only three were statistically significant: H. pylori seropositivity and age (p = .019), ln(CagA) and age (p = .032), and ln(GroEL) and sex (p = .027) (Table 3). In these models, H. pylori seropositivity was associated with worse performance on the Pairs matching incorrect task with increasing age (S1 Table in S1 File, Fig 1) but with worse performance on the Trails numeric task with increasing age (S1 Table in S1 File, S1 Fig) compared to H. pylori seropositivity. ln(CagA) negative was associated with worse Trails numeric function with increasing age compared to ln(CagA) positive (S1 Table in S1 File, S2 Fig). Similarly, ln(CagA) negative was associated with worse Trails alphanumeric function with increasing age compared to ln(CagA) positive groups (S1 Table in S1 File, S3 Fig). Increasing ln(GroEL) decreased differences between women and men in the Tower rearrangement task (S2 Table in S1 File, S4 Fig). There were no multivariately statistically significant interactions with either educational attainment or income (S3 and S4 Tables in S1 File).
Note: N = 6785. a Models adjusted for sex, race, education, household income, self-rated health, body-mass index, smoking status, and frequency of drinking alcohol. Source: UK Biobank.
Discussion
In this sample from the UK Biobank, H. pylori seropositivity and H. pylori serointensity assessed by natural log-transformed concentrations of the H. pylori CagA, VacA, OMP, GroEL, and UreA antigens were negatively associated as a group with performance on the Numeric memory, Reasoning, and Pairs matching incorrect tasks, with better performance on the Tower rearrangement task, and with no associations with the Matrix pattern-completion, Symbol-digit substitution, Reaction-time, Trails-numeric, and Trails-alphanumeric tasks in adjusted models constrained by multivariate testing to protect against false positive from multiple comparisons. The only H. pylori antigen not statistically significantly associated with cognitive function was Catalase. Worse performance on the Reasoning task was the most consistently associated with H. pylori. There were fewer negative associations with performance on the Numeric memory and Pairs-matching incorrect tasks and none with performance on the Symbol-digit substitution, Trails numeric and Trails alphanumeric tasks. Apart from better performance on the Tower rearrangement task, this pattern suggests that executive function might be particularly vulnerable to infection with H. pylori, although we were able to examine associations between H. pylori and cognitive function in only nine tasks of cognitive function. Had more tasks including ones from other cognitive domains been available, it is possible that other patterns of potential vulnerability would have been found.
These associations between H. pylori seropositivity and serointensity add to previous findings showing associations between H. pylori and worse cognitive function [9–11] but in a sample from the United Kingdom instead of the United States. However, our results differ from a previous study that did not find associations between H. pylori seropositivity and performance on the digit-symbol coding task in older US adults, although this study did find that H. pylori seropositivity did interact with 5-methytetrahydrofolate concentrations to affect performance on the digit-symbol coding task [19], a finding that suggests a possible association between H. pylori infection and worse cognitive function.
While there were few significant findings from the interaction models, findings from models constrained by multivariate analyses to protect against false-positive results from multiple comparisons suggest that age might interact with H. pylori to affect cognitive function. The findings from these few statistically significant interaction models showed mixed results. With overall seropositivity and function on the Pairs matching incorrect task, H. pylori seropositivity was associated with worse function with increased age, which suggests the possibility that H. pylori could be a factor in the development of dementia consistent with findings showing possible associations between H. pylori and dementia [12,13]. However, the other statistically significant interaction models showed better function on Trails: numeric and Trails: alphanumeric task with increased antibody titers and age, findings suggesting a more nuanced association between H. pylori and cognitive function at least on the Trails tasks with age.
If replicated, however, the associations we found between H. pylori and decreased cognitive function suggest that H. pylori infection could be a clinically significant risk factor for cognitive dysfunction and possibly even dementia, particularly given the widespread distribution of infection with H. pylori [6]. Available treatment for H. pylori makes H. pylori infection a potentially modifiable risk factor for cognitive decline. In this regard, treatment for H. pylori decreased mortality in patients with Alzheimer disease [20], although to our knowledge no studies to date have evaluated whether treatment of H. pylori infection improves cognitive function or affects the incidence of dementia.
We did not design our study to investigate mechanisms by which H. pylori could affect cognitive function. Work from previous studies, however, has identified several potential mechanisms. Emerging evidence indicates that the composition of and disturbances of the gut microbiome can be associated with cognitive function [21]. In this regard, H. pylori colonization in the stomach and duodenum could feasibly alter the gut microbiome and bidirectional signaling with the brain to affect cognitive function. H. pylori infection can result in decreased vitamin B12 and folate absorption, which in turn has been associated with elevated concentrations of homocysteine, which have been associated with atherosclerosis [5], decreased cognitive function, and dementia [22]. In addition, altered 5-methyltetrahydrofolate [19] and folate metabolism and inflammatory markers could mediate associations between H. pylori infection and cognitive function [23]. Other possible mechanisms by which H. pylori could affect brain function leading to worse cognitive function could be due to T cell-mediated apoptosis, molecular mimicry of host structures [24], induced inflammation from influx of cytokines, platelets, acute phase proteins, or eicosanoids, or oxidative stress [25]. As H. pylori infection can possibly disrupt the blood-brain barrier via inflammatory factors and oxidative stress, additional causes of decreased cognitive function from H. pylori infection could be related to amyloid deposition induced by pathogens passing through a disrupted blood brain barrier [20]. Similarly, in rats, H. pylori itself might increase hippocampal and cortical amyloid 42 via increased gamma secretase activity [26], suggesting that H. pylori could also affect cognition via increased amyloid deposition. These putative mechanisms by which H. pylori might affect cognitive function provide additional potential avenues for treatment to mitigate against the effects of H. pylori on cognitive function.
While our study was large with sample sizes ranging from 379 to 6,785 and included analyses adjusted for several covariates that could confound associations between H. pylori and cognitive function, several factors require consideration when interpreting these findings. The study design was cross-sectional, precluding the study’s ability to determine the temporal ordering of H. pylori and cognitive function. Being unable to determine when the initial infection with H. pylori occurred, we could not address whether infection at different neurodevelopmental stages could affect the relationship between H. pylori infection and cognitive function. An important limitation of the existing research of the associations between H. pylori infection and cognitive function is that most of the findings come from samples in either the United Kingdom or the United States. Findings from samples from other regions are needed to better characterize associations between H. pylori and cognitive function. While we adjusted for nine variables that could potentially alter the association between H. pylori and cognitive function, numerous factors are associated with cognitive outcomes, leading to the possibility of residual confounding. An additional limitation of our study is that we did not adjust the statistical models for use of gastroprotective drugs, including proton-pump inhibitors, which could affect the association we found between H. pylori and cognitive function. While a study based on UK Biobank data found an association between use of proton-pump inhibitors and dementia, with the association being strongest in carriers of the apolipoprotein E4 allele [27], three meta-analyses [28–30] and a study with longitudinal neuropsychological testing [31] did not show an association between proton-pump inhibitors and dementia. Given these mixed findings, future studies regarding associations between H. pylori and cognitive function may want to consider including controlling for use of proton-pump inhibitors. Finally, our results do not address whether treatment for H. pylori would result in any changes in cognition.
In conclusion, the results from this study show that H. pylori seropositivity and serointensity are associated with worse function on tasks of Numeric memory, Reasoning, and Pairs matching incorrect but with better performance on the Tower rearrangement task. Further, research investigating whether potentially confounding variables such as the use of proton-pump inhibitors might affect the association between H. pylori and cognitive function is needed. Additional future research regarding associations between H. pylori and cognitive function from regions outside of the United Kingdom and the United States is needed as is research using longitudinal studies, possibly by using data from follow-up assessments in the UK Biobank and by linking datasets where one contains exposure data and a second, later dataset has some sort of cognitive outcome variable, including dementia.
Supporting information
S1 Fig. Interaction of H. Pylori and age on Trails: Numeric: Adjusting predictionsa from linear regression.
Note: N = 394. aModels adjusted for sex, race, education, household income, self-rated health, body-mass index, smoking status, and frequency of drinking alcohol. Source: UK Biobank.
https://doi.org/10.1371/journal.pone.0286731.s001
(TIF)
S2 Fig. Interaction of ln(CagA) and age on Trails: Numeric: Adjusting predictionsa from linear regression.
Note: N = 213. aModels adjusted for sex, race, education, household income, self-rated health, body-mass index, smoking status, and frequency of drinking alcohol. Source: UK Biobank. ln(CagA) = natural log of CagA antibody concentration.
https://doi.org/10.1371/journal.pone.0286731.s002
(TIF)
S3 Fig. Interaction of ln(CagA) and age on Trails: Alphanumeric: Adjusting predictionsa from linear regression.
Note: N = 204. aModels adjusted for sex, race, education, household income, self-rated health, body-mass index, smoking status, and frequency of drinking alcohol. Source: UK Biobank. ln(CagA) = natural log of CagA antibody concentration.
https://doi.org/10.1371/journal.pone.0286731.s003
(TIF)
S4 Fig. Interaction of ln(CagA) and age on Tower Rearrangement: Adjusting predictionsa from linear regression.
