Use of Glucagon-Like Peptide-1 Receptor Agonists and Risk of Parkinson's Disease: Scandinavian Cohort Study
Engström A, Svanström H, Hviid A, Eliasson B, Gudbjörnsdottir S, Wintzell V
Diabetes, Obesity and Metabolism (2026) · n=347,026
Across 158,961 new users of GLP-1 receptor agonists in Denmark, Norway and Sweden — liraglutide-dominant (72.9% of follow-up) — compared with 188,065 new users of sulfonylureas, GLP-1RA use was associated with a 19% lower incidence of Parkinson's disease over follow-up (adjusted HR 0.81; 95% CI 0.68–0.96), with the signal strengthening to HR 0.74 (0.60–0.93) after excluding or censoring DPP-4 inhibitor users.
This is a nationwide multi-country active-comparator cohort study from Bjorn Pasternak's Karolinska pharmacoepidemiology group, using linked health registers in Denmark, Norway and Sweden to test whether GLP-1 receptor agonist use is associated with incident Parkinson's disease. The Scandinavian register infrastructure is one of the strongest pharmacoepidemiology data environments globally — universal healthcare, mandatory prescription registration, and linked clinical-outcome registries.
The investigators identified 158,961 new users of GLP-1 receptor agonists and 188,065 new users of sulfonylureas, aged 45 years or older, in a new-user, active-comparator design that is the methodological gold standard for non-randomised pharmacoepidemiology. The GLP-1RA cohort follow-up was 72.9% liraglutide, 13.4% semaglutide, 7.3% exenatide, 5.1% dulaglutide, and 1.3% lixisenatide — making this primarily a liraglutide-evidence paper. Mean age was 65; 43% female.
The primary outcome was incident Parkinson's disease, defined as a first-ever diagnosis of Parkinson's disease (ICD-10 G20) or Parkinson's disease dementia (ICD-10 F02.3). Cox regression with propensity-score weighting was used.
Incidence rates were 5.2 vs 8.0 cases per 10,000 person-years (GLP-1RA vs sulfonylurea), corresponding to an adjusted HR of 0.81 (95% CI 0.68–0.96). A 2-year lag-time analysis (designed to mitigate reverse-causation from prodromal symptoms influencing prescribing) returned HR 0.84 (0.70–1.02). Excluding or censoring DPP-4 inhibitor users (a methodological concern given DPP-4 inhibition affects GLP-1 endogenous levels) strengthened the signal to HR 0.74 (0.60–0.93).
This is observational pharmacoepidemiology, not a randomised trial — propensity-score weighting reduces but does not eliminate residual confounding. The active-comparator new-user design is the methodological best-practice for this class of analysis, and the choice of sulfonylureas as comparator means the confound is the contrast between two pharmacotherapy classes rather than a treated-versus-untreated contrast, which sharply reduces the indication-by-confounding-by-severity problem. But unmeasured confounding from health-seeking behaviour, lifestyle, and prodromal motor symptoms cannot be fully excluded.
The cohort is liraglutide-dominant; the signal for semaglutide and tirzepatide cannot be cleanly estimated from this data. The 2-year lag-time sensitivity analysis loses statistical significance — consistent with either a genuine but smaller effect, or with reverse-causation/protopathic-bias contribution to the primary estimate. The corpus should weight the primary HR appropriately conservatively against this sensitivity.
The mechanism of any putative GLP-1RA Parkinson's-protective effect — neuroinflammation suppression, α-synuclein pathway interaction, mitochondrial protection — is preclinically supported but not directly tested in this study. The cohort is Scandinavian, predominantly white European; generalisability to non-European populations is the standard external-validity caveat. Follow-up duration is not stated in the abstract.
The corpus does not yet have a dedicated liraglutide peptide slug; this entry is tagged semaglutide as the closest peptide-level anchor, with the understanding that the underlying evidence is primarily liraglutide-driven.
07·Member discussion
No member discussion yet.
Member-only conversation lives here — cycle notes, practitioner commentary, pattern-matching. Be the first paying member to start the thread.