Invited Review

Gabapentin and Dementia: Are the Benefits Worth the Risk for Spine-Related Pain?

John D. Hesling, MD

University of Colorado School of Medicine Aurora, CO

Austin Boos, DO

University of Colorado School of Medicine Aurora, CO

Key Takeaways

Gabapentin prescriptions are increasing and emerging evidence suggests a weak link between gabapentin and dementia that prompts renewed evaluation of risk versus reward.
Consistent with prior NASS guidelines, available efficacy data do not support the routine use of gabapentin for many common spine-related indications, including low back pain with or without radicular pain and neuropathic pain caused by neurogenic claudication.
Prescribing decisions should be individualized with shared decision making while using risk-mitigating prescribing practices such as time-limited trials, using the lowest effective dose and avoiding unnecessarily long treatment duration.

Introduction

Gabapentin is commonly prescribed off-label to treat neuropathic pain conditions of spinal etiology, including radiculitis, painful radiculopathy, and central canal stenosis with radicular pain and/or neurogenic claudication. Recent studies have highlighted a potential association between gabapentin use and dementia, a concern that has gained broader visibility through lay media coverage and is increasingly being raised by patients. This article reviews the available literature exploring the potential association between gabapentin use and dementia, examines other established risks of gabapentin therapy and contrasts these potential harms with the evidence supporting its clinical efficacy.

Background

While gabapentin is approved by the United States Food and Drug Administration (FDA) for the treatment of partial seizures and postherpetic neuralgia, it is commonly prescribed off label by spine providers for a range of neuropathic and chronic pain conditions.1,2 Gabapentin is a structural analogue of γ-aminobutyric acid (GABA), but its analgesic activity is not attributable to direct agonism at GABA receptors or other primary GABAergic mechanisms.3–5 Instead, gabapentin binds with high affinity to the α2δ subunit of presynaptic voltage-gated calcium channels in neuronal tissue, including within central nociceptive pathways.3–5 This binding is thought to reduce functional calcium channel activity and/or membrane trafficking, thereby decreasing activity-dependent calcium influx at presynaptic terminals and reducing release of excitatory neurotransmitters such as glutamate that facilitate nociceptive transmission. In neuropathic pain models, α2δ expression is upregulated after nerve injury and is implicated in enhanced synaptic transmission and central sensitization; therefore, modulation of this subunit by gabapentin provides a biologically plausible mechanism for attenuating hyperexcitability, allodynia, and hyperalgesia in select chronic pain states.6

Gabapentin utilization has risen substantially over the past decade, with prescription rates more than doubling from 79.5 per 1000 people in 2010 to 177.9 per 1000 people by 2024.7 Strahan et al reported that this increase was driven primarily by rising prescriptions among older adults (≥65 years), a population already at higher baseline risk for cognitive impairment.7 Additionally, it is estimated that up to 95% of gabapentin prescriptions maybe for off-label indications.2,8 In this context, these trends warrant renewed scrutiny of gabapentin’s efficacy and risks to ensure that patients are not being exposed to potential harms in the absence of high-quality evidence of likely benefit.

Emerging Concerns Linking Gabapentin to Dementia

Against this backdrop of rapidly rising off-label use, particularly among older adults, recent publications have raised concern regarding a potential association between gabapentin use and an increased risk of dementia and/or mild cognitive impairment in humans.

The most recent and widely publicized of these was a 2025 retrospective cohort study by Eghrari et al.9 This study included 26,416 adults identified within the US TriNetX national database as having chronic pain, chronic pain syndrome, chronic low back pain, or lumbar radiculopathy. Patients were categorized based on the presence of at least 2 gabapentin prescriptions versus no gabapentin exposure and were further stratified by age. Gabapentin exposure was associated with a higher likelihood of both dementia (RR 1.29, 95% CI, 1.18–1.40) and mild cognitive impairment (RR 1.85, 95% CI, 1.63–2.10). These associations corresponded to absolute risk increases of 1.0% for dementia and 1.2% for mild cognitive impairment compared with the control group. An exposure–response relationship based on prescription duration was also observed, with patients receiving 12 or more gabapentin prescriptions demonstrating a higher risk of dementia than those receiving 3–11 prescriptions (RR 1.40, 95% CI, 1.25–1.57). Age-stratified analyses suggested a disproportionate relative risk among younger patients (18–64 years; RR 2.10, 95% CI, 1.75–2.51) compared with elderly patients (≥65 years; RR 1.28, 95% CI, 1.15–1.42).

Huang et al previously reported similar findings in a large retrospective, population-based matched cohort study of 206,802 patients enrolled in Taiwan’s National Health Insurance program, published in 2023.10 In that study, gabapentin exposure was associated with a significantly higher risk of dementia across all age groups. The multivariable-adjusted hazard ratio was 1.45 (95% CI, 1.35–1.55), corresponding to an absolute risk increase of 1.5% compared to the control group. An exposure-response relationship based on cumulative prescribed dose was observed, with dementia risk increasing as cumulative daily doses increased. Their analysis adjusted for multiple dementia risk factors, including diabetes, hypertension, hyperlipidemia, cerebrovascular disease, head injury, and depression. Of note, the increased risk of dementia was not statistically significant in the depression or head-injury subgroups. When stratified by age, patients younger than 50 years demonstrated nearly double the hazard ratio for dementia (HR 3.16, 95% CI, 2.23–4.47) compared with older age groups, with the next highest risk observed among those aged 50–59 years (HR 1.58, 95% CI, 1.24–2.00).

In contrast, Tsai et al reported null findings in a 2024 nested case-control study using the Taiwanese Health and Welfare Data Science Centre insurance database among patients with chronic pain.11 Inclusion criteria included a diagnosis of chronic pain, which was defined as receipt of a pain medication prescription at least three times within one year. After adjustment for dementia-related comorbidities, including hypertension, hyperlipidemia, and ischemic heart disease, no increased risk of dementia was observed among patients prescribed gabapentin (adjusted OR 0.91, 95% CI, 0.83–1.01). Similarly, no dose-dependent effect was identified when patients were stratified into high- and low-exposure groups (low exposure defined as <9800 mg cumulatively).

Notably, a 2023 systematic review and meta-analysis of 10 observational studies evaluating anti-seizure medication exposure and dementia risk failed to demonstrate a significant association once analyses were adjusted for drug indications.12 In the primary pooled analysis, overall antiseizure medication use was associated with a modestly increased dementia risk (OR 1.09, 95% CI, 1.03–1.15); however, this association disappeared after adjustment for treatment indication (OR 1.02, 95% CI, 0.97–1.07). In subgroup analyses by individual drug, no significant association was observed for gabapentin exposure (OR 0.76, 95% CI, 0.49–1.18), whereas increased risk was confined to select first-generation agents, including valproate, carbamazepine, and clonazepam.

Taken together, the existing human evidence suggests a possible, but inconsistent, association between gabapentin use and an increased risk of dementia. While the 2 retrospective cohort studies reported modest but statistically significant increases in dementia risk, other observational studies, have failed to detect a similar association. These discrepant findings raise important questions about whether the observed signal in the positive studies reflects a true relationship or is partly attributable to residual confounding. In particular, differences in underlying patient characteristics, comorbidity burden, and clinical indications for gabapentin prescribing may contribute to an apparent association that does not reflect a direct effect of the drug itself. As such, the current human evidence neither definitively establishes nor definitively refutes a causal link between gabapentin exposure and dementia risk.

Mechanistic and Preclinical Considerations

Although the available epidemiologic evidence does not establish a causal relationship, several lines of preclinical and mechanistic evidence have been proposed that could, in theory, support a link between gabapentin exposure and adverse cognitive outcomes. Beyond its effects on nociceptive processing, gabapentin modulates neuronal calcium channel activity and indirectly influences multiple neurotransmitter systems involved in learning and memory. Binding of gabapentin to the α2δ subunit of voltage-gated calcium channels reduces presynaptic calcium influx and alters the release of excitatory neurotransmitters, including glutamate, with downstream effects on synaptic transmission.13,14 Disruption of the balance between inhibitory and excitatory signaling has been implicated in neurodegenerative processes, including Alzheimer disease, providing a biologically plausible—though unproven—mechanistic framework by which chronic gabapentin exposure could influence cognitive function.15

Preclinical data provide one such mechanistic hypothesis. In a recent study, Xia et al demonstrated that long-term gabapentin administration in aged mice was associated with impaired recognition memory and reduced performance on contextual and cued fear-conditioning tasks.16 These behavioral deficits were accompanied by increased hippocampal tau phosphorylation at Alzheimer-associated sites (Ser262 and Ser416), suppression of Sirt1 expression, and upregulation of CaMKIIα activity. Restoration of Sirt1 signaling through viral overexpression or resveratrol administration attenuated both tau hyperphosphorylation and cognitive impairment.¹⁶ While these findings establish a mechanistic link between chronic gabapentin exposure and tau-related pathology in aged animals, their relevance to human dementia risk remains uncertain and requires cautious interpretation.