Note: N = 398. aModels adjusted for sex, race, education, household income, self-rated health, body-mass index, smoking status, and frequency of drinking alcohol. Source: UK Biobank. ln(GroEL) = natural log of GroEL antibody concentration. ln(CagA) = natural log of CagA antibody concentration.
https://doi.org/10.1371/journal.pone.0286731.s004
(TIF)
Acknowledgments
This research has been conducted using the UK Biobank Resource under Application Number 41535.
References
- 1. Marshall BJ, Warren JR. Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet. 1984;1(8390):1311–5. pmid:6145023.
- 2. McGovern LM, Boyce TG, Fischer PR. Congenital infections associated with international travel during pregnancy. J Travel Med. 2007;14(2):117–28. Epub 2007/03/21. pmid:17367482.
- 3. Parsonnet J, Friedman GD, Vandersteen DP, Chang Y, Vogelman JH, Orentreich N, et al. Helicobacter pylori infection and the risk of gastric carcinoma. The New England journal of medicine. 1991;325(16):1127–31. Epub 1991/10/17. pmid:1891020.
- 4. Rowland M, Imrie C, Bourke B, Drumm B. How should Helicobacter pylori infected children be managed? Gut. 1999;45 Suppl 1:I36–9. Epub 1999/08/24. pmid:10457035; PubMed Central PMCID: PMC1766651.
- 5. Santarelli L, Gabrielli M, Cremonini F, Santoliquido A, Candelli M, Nista EC, et al. Atrophic gastritis as a cause of hyperhomocysteinaemia. Alimentary pharmacology & therapeutics. 2004;19(1):107–11. Epub 2003/12/23. pmid:14687172.
- 6. Hooi JKY, Lai WY, Ng WK, Suen MMY, Underwood FE, Tanyingoh D, et al. Global Prevalence of Helicobacter pylori Infection: Systematic Review and Meta-Analysis. Gastroenterology. 2017;153(2):420–9. Epub 2017/05/01. pmid:28456631.
- 7. Algood HM, Cover TL. Helicobacter pylori persistence: an overview of interactions between H. pylori and host immune defenses. Clinical microbiology reviews. 2006;19(4):597–613. Epub 2006/10/17. pmid:17041136; PubMed Central PMCID: PMC1592695.
- 8. Laine L, Suchower L, Frantz J, Connors A, Neil G. Twice-daily, 10-day triple therapy with omeprazole, amoxicillin, and clarithromycin for Helicobacter pylori eradication in duodenal ulcer disease: results of three multicenter, double-blind, United States trials. Am J Gastroenterol. 1998;93(11):2106–12. Epub 1998/11/20. pmid:9820381.
- 9. Beydoun MA, Beydoun HA, Shroff MR, Kitner-Triolo MH, Zonderman AB. Helicobacter pylori seropositivity and cognitive performance among US adults: evidence from a large national survey. Psychosom Med. 2013;75(5):486–96. pmid:23697465; PubMed Central PMCID: PMC3679320.
- 10. Gale SD, Erickson LD, Brown BL, Hedges DW. Interaction between Helicobacter pylori and Latent Toxoplasmosis and Demographic Variables on Cognitive Function in Young to Middle-Aged Adults. PLoS One. 2015;10(1):e0116874. pmid:25590622; PubMed Central PMCID: PMC4295891.
- 11. Cardenas VM, Boller F, Roman GC. Helicobacter pylori, Vascular Risk Factors and Cognition in U.S. Older Adults. Brain Sci. 2019;9(12). Epub 2019/12/18. pmid:31842501; PubMed Central PMCID: PMC6955675.
- 12. Kountouras J, Tsolaki M, Gavalas E, Boziki M, Zavos C, Karatzoglou P, et al. Relationship between Helicobacter pylori infection and Alzheimer disease. Neurology. 2006;66(6):938–40. pmid:16567719.
- 13. Malaguarnera M, Bella R, Alagona G, Ferri R, Carnemolla A, Pennisi G. Helicobacter pylori and Alzheimer’s disease: a possible link. European journal of internal medicine. 2004;15(6):381–6. pmid:15522573.
- 14. Batty GD, Gale CR, Kivimaki M, Deary IJ, Bell S. Comparison of risk factor associations in UK Biobank against representative, general population based studies with conventional response rates: prospective cohort study and individual participant meta-analysis. BMJ. 2020;368:m131. Epub 2020/02/14. pmid:32051121; PubMed Central PMCID: PMC7190071
- 15. Fawns-Ritchie C, Deary IJ. Reliability and validity of the UK Biobank cognitive tests. PLoS One. 2020;15(4):e0231627. Epub 2020/04/21. pmid:32310977; PubMed Central PMCID: PMC7170235.
