Gut infection and inflammation: a small link, but is it a cause?
Helicobacter pylori infection, peptic ulcer, gastritis and risk of Parkinson's disease: A nationwide matched cohort study.
AI-authored reading companion. Supporting quotations match the source text; this does not verify the interpretation. No expert review yet.
Sara Licaj, Alexander Ploner, Agnieszka Szwajda, Sara Hägg, Fang Fang, Jonas F Ludvigsson, Karin Wirdefeldt · Journal of Parkinson's disease · 2026
Adapted from the original publication. License: https://creativecommons.org/licenses/by/4.0/This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage).
This Swedish study found a small link between some stomach conditions and a later Parkinson’s diagnosis. It could not establish which came first: the digestive condition might contribute to disease, or it might be an early consequence of disease already developing. The study did not test whether treating a stomach infection would prevent Parkinson’s.
Read the supporting text
In conclusion, our findings suggest that H. pylori infection and gastritis may have some relevance for PD but are unlikely to be major contributors to its pathogenesis. Future studies are needed to clarify whether the associations we found reflect a true causal relationship, prodromal PD symptoms, or shared underlying mechanisms influencing both gastrointestinal impairment and neurodegeneration. Considering other factors related to the gut-brain axis, such as microbiome composition or lifestyle habits, might help to investigate the complex interplay underlying PD pathogenesis further.
This study used a population-based matched cohort design, where three distinct cohorts were derived from the independent assessment of each exposure. Exposed individuals were defined as anyone with a biopsy positive for one of the following diagnoses: H. pylori infection, peptic ulcer or gastritis. As individuals may have more than one biopsy record, the index date of H. pylori infection, peptic ulcer, or gastritis was defined as the first-ever date of biopsy leading to any of these diagnoses. Every exposed individual was matched to up to five referents who were free of gastrointestinal biopsy at the index date of the exposed individual. Referents were randomly selected from the general population and individually matched to the exposed individual on age, sex, calendar year and county of residence. The index date of the exposed individual was used as the index date for their matched referents. Owing to the dynamic nature of the cohort, unexposed individuals could be matched multiple times. Those who transitioned to exposed status were excluded from further matching and re-indexing. Additional details on exposure trajectories in our study cohort and re-indexing can be found in Table S1.
The main analysis revealed a marginal association between H. pylori infection (HR: 1.14, 95%CI: 1.02–1.28) and an increased risk of developing PD later in life, with no statistically significant association for peptic ulcer (HR: 1.08, 95%CI: 0.98–1.19) (Table 2). For simplicity of presentation, the time-constant HR for gastritis was 1.14 (95%CI: 1.09–1.19).
Reverse causation might also explain our results, as gastrointestinal symptoms are frequent in patients with PD and there is often a delay between the clinical PD onset and diagnosis recorded in the register. 36 Our findings might also reflect early prodromal symptoms of PD, such as impaired gut physiology or immune response, leading to increased susceptibility to gastrointestinal pathology. 4 Prodromal gastrointestinal symptoms can contribute to increased healthcare seeking, potentially leading to an earlier diagnosis of PD. Although we tried to control for surveillance bias by adjusting for number of healthcare visits, we cannot exclude the possibility that such bias influenced our findings.
In brief
Are stomach infection or inflammation linked to a later Parkinson’s diagnosis? This Swedish study examined H. pylori, a stomach bacterium; gastritis, inflammation of the stomach lining; and peptic ulcers, sores in the stomach or nearby intestine.
Read the supporting text
Therefore, the present research aims to examine the potential role of H. pylori infection, gastritis and peptic ulcer in PD risk, by conducting a large-scale population-based study that relies on biopsy-based diagnoses. We hypothesized that H. pylori infection, gastritis, and peptic ulcer increase the risk of PD.
Researchers linked biopsy records to national health records and compared people with each condition with similar people who had no gastrointestinal biopsy. This was an observational follow-up study, with overlapping groups, not a treatment trial.
