New Meta-Analysis: NMN Looks Safe, But Did Anything Actually Improve?

NAD+ is the molecule at the centre of most of the current longevity conversation. It is a coenzyme involved in energy metabolism, DNA repair and the activity of a family of enzymes that have been linked to ageing biology, and its levels decline with age in several tissues. From that observation flows an entire product category: NMN and nicotinamide riboside capsules sold as precursors intended to restore NAD+ levels, and NAD+ infusions offered in clinics across Ireland and the UK.

The underlying biology is genuinely interesting and is not in dispute. What has always been in dispute is the step from the biology to the benefit, and specifically whether raising a measurable molecule in the blood translates into a person feeling or functioning better. That is an empirical question, and it can only be settled by trials in humans.

A systematic review and meta-analysis published in Nutrients in July 2026 gathered the randomised trial evidence on oral NMN in one place. It is a useful paper precisely because it was not designed to promote anything. Its primary interpretive focus was safety rather than efficacy, on the reasoning that NMN is used as a dietary supplement rather than as a prescribed treatment, and its funding statement records no external funding and no author conflicts of interest. In a field where a great deal of research is sponsored by people selling the product, that matters.

The reviewers searched PubMed, Embase, Scopus, Web of Science and two Chinese databases, CNKI and Wanfang, from inception to 13 May 2026, and pre-registered their protocol on PROSPERO. They included parallel randomised controlled trials comparing oral NMN or NMN-related preparations against placebo, a blank control, a lifestyle control, or the same background intervention without NMN.

Fifteen trials met the criteria. Ten of those contributed to the pooled safety analysis, covering 383 participants in total, of whom 230 were assigned to NMN and 153 to a control condition. The remaining five were excluded from pooled safety estimates mainly because they reported safety only descriptively or in a format that could not be converted into pooled data. Doses ranged from 250 mg per day to 2000 mg per day, and interventions ran from 14 days to 24 weeks. Some trials used a microcrystalline NMN-related preparation rather than ordinary free NMN.

Risk of bias was assessed with the Cochrane RoB 2 tool. Most trials were judged at low risk or as raising some concerns, with the main issues being incomplete reporting of randomisation, possible deviations from intended interventions in open-label or lifestyle-control studies, and missing outcome data. It is worth holding two features of this evidence base in mind throughout: the trials are short, and they are small. Neither is a criticism of the review, which is a fair synthesis of what exists. It is a description of what exists.

On the question the reviewers treated as primary, the answer was consistent. Across the pooled trials, NMN did not increase overall adverse events, serious adverse events, withdrawals due to adverse events, or system-specific adverse events compared with control.

Liver biochemistry was examined separately, since it is the usual early warning for a supplement taken daily. The ALT analysis pooled 10 comparisons covering 330 participants and found no significant change, with a mean difference of -1.03 U/L, a 95 percent confidence interval of -2.51 to 0.44, p equals 0.169 and no detectable heterogeneity between studies. AST behaved the same way across 10 comparisons and 330 participants, with a mean difference of -0.24 U/L, a confidence interval of -1.43 to 0.94 and p equals 0.686.

The authors are careful about how far this stretches, and their caution is worth repeating rather than glossing over. The safety dataset is small, the intervention periods were short, and serious adverse events were rare in absolute terms. They state explicitly that these results should not be read as excluding rare, delayed, long-term, high-dose or population-specific risks. The honest summary is that short-term oral NMN did not show a safety signal in the trials conducted so far, which is a narrower and more useful claim than declaring it safe.

This is the part that matters most for anyone deciding whether to buy NMN, and it is the part least likely to appear in marketing.

Body weight did not change significantly, with a mean difference of -0.32 kg, a 95 percent confidence interval of -1.19 to 0.54 and p equals 0.462. BMI did not change, with a mean difference of 0.02 kg/m2 and p equals 0.795. Fasting plasma glucose did not change, with a mean difference of 0.73 mg/dL, a confidence interval of -1.36 to 2.82 and p equals 0.492. Glycated haemoglobin was not significantly affected either, with a mean difference of 0.03 percent, a confidence interval of -0.04 to 0.10 and p equals 0.342. Lipid profiles and systolic blood pressure showed no significant pooled effect.