Additional hypotheses focus on the indirect functional consequences of gabapentin use that may influence dementia risk over time. Gabapentin initiation has been associated with increased falls and functional decline, with higher odds of falls and worsening Functional Activities Questionnaire scores after treatment is started.¹⁷ Falls and fear of falling can in turn reduce mobility and social participation, and lower social participation is itself associated with an increased risk of developing dementia.¹⁸,¹⁹

Taken together, these mechanistic and preclinical observations provide biological plausibility for a potential association between gabapentin exposure and adverse cognitive outcomes, but they remain insufficient to establish causality and must be interpreted in the context of the modest, inconsistent, and observational nature of the existing human data.

Efficacy of Gabapentin for Common Spine and Pain Indications

Regardless of whether the observed association between gabapentin and dementia ultimately proves causal, its clinical relevance depends fundamentally on the magnitude and reliability of gabapentin’s therapeutic benefit for the conditions in which it is most prescribed. Accordingly, a careful appraisal of gabapentin’s efficacy data is essential for interpreting the appropriateness of long-term use.

Unfortunately, the evidence supporting gabapentin for many of the conditions for which it is routinely prescribed by pain and spine providers is limited and, in many cases, only weakly supportive. Prior NASS guidelines found insufficient evidence to recommend pharmacologic treatments, including gabapentin, for lumbar radiculopathy or spinal stenosis. Clinical trials have failed to demonstrate that gabapentin is an effective treatment for radicular pain or chronic low back pain (LBP).20 Enke et al conducted a systematic review of anticonvulsants for radicular pain that included nine trials (6 randomized controlled trials and 3 randomized crossover trials) comprising 859 participants. They found that the available evidence did not support the use of anticonvulsants for either chronic LBP or radicular pain. In a pooled analysis of 3 trials comparing gabapentin with placebo for LBP with or without radicular pain, the mean difference (MD) in pain scores was 0.0 (95% CI, −0.3 to 0.3). Four additional trials comparing gabapentin or pregabalin with placebo for lumbar radicular pain produced a pooled MD of −0.1 (95% CI, −0.3 to 0.2).

Current evidence from a 2023 systematic review and meta-analysis suggests that gabapentinoids have limited and time-dependent effects in lumbar spinal stenosis.21 Martínez et al included 6 comparative studies (primarily randomized trials) totaling 392 patients and evaluated pain (VAS/NRS) and disability (ODI) outcomes. Pain scores were not significantly different between gabapentin or pregabalin and control groups at 2, 4, or 8 weeks. However, at 3 months the pooled analysis showed a statistically significant improvement in VAS favoring gabapentinoids, driven largely by pregabalin. ODI outcomes did not differ significantly at 8 weeks (MD −3.47, 95% CI −7.15 to 0.21; p = 0.06). In sensitivity analyses, gabapentin alone did not demonstrate significant pain improvement at the analyzed follow-up time points, whereas pregabalin showed improvement at some time points.

Overall, the authors concluded that any potential pain benefit appears modest and is accompanied by higher adverse event rates.

In fibromyalgia, a Cochrane review identified only one eligible 12-week, randomized, double-blind, placebo-controlled trial (n=150), and rated the certainty of evidence as very low, concluding there is insufficient evidence to support or refute that gabapentin reduces fibromyalgia pain.22 In that trial, 30% or greater pain reduction was achieved by 49% (38/75) of participants receiving gabapentin versus 31% (23/75) receiving placebo; the review did not report a calculated NNT for this outcome, and 50% pain reduction was not reported. For acute postoperative pain, a separate Cochrane review pooling three trials of gabapentin 250 mg found 15% achieved ≥50% pain relief over 6 hours compared with 5% on placebo (NNT 11), but concluded the analgesic effect is clinically unremarkable and inferior to commonly used analgesics, and therefore not recommended as first-line therapy when other effective options are available.23

In contrast, robust evidence continues to support the use of gabapentin for postherpetic neuralgia and painful diabetic neuropathy.24 Wiffen et al conducted a Cochrane Review of gabapentin for chronic neuropathic pain that included 37 studies and 5914 participants. Moderate-quality evidence supported its use in postherpetic neuralgia, with an NNT of 6.7. They reported that 32% of patients achieved greater than 50% pain relief compared with 17% receiving placebo (RR 1.8, 95% CI, 1.5–2.1). Moderate-quality evidence also supported its use in painful diabetic neuropathy, with an NNT of 6.6. In this population, 38% of patients achieved greater than 50% pain relief compared with 23% in the placebo group (RR 1.7, 95% CI, 1.4–2.0).

In summary, available efficacy data, consistent with prior NASS guidelines, do not support the routine use of gabapentin for many common spine-related indications, including LBP with or without radicular pain and neuropathic pain caused by neurogenic claudication. This limited evidence of benefit makes it especially important for clinicians to carefully consider potential harms when prescribing gabapentin for these conditions.