Read the supporting text
This study used a population-based matched cohort design, where three distinct cohorts were derived from the independent assessment of each exposure. Exposed individuals were defined as anyone with a biopsy positive for one of the following diagnoses: H. pylori infection, peptic ulcer or gastritis. As individuals may have more than one biopsy record, the index date of H. pylori infection, peptic ulcer, or gastritis was defined as the first-ever date of biopsy leading to any of these diagnoses. Every exposed individual was matched to up to five referents who were free of gastrointestinal biopsy at the index date of the exposed individual. Referents were randomly selected from the general population and individually matched to the exposed individual on age, sex, calendar year and county of residence. The index date of the exposed individual was used as the index date for their matched referents. Owing to the dynamic nature of the cohort, unexposed individuals could be matched multiple times. Those who transitioned to exposed status were excluded from further matching and re-indexing. Additional details on exposure trajectories in our study cohort and re-indexing can be found in Table S1.
H. pylori and gastritis each showed an adjusted hazard ratio of 1.14: a small relative increase in the rate of Parkinson’s diagnosis over follow-up, not a 14-percentage-point increase in lifetime risk. The ulcer result was not statistically clear.
Read the supporting text
The main analysis revealed a marginal association between H. pylori infection (HR: 1.14, 95%CI: 1.02–1.28) and an increased risk of developing PD later in life, with no statistically significant association for peptic ulcer (HR: 1.08, 95%CI: 0.98–1.19) (Table 2). For simplicity of presentation, the time-constant HR for gastritis was 1.14 (95%CI: 1.09–1.19).
Early, undiagnosed Parkinson’s might itself change digestion or increase healthcare visits. That could help explain the association. Adjustments for healthcare use cannot fully settle which condition came first biologically.
Read the supporting text
Reverse causation might also explain our results, as gastrointestinal symptoms are frequent in patients with PD and there is often a delay between the clinical PD onset and diagnosis recorded in the register. 36 Our findings might also reflect early prodromal symptoms of PD, such as impaired gut physiology or immune response, leading to increased susceptibility to gastrointestinal pathology. 4 Prodromal gastrointestinal symptoms can contribute to increased healthcare seeking, potentially leading to an earlier diagnosis of PD. Although we tried to control for surveillance bias by adjusting for number of healthcare visits, we cannot exclude the possibility that such bias influenced our findings.
This is a useful clue for studying the gut–brain connection, but it does not establish that these conditions cause Parkinson’s or that treating them prevents it. The authors judge them unlikely to be major contributors.
Read the supporting text
In conclusion, our findings suggest that H. pylori infection and gastritis may have some relevance for PD but are unlikely to be major contributors to its pathogenesis. Future studies are needed to clarify whether the associations we found reflect a true causal relationship, prodromal PD symptoms, or shared underlying mechanisms influencing both gastrointestinal impairment and neurodegeneration. Considering other factors related to the gut-brain axis, such as microbiome composition or lifestyle habits, might help to investigate the complex interplay underlying PD pathogenesis further.
When was the research done?
The records cover diagnoses from 1987 for gastritis and ulcers, and from 1990 for H. pylori, through 2017. Follow-up continued until the end of 2021. These are the study’s observation dates; the paper was published in 2026.
Read the supporting text
As the NPR reached nationwide coverage for inpatient care in 1987, the start of the study period for gastritis and peptic ulcer was 1987-01-01, to ensure reliable PD detection. 35 For H. pylori infection, we set the study start in 1990-01-01, as diagnoses before this time were uncommon in Sweden and were judged to be unreliable. In all three cohorts, follow-up started at the index date until date of PD diagnosis, 31 December 2021, death, emigration or re-indexing date, whichever came first.
Where did it happen?
The researchers used Swedish health records. ESPRESSO is a Swedish research database of reports from biopsies: small samples of digestive-tract tissue examined under a microscope. It brings together records from 28 pathology departments, where tissue samples are examined to help identify disease. Linking these records to national health registers let the team check which people later received a Parkinson’s diagnosis.