- 16.
Brown BL, Hendrix S, Hedges DW, Smith T. Multivariate Analysis for the Biobehavioral and Social Sciences. New York: Wiley; 2012.
- 17. Rencher AC, Scott DT. Assessing the Contribution of Individual Variables Following Rejection of a Multivariate Hypothesis. Commun Stat Simulat. 1990;19(2):535–53. WOS:A1990DZ10400012.
- 18.
Rencher AC, Christensen W. Methods of Multivariate Analysis. Third ed. New York: Wiley; 2012.
- 19. Berrett AN, Gale SD, Erickson LD, Brown BL, Hedges DW. Helicobacter pylori moderates the association between 5-MTHF concentration and cognitive function in older adults. PLoS One. 2018;13(1):e0190475. Epub 2018/01/25. pmid:29364915; PubMed Central PMCID: PMC5783346.
- 20. Kountouras J, Boziki M, Zavos C, Gavalas E, Giartza-Taxidou E, Venizelos I, et al. A potential impact of chronic Helicobacter pylori infection on Alzheimer’s disease pathobiology and course. Neurobiology of aging. 2012;33(7):e3–4. Epub 2012/02/14. pmid:22325590.
- 21. Davidson GL, Cooke AC, Johnson CN, Quinn JL. The gut microbiome as a driver of individual variation in cognition and functional behaviour. Philos Trans R Soc Lond B Biol Sci. 2018;373(1756). Epub 2018/08/15. pmid:30104431; PubMed Central PMCID: PMC6107574.
- 22. Beydoun MA, Beydoun HA, Gamaldo AA, Teel A, Zonderman AB, Wang Y. Epidemiologic studies of modifiable factors associated with cognition and dementia: systematic review and meta-analysis. BMC Public Health. 2014;14:643. Epub 2014/06/26. pmid:24962204; PubMed Central PMCID: PMC4099157.
- 23. Berrett AN, Gale SD, Erickson LD, Brown BL, Hedges DW. Folate and Inflammatory Markers Moderate the Association Between Helicobacter pylori Exposure and Cognitive Function in US Adults. Helicobacter. 2016;21(6):471–80. pmid:26935014.
- 24. D’Elios MM, Amedei A, Benagiano M, Azzurri A, Del Prete G. Helicobacter pylori, T cells and cytokines: the "dangerous liaisons". FEMS Immunol Med Microbiol. 2005;44(2):113–9. Epub 2005/05/04. pmid:15866204.
- 25. Kountouras J, Gavalas E, Zavos C, Stergiopoulos C, Chatzopoulos D, Kapetanakis N, et al. Alzheimer’s disease and Helicobacter pylori infection: Defective immune regulation and apoptosis as proposed common links. Med Hypotheses. 2007;68(2):378–88. Epub 2006/09/19. pmid:16979298.
- 26. Wang XL, Zeng J, Feng J, Tian YT, Liu YJ, Qiu M, et al. Helicobacter pylori filtrate impairs spatial learning and memory in rats and increases beta-amyloid by enhancing expression of presenilin-2. Frontiers in aging neuroscience. 2014;6:66. Epub 20140411. pmid:24782763; PubMed Central PMCID: PMC3990046.
- 27. Zhang P, Li Z, Chen P, Zhang A, Zeng Y, Zhang X, et al. Regular proton pump inhibitor use and incident dementia: population-based cohort study. BMC medicine. 2022;20(1):271. Epub 20220901. pmid:36045363; PubMed Central PMCID: PMC9434890.
- 28. Li M, Luo Z, Yu S, Tang Z. Proton pump inhibitor use and risk of dementia: Systematic review and meta-analysis. Medicine (Baltimore). 2019;98(7):e14422. pmid:30762748; PubMed Central PMCID: PMC6408083.
- 29. Ahn N, Nolde M, Krause E, Guntner F, Gunter A, Tauscher M, et al. Do proton pump inhibitors increase the risk of dementia? A systematic review, meta-analysis and bias analysis. Br J Clin Pharmacol. 2023;89(2):602–16. Epub 20221123. pmid:36331350.
- 30. Song YQ, Li Y, Zhang SL, Gao J, Feng SY. Proton pump inhibitor use does not increase dementia and Alzheimer’s disease risk: An updated meta-analysis of published studies involving 642305 patients. PLoS One. 2019;14(7):e0219213. Epub 20190702. pmid:31265473; PubMed Central PMCID: PMC6605652.
- 31. Collin BG, Raju D, Altman J, Katsikas S. The effects of proton pump inhibitors on neuropsychological functioning. Appl Neuropsychol Adult. 2022;29(6):1403–12. Epub 20210302. pmid:33651951.