Gabapentin-Associated Harms

In addition to the limited and condition-specific evidence supporting gabapentin’s efficacy for many common pain indications, its overall risk–benefit profile is further shaped by a well-documented burden of adverse effects. In the 2017 Cochrane Review by Wiffen et al, 11% of patients receiving gabapentin withdrew because of adverse events compared with 8.2% in the placebo group, yielding a number needed to harm (NNH) of 30 (RR 1.4, 95% CI, 1.1–1.7), although serious adverse events were no more common than with placebo.24 Across 18 studies including 4279 participants, 63% of patients reported at least one adverse event compared with 49% in the placebo group (NNH 7.5; RR 1.3, 95% CI, 1.2–1.4).²²

The most frequently reported adverse effects included dizziness (19%; NNH 8; RR 2.9, 95% CI, 2.4–3.4), somnolence (14%; NNH 11; RR 2.8, 95% CI, 2.3–3.5), peripheral edema (7%; NNH 20; RR 4.1, 95% CI, 2.7–6.4), and gait disturbance (14%; NNH 8.5; RR 5.5, 95% CI, 2.5–12).²² One year later, Enke et al reported a similarly increased risk of any adverse event compared with placebo (RR 1.4, 95% CI, 1.2–1.7) in their systematic review and meta-analysis of anticonvulsants for low back and radicular pain.20 In both reviews, the evidence supporting increased adverse events with gabapentin was graded as high quality.

Table 1. Summary of Benefits and Harms for Gabapentin

Indication/Outcome	NNT (95% CI)	NNH (95% CI)
Low back pain (± radicular pain)²⁰	No benefit over placebo
Lumbar spinal stenosis with neurogenic claudication²¹	No benefit over placebo
Postherpetic neuralgia (≥50% pain relief)²⁴	6.7 (5.4–8.7)
Painful diabetic neuropathy (≥50% pain relief)²⁴	6.6 (5.0–9.7)
Fibromyalgia²²	Not reported
Any adverse event²⁴		7.5 (6.1–9.6)
Dizziness²⁴		8.0 (7.0–9.4)
Gait disturbance²⁴		8.5 (6.1–14)
Somnolence²⁴		11 (9.4–14)
Peripheral edema²⁴		20 (16–27)
Discontinuation due to adverse event²⁴		30 (20–65)

Abbreviations: NNT, number needed to treat; NNH, number needed to harm; CI, confidence interval.

Discussion

Taken together, the weak and condition-specific evidence supporting gabapentin’s efficacy for many common spine-related pain conditions must be balanced against its established adverse-effect burden and emerging, though still inconsistent, observational evidence suggesting a possible association with dementia. Although current data do not establish causality and the reported absolute risk increase is modest at approximately 1% compared to baseline, this potential signal warrants thoughtful consideration, particularly when considering long-term treatment.

At the same time, a lack of population-level efficacy as reported in clinical trials should not obscure the possibility that individual patients may derive meaningful benefit from gabapentin, particularly those with comorbid symptoms such as impaired sleep or nocturnal pain. For some patients, improvements in sleep quality, pain control, or overall function may enable greater participation in protective behaviors such as restorative sleep, regular exercise, and social or community engagement that are themselves associated with reduced dementia risk. In such cases, these functional and behavioral benefits may plausibly outweigh any small or uncertain long-term cognitive risk suggested by observational studies.

Accordingly, decisions regarding gabapentin therapy should be individualized and informed by shared decision making, with explicit consideration of potential risk-mitigating prescribing practices. Clinicians should engage patients in discussions regarding the uncertain long-term cognitive risks, the limited and condition-specific evidence of benefit, and the patient’s treatment goals, preferences, and functional priorities. Reasonable approaches may include avoiding unnecessary long-term therapy, using the lowest effective dose, and exercising additional caution in younger patients or in those anticipated to require prolonged exposure.

Prescriptions should be reassessed at regular intervals, ideally incorporating both patient-reported outcomes and objective functional measures when available. In the absence of clear and sustained clinical benefit, tapering and discontinuation should be considered. Time-limited therapeutic trials with predefined response criteria, along with dose reduction or discontinuation in cases of marginal or absent efficacy, may help limit unnecessary cumulative exposure while preserving benefit in patients who demonstrate a meaningful response.

Conclusion

Gabapentin remains widely prescribed off-label to manage common spine conditions but there’s emerging concern that it may be linked to dementia, combined with poor evidence of efficacy implore providers to use shared decision making to carefully assess if each patient gains enough benefit to be worth the risks.