Read the supporting text
The study population was derived from the ESPRESSO (Epidemiology Strengthened by HistoPathology Reports in Sweden) cohort, consisting of 6.1 million histopathology records among 2.1 million unique individuals from 28 pathology departments across Sweden between 1965 and 2017. 32 To obtain additional health and sociodemographic data, the ESPRESSO cohort was linked to Swedish national registers such as the National Patient Register (NPR), Causes of Death Register (CDR), Prescribed Drug Register (PDR) and the Longitudinal Integrated Database for Health Insurance and Labor Market Studies (LISA).32–34
The full story
Why look in the stomach?
The researchers wanted to investigate events that might come before a Parkinson’s diagnosis. Digestive problems can appear well before the movement symptoms that make the disease recognizable. One possible explanation is that changes in the gut help set disease processes in motion; another is that an already-developing brain and nerve disorder changes the gut. The paper focuses on three specific conditions, rather than treating every digestive problem as the same thing: H. pylori infection, gastritis and peptic ulcers. It asks whether people diagnosed with these conditions are subsequently diagnosed with Parkinson’s more often.
Read the supporting text
Gastrointestinal symptoms are frequent in Parkinson's disease (PD) and may precede diagnosis by decades, suggesting a possible role in its pathogenesis. Patients with PD have higher rates of Helicobacter pylori (H. pylori) infection than the general population, although its potential contribution to PD development remains unclear. Similarly, evidence linking gastritis and peptic ulcer to PD risk is limited.
Therefore, the present research aims to examine the potential role of H. pylori infection, gastritis and peptic ulcer in PD risk, by conducting a large-scale population-based study that relies on biopsy-based diagnoses. We hypothesized that H. pylori infection, gastritis, and peptic ulcer increase the risk of PD.
Following people over time
This was a matched cohort study. A cohort is a group followed over time; matching means selecting comparison people who resemble the people of interest in certain important ways. The investigators used ESPRESSO, a Swedish database of reports from digestive-tract biopsies. A biopsy is a small tissue sample examined under a microscope. Linking these reports to national health records allowed the team to follow later diagnoses. Each person with a biopsy-confirmed condition was matched to up to five people of similar age and the same sex, calendar year and county. The comparison people had no gastrointestinal biopsy at that starting point. That does not mean tests had proved them free of infection or inflammation. No researcher assigned a person to become infected, and no treatment was being tested.
Read the supporting text
The study population was derived from the ESPRESSO (Epidemiology Strengthened by HistoPathology Reports in Sweden) cohort, consisting of 6.1 million histopathology records among 2.1 million unique individuals from 28 pathology departments across Sweden between 1965 and 2017. 32 To obtain additional health and sociodemographic data, the ESPRESSO cohort was linked to Swedish national registers such as the National Patient Register (NPR), Causes of Death Register (CDR), Prescribed Drug Register (PDR) and the Longitudinal Integrated Database for Health Insurance and Labor Market Studies (LISA).32–34
This study used a population-based matched cohort design, where three distinct cohorts were derived from the independent assessment of each exposure. Exposed individuals were defined as anyone with a biopsy positive for one of the following diagnoses: H. pylori infection, peptic ulcer or gastritis. As individuals may have more than one biopsy record, the index date of H. pylori infection, peptic ulcer, or gastritis was defined as the first-ever date of biopsy leading to any of these diagnoses. Every exposed individual was matched to up to five referents who were free of gastrointestinal biopsy at the index date of the exposed individual. Referents were randomly selected from the general population and individually matched to the exposed individual on age, sex, calendar year and county of residence. The index date of the exposed individual was used as the index date for their matched referents. Owing to the dynamic nature of the cohort, unexposed individuals could be matched multiple times. Those who transitioned to exposed status were excluded from further matching and re-indexing. Additional details on exposure trajectories in our study cohort and re-indexing can be found in Table S1.
The H. pylori comparison included 48,258 affected people and 231,805 referents; the ulcer comparison included 52,761 and 274,313; and the gastritis comparison included 242,325 and 1,053,688. These totals must not be added as though everyone was unique: people could appear in more than one condition group, and some referents could be matched more than once. People already diagnosed with Parkinson’s, and those aged 35 or younger at entry, were excluded. Follow-up ended at a Parkinson’s diagnosis, death, emigration, re-indexing or the end of 2021. Median follow-up varied from 6.7 to 9.3 years across groups.
Read the supporting text
Table 1 summarizes the three cohorts included in the study. The H. pylori cohort was the smallest, including 48,258 exposed individuals and 231,805 matched referents. The peptic ulcer cohort included 52,761 exposed individuals and 274,313 referents, whereas the gastritis cohort consisted of 242,325 exposed individuals and 1,053,688 referents. There was an overlap among exposed individuals across the cohorts. Specifically, 3397 study participants were diagnosed with all three conditions. Among individuals diagnosed with H. pylori infection, 42,990 (89%) were also diagnosed with gastritis, while 4417 (9.2%) were also diagnosed with peptic ulcer. There were 23,971 study participants diagnosed with both peptic ulcer and gastritis, corresponding to 45.4% of individuals diagnosed with peptic ulcer and 9.9% of those diagnosed with gastritis. Among those with H. pylori infection, 82% received their diagnosis on the same date as they received a diagnosis of either gastritis or peptic ulcer, or both, suggesting that these conditions were detected during the same endoscopic examination.
To ensure the capture of relevant information on outcome and covariates, only individuals who had lived continuously in Sweden for at least 5 years before the diagnosis date of the respective gastrointestinal disease were included in the study. Individuals with an NPR-based PD diagnosis before the exposure diagnosis date were excluded, as we were only interested in the incident diagnosis of PD. We further excluded individuals who died before the index date and those aged 35 years or younger at the index date, as PD diagnosed at a very young age is typically related to genetic causes.
As the NPR reached nationwide coverage for inpatient care in 1987, the start of the study period for gastritis and peptic ulcer was 1987-01-01, to ensure reliable PD detection. 35 For H. pylori infection, we set the study start in 1990-01-01, as diagnoses before this time were uncommon in Sweden and were judged to be unreliable. In all three cohorts, follow-up started at the index date until date of PD diagnosis, 31 December 2021, death, emigration or re-indexing date, whichever came first.
The median follow-up time ranged between 6.7 and 9.3 years across the three cohorts.
What the results actually tell us
The outcome was a newly recorded Parkinson’s diagnosis, identified by diagnosis codes in a national register. It was not a direct measurement of the earliest biological start of disease. The statistical model compared diagnosis rates over time while accounting for age and the matched groups. The team also adjusted for country of birth, education, previous healthcare contacts and a long-term lung condition called chronic obstructive pulmonary disease. The team used this condition as an imperfect stand-in for heavy smoking because direct smoking histories were unavailable. These adjustments help make the groups more comparable, but cannot remove differences that were not measured well.
Read the supporting text
Incident PD was defined by diagnoses in the NPR according to the following Swedish International Classification of Diseases (ICD) codes: 342.00 (ICD-8: 1968–1986), 332.0 (ICD-9: 1987–1996), and G20 (ICD-10: 1997 onwards). Both inpatient and outpatient contacts were included, as well as primary and secondary diagnoses in the NPR. For individuals with multiple PD diagnoses, the earliest recorded diagnosis date was considered the diagnosis date of PD. The accuracy of PD diagnoses in the NPR has previously been reported to be 70.8%, increasing to 83.0% when diagnoses are restricted to primary diagnoses only. 36
We quantified the associations between H. pylori infection, peptic ulcer, gastritis and the risk of PD as hazard ratios (HRs) with 95% confidence intervals (CIs) by comparing the exposed individuals to the unexposed ones. Estimates were based on Cox regression models with attained age as the underlying time scale, stratified by matched index sets. All models were adjusted for country of birth, educational level, number of healthcare contacts prior to the index date, and COPD.