Systolic blood pressure is worth noting in a little more detail because it shows how fragile these estimates are. The analysis pooled 11 comparisons covering 339 participants and found a mean difference of -0.89 mmHg, with a confidence interval of -3.83 to 2.05 and p equals 0.553, alongside moderate heterogeneity between studies. Removing one study reduced heterogeneity to zero and moved the estimate to -2.00 mmHg, but the result still did not reach statistical significance.

The authors put it plainly in their own discussion: on current evidence, NMN should not be regarded as an established treatment for metabolic conditions such as obesity, diabetes or dyslipidaemia. That is a striking sentence to find in a paper about a product category sold largely on metabolic and longevity claims.

Two results were not flat, and both deserve to be reported accurately rather than either inflated or ignored.

Diastolic blood pressure fell. Across 11 comparisons covering 339 participants, NMN was associated with a reduction of 2.43 mmHg, with a 95 percent confidence interval of -4.21 to -0.66, p equals 0.007 and no heterogeneity between studies. Leave-one-out sensitivity analysis supported its robustness: removing any single study or comparison produced pooled estimates between -2.75 and -2.02 mmHg, with every confidence interval remaining below zero. That is a real and internally consistent finding.

Insulin resistance, measured by HOMA-IR, trended downward but did not reach significance. The pooled estimate was -0.22, with a confidence interval of -0.55 to 0.11 and p equals 0.19, drawn from five comparisons covering 160 participants. Dose subgroup analysis found no significant difference between dose groups, and neither did analysis by trial duration.

What stops either from being a headline is the certainty rating. The reviewers applied GRADE, the standard framework for judging how much confidence an estimate deserves, and the ratings are the most informative numbers in the paper. Evidence was rated moderate certainty for ALT, AST, body weight, BMI, fasting plasma glucose, triglycerides and total adverse events. It was rated low certainty for HbA1c, LDL cholesterol, and both systolic and diastolic blood pressure. It was rated very low certainty for HOMA-IR and for serious adverse events. The downgrades reflected small sample sizes, indirectness, and the fact that blood pressure was not the primary endpoint in most of the trials measuring it.

Low certainty in GRADE terms means further research is likely to change the estimate substantially. Very low certainty means the estimate is barely more than a starting hypothesis. The blood pressure finding is a reasonable thing to study next. It is not a reason to take NMN for blood pressure, and for context, 2.43 mmHg of diastolic reduction is modest against what established blood pressure treatment achieves.

A separate paper published in Alzheimer's and Dementia in July 2026 illustrates the same gap between the mechanism and the outcome, using the other common NAD+ precursor.

Researchers led from the University of Delaware ran a 12-week double-blind, randomised, placebo-controlled pilot study of nicotinamide riboside in older adults with amnestic mild cognitive impairment, the stage that often precedes Alzheimer's disease. Forty-two participants completed the study, 22 on nicotinamide riboside and 20 on placebo. Adherence was similar between groups and there were no serious adverse effects.

The supplement did what it is supposed to do biochemically. Blood NAD+ roughly doubled in the treated group, confirming the precursor was absorbed and converted. What did not follow was any benefit. There was no improvement in cognitive function, which was the primary outcome, and none in total cerebral blood flow or blood pressure, the secondary outcomes. Exploratory analyses suggested possible increases in regional cerebral blood flow, particularly in the hippocampus, which the authors flagged as a question for longer studies rather than as a result.

The authors' own conclusion is the sentence to take away from this whole field: nicotinamide riboside effectively raises NAD+ in people with mild cognitive impairment but does not improve cognitive function over 12 weeks. Raising the molecule is the easy part and it demonstrably works. Whether raising it does anything for a person is a separate question, and so far the trials answering it have mostly come back negative. It should be noted this was a pilot study of 42 completers, which is small, and a null result in a small trial is not the same as showing an intervention definitely does nothing.

This is the most important limitation to state clearly, because it is the one most often blurred in clinic marketing, and blurring it in either direction would be misleading.

Every trial discussed above tested an oral supplement. Participants swallowed NMN or nicotinamide riboside capsules. Intravenous NAD+ therapy, which Irish and Northern Irish clinics offer as an infusion over several hours, is a different molecule administered by a different route at a different dose with a completely different absorption profile. Evidence does not carry across a change in route of administration, and it does not carry across a change in compound either. An oral NMN trial is not evidence about an NAD+ drip, and it would be equally wrong to present these results as evidence against NAD+ infusions.

What can be said accurately is narrower. The clinical evidence base specific to intravenous NAD+ in humans remains thin, consisting largely of small studies and uncontrolled observations rