Neurology · Medical Literature
Gabapentinoid Use and Risk of Alzheimer's Disease-Related Dementias: A Systematic Review and Meta-analysis of Observational Studies.
Abstract
Gabapentinoids, particularly gabapentin and pregabalin, are widely prescribed for neuropathic pain and other chronic conditions, including in populations at risk for cognitive decline. Their long-term cognitive safety and potential association with Alzheimer's disease and related dementias (ADRD) remain uncertain. We examined whether gabapentinoid use is associated with ADRD risk and assessed between-study heterogeneity and potential study-level modifiers. PubMed, Embase, and the Cochrane Library were searched from inception through November 2025. We included observational studies comparing gabapentinoid users with non-users or non-gabapentinoid comparators and reporting incident ADRD. Risk of bias was assessed using ROBINS-E. A random-effects meta-analysis was performed using odds ratios. Subgroup and meta-regression analyses explored study-level modifiers. Certainty of evidence was assessed using GRADE (Grading of Recommendations Assessment, Development and Evaluation). The protocol was registered in PROSPERO (CRD420251240055). Five observational studies (325,245 participants) were included. Gabapentinoid use was associated with modestly higher estimated odds of ADRD (odds ratio, 1.28; 95% confidence interval 1.10-1.50; P = 0.002), with high heterogeneity (I2 = 90.3%). This association was attenuated when restricted to studies with a low risk of bias (odds ratio, 1.23; 95% confidence interval 0.83-1.84). Studies enrolling younger populations yielded larger estimates than those with older cohorts, though the subgroup difference was not statistically significant. Meta-regression identified mean age as a study-level moderator, accounting for 60.2% of between-study variance. Certainty of evidence was low overall. Gabapentinoid use was associated with modestly higher estimated odds of ADRD in pooled observational data. High heterogeneity, residual confounding, and confounding by indication preclude causal inference. These findings support cautious prescribing, particularly for off-label and long-term use, and underscore the need for prospective studies. PROSPERO protocol number: CRD420251240055.
Clinical relevance
Among adults prescribed gabapentinoids (gabapentin or pregabalin), is use associated with an increased incidence of Alzheimer's disease and related dementias (ADRD) compared with non-users or non-gabapentinoid comparators? __SNAPSHOT__ {"studyDesign":"Systematic review and random-effects meta-analysis of observational studies (PROSPERO registration: CRD420251240055)","setting":"Multiple countries and settings (individual study-level settings not reported in abstract); searches of PubMed, Embase, and Cochrane Library from inception through November 2025","sampleSize":"n = 325,245 (across 5 included observational studies); individual study allocations not reported in abstract","population":"Adults prescribed gabapentinoids (gabapentin and/or pregabalin), predominantly for neuropathic pain and other chronic conditions, including populations at elevated baseline risk for cognitive decline; specific age ranges and comorbidity profiles of included studies not reported in abstract","intervention":"Gabapentinoid use (gabapentin and/or pregabalin); dosing, duration, and frequency of prescriptions vary across included studies and are not uniformly reported in the abstract","comparator":"Non-users of gabapentinoids or users of non-gabapentinoid comparators; specific comparator definitions vary by included study","primaryOutcome":"Incident Alzheimer's disease and related dementias (ADRD)","followUp":"Not reported in abstract; varies across included observational studies","fundingSource":"Not reported in abstract"}
Key findings
The pooled random-effects meta-analysis of 5 observational studies (n = 325,245) found that gabapentinoid use was associated with a statistically significant, modestly higher odds of incident ADRD compared with non-users or non-gabapentinoid comparators (OR 1.28; 95% CI 1.10–1.50; P = 0.002). This represents a 28% increase in the odds of ADRD among gabapentinoid users. Absolute risk difference, NNH, and relative risk data are not reported in the abstract and cannot be calculated without event-rate data from individual studies. Critically, between-study heterogeneity was very high (I² = 90.3%), indicating that the pooled estimate should be interpreted with substantial caution, as a large proportion of observed variance reflects true differences across studies rather than chance. Risk of bias was assessed using the ROBINS-E tool (results not detailed in abstract). Certainty of evidence was rated using GRADE (rating not specified in abstract). Subgroup and meta-regression analyses were performed to explore study-level modifiers of effect; specific subgroup results are not reported in the abstract. Clinically important adverse events beyond ADRD incidence are not reported.
Practice implications
This meta-analysis establishes a statistically significant association between gabapentinoid use and modestly elevated odds of incident ADRD (OR 1.28). However, several critical limitations constrain the strength of any causal inference. First, all included studies are observational; confounding by indication is a paramount concern, as gabapentinoids are often prescribed for conditions — including chronic pain, diabetic neuropathy, and epilepsy — that are themselves independently associated with cognitive decline. The very high heterogeneity (I² = 90.3%) means the pooled OR is an unstable summary across highly disparate studies, and the true effect size in any given clinical context may differ substantially from 1.28. Second, the evidence base consists of only 5 studies, limiting the ability to detect and adjust for publication bias or to perform robust subgroup analyses by drug (gabapentin vs. pregabalin), dose, duration, indication, or age. Third, reverse causation cannot be excluded: prodromal dementia may increase pain perception or prescriber propensity to prescribe gabapentinoids, inflating apparent risk. Fourth, the GRADE certainty rating is not disclosed in the abstract, though given the observational design, high heterogeneity, and likelihood of residual confounding, it is expected to be low or very low. The findings are most directly applicable to adults with chronic conditions for whom gabapentinoids are commonly prescribed; populations with pre-existing cognitive vulnerability may be at higher risk but are not specifically characterized. This study does NOT establish causality, does NOT define a safe dose or duration threshold, and does NOT provide evidence sufficient to justify abrupt discontinuation of gabapentinoids in patients with established clinical benefit. Practice should not change based on this meta-analysis alone; rather, it should prompt heightened vigilance for cognitive symptoms in long-term gabapentinoid users and support the conduct of adequately adjusted prospective cohort studies or pharmacoepidemiological analyses using active comparators and new-user designs. NOTE: This summary is based on abstract-only data; full-text data including GRADE ratings, ROBINS-E domain-level results, subgroup findings, and meta-regression outputs are not available for review. __LIMITATIONS__ Only 5 observational studies included — a very small evidence base for meta-analytic pooling, limiting statistical power and reliability of subgroup analyses.,Extremely high between-study heterogeneity (I² = 90.3%) renders the pooled OR highly unstable and of limited standalone clinical interpretability; sources of heterogeneity are not characterizable from the abstract alone.,All included studies are observational; confounding by indication is a critical and likely unresolvable threat — patients prescribed gabapentinoids often have conditions (e.g., neuropathic pain, epilepsy, diabetic neuropathy) independently associated with ADRD risk.,Reverse causation (prodromal dementia driving gabapentinoid prescribing) cannot be excluded and may inflate the observed association.,Dose-response, duration-response, and drug-specific (gabapentin vs. pregabalin) relationships are not reported in the abstract, precluding risk stratification by prescribing pattern.,Active comparator and new-user designs — methodological standards that reduce confounding in pharmacoepidemiology — may not have been uniformly applied across included studies.,The absolute risk increase and NNH cannot be calculated from available data, limiting direct clinical applicability.,GRADE certainty rating is not disclosed in the abstract; the overall quality of evidence is likely low or very low given study design and heterogeneity.,Generalizability is uncertain: patient populations, healthcare systems, gabapentinoid indications, and diagnostic ascertainment of ADRD likely vary substantially across included studies.,Funding source is not reported, precluding assessment of potential financial conflicts of interest.,Only the abstract was available for this summary; full-text data on ROBINS-E domain-level risk ratings, subgroup results, meta-regression outputs, funnel plot asymmetry, and sensitivity analyses were not accessible. __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ This meta-analysis of 5 observational studies found a statistically significant but modest association between gabapentinoid use and incident ADRD (OR 1.28; 95% CI 1.10–1.50), accompanied by very high between-study heterogeneity (I² = 90.3%) that severely limits confidence in the pooled estimate. Causality cannot be established given the observational design, high confounding risk, and small number of studies. Clinicians should not discontinue gabapentinoids with established benefit based on this evidence, but should monitor cognitive function in long-term users and apply shared decision-making — particularly in older adults and those with pre-existing cognitive risk factors.