As smoking is negatively associated with PD, but positively associated with H. pylori infection, gastritis and peptic ulcer, we included chronic obstructive pulmonary disease (COPD) diagnoses as a proxy for heavy smoking.37–40 We retrieved them from the NPR via ICD codes 491, 492, 496 (ICD-9) and J41–44 (ICD-10). Only the first COPD diagnoses after age 40 were included, as disease onset before this age is likely caused by factors other than smoking. Both prevalent COPD diagnoses at baseline and diagnoses occurring during follow-up were considered; in the latter case, we treated COPD as a time-varying exposure and split the time at risk at the first COPD diagnosis after baseline.
The adjusted hazard ratio was 1.14 for H. pylori, with a 95% confidence interval of 1.02–1.28. A ratio of 1 would mean no difference in the modeled rate; 1.14 means a roughly 14% higher rate relative to the comparison group. It does not mean 14% of infected people developed Parkinson’s. The confidence interval describes statistical uncertainty under the model, not protection against bias. For ulcers, the estimate was 1.08, with an interval of 0.98–1.19. Because that range includes 1, the study did not establish a clear increase for ulcers. The overall gastritis estimate was also 1.14, but its association changed with time.
Read the supporting text
The main analysis revealed a marginal association between H. pylori infection (HR: 1.14, 95%CI: 1.02–1.28) and an increased risk of developing PD later in life, with no statistically significant association for peptic ulcer (HR: 1.08, 95%CI: 0.98–1.19) (Table 2). For simplicity of presentation, the time-constant HR for gastritis was 1.14 (95%CI: 1.09–1.19).
For gastritis, we observed time-varying effects, with the highest increase in PD risk occurring within the first two years after a gastritis diagnosis (HR: 1.28; 95%CI: 1.16–1.41). This association remained statistically significant for up to ten years after the diagnosis (Table 2).
The unadjusted Parkinson’s diagnosis rate in the H. pylori group was about 1.0 per 1,000 person-years, compared with 0.9 in its reference group. Person-years combine how many people were followed with how long they were followed: 1,000 people followed for one year contribute 1,000 person-years. The difference is approximately 0.1 diagnosis per 1,000 person-years, calculated from the rounded rates in the paper. It is useful for understanding scale, but it is neither an adjusted treatment effect nor an individual prediction. For gastritis the reported rates were 1.1 versus 1.0; for ulcers, 1.3 versus 1.2.
Read the supporting text
The number of incident PD diagnoses was 2283 in the H. pylori cohort, 4214 in the peptic ulcer cohort, and 14,369 in the gastritis cohort. The incident PD rates per 1000 person-years were 1.0 for the H. pylori-exposed cohort (0.9 for referents), 1.3 in the peptic ulcer exposed cohort (1.2 for referents) and 1.1 for individuals exposed to gastritis (1.0 for referents). Mortality rates, however, were higher in the exposed groups than in the reference group across all cohorts, with peptic ulcer showing the highest rate (9.1 per 100 person-years).
Could the link run the other way?
For gastritis, the association was strongest during the first two years after diagnosis: a hazard ratio of 1.28. It remained statistically detectable through ten years, but the later intervals did not show a clear increase. This timing is compatible with more than one explanation. Inflammation could contribute to disease processes, but early Parkinson’s could also produce digestive symptoms and bring someone into medical care sooner. The researchers repeated the analysis using only hospital contacts where Parkinson’s was the main diagnosis. Gastritis results were similar, while estimates for H. pylori and ulcers became smaller. That sensitivity to the outcome definition belongs alongside the headline result.
Read the supporting text
For gastritis, we observed time-varying effects, with the highest increase in PD risk occurring within the first two years after a gastritis diagnosis (HR: 1.28; 95%CI: 1.16–1.41). This association remained statistically significant for up to ten years after the diagnosis (Table 2).
10 - ≤ 15 years | 1.07 | 0.95, 1.20 | 0.3
15 - ≤ 20 years | 0.94 | 0.80, 1.10 | 0.4
20+ years | 1.01 | 0.80, 1.28 | >0.9
Restricting to the definition of PD as hospital contact with PD as the primary diagnosis only, the results were unchanged in the gastritis cohort, but the HR estimates decreased in the H. pylori infection and peptic ulcer cohorts (Table S4).
Reverse causation might also explain our results, as gastrointestinal symptoms are frequent in patients with PD and there is often a delay between the clinical PD onset and diagnosis recorded in the register. 36 Our findings might also reflect early prodromal symptoms of PD, such as impaired gut physiology or immune response, leading to increased susceptibility to gastrointestinal pathology. 4 Prodromal gastrointestinal symptoms can contribute to increased healthcare seeking, potentially leading to an earlier diagnosis of PD. Although we tried to control for surveillance bias by adjusting for number of healthcare visits, we cannot exclude the possibility that such bias influenced our findings.
The large national database and biopsy confirmation are strengths: the exposure measure is more concrete than someone recalling a stomach complaint years later. But the same approach misses people diagnosed by breath tests, blood tests or clinical judgment, and people with silent infection. It may favor more severe or investigated cases. Parkinson’s register diagnoses can also be wrong or delayed. Lifestyle differences were not fully available, and using lung-disease diagnoses as a substitute for smoking histories mainly identifies heavy smokers. The condition groups also had different mortality rates. The authors acknowledge that remaining confounding—other factors affecting both exposure and outcome—could explain part of the small association.
Read the supporting text
This study has several strengths. The large-scale and high-quality Swedish registers allow for comprehensive population coverage that minimizes selection bias. The prospective cohort design and long follow-up period enhance the validity and generalizability of our findings. In addition, the biopsy-confirmed exposure assessment strengthens the accuracy of exposure classification, minimizing the risk of false positives.
Nevertheless, our exposure ascertainment method can also represent a limitation. Biopsy is not always the primary method for diagnosing H. pylori infection, as non-invasive procedures (e.g., urea breath test, serology) are also commonly used, especially in young individuals, and some patients may even be diagnosed on a clinical basis without prior testing. H. pylori infection is often asymptomatic and, although less likely, this may also apply to gastritis and peptic ulcer. Thus, biopsy-confirmed definition of H. pylori, peptic ulcer, and gastritis might have led to under-ascertainment and over-representation of more severe cases.
The use of COPD as a proxy for smoking status is also a limitation, as it relies on ICD codes and mostly captures heavy smokers. Lastly, despite the matched design and the effort to adjust for relevant covariates, residual confounding might have contributed to our findings, as we lacked data on potentially relevant confounders such as lifestyle factors.
Mortality rates, however, were higher in the exposed groups than in the reference group across all cohorts, with peptic ulcer showing the highest rate (9.1 per 100 person-years).
What should happen next?
The study provides evidence to compare with stronger claims from smaller or differently designed gut studies. It does not identify a single root cause, measure bacteria traveling to the brain, or test whether eliminating H. pylori changes Parkinson’s incidence. The authors discuss several possible mechanisms, but those explanations are not directly demonstrated by these register comparisons. For our dataset, the appropriate interpretation is a small human observational association with a substantial question about direction of cause and effect. A useful next research step is to distinguish a causal gut process from an early sign of disease or a shared underlying process.
Read the supporting text
In conclusion, our findings suggest that H. pylori infection and gastritis may have some relevance for PD but are unlikely to be major contributors to its pathogenesis. Future studies are needed to clarify whether the associations we found reflect a true causal relationship, prodromal PD symptoms, or shared underlying mechanisms influencing both gastrointestinal impairment and neurodegeneration. Considering other factors related to the gut-brain axis, such as microbiome composition or lifestyle habits, might help to investigate the complex interplay underlying PD pathogenesis further.
Is this research continuing?
We have not yet checked for a later publication, registered follow-up study or update from this team. The next steps below are proposals in this paper, not confirmation that the work is underway.
Based on the full acquired article text, including available tables and figure captions. Figure images, raw data and separate supplementary files were not independently examined. This is an explanatory adaptation, not a line-by-line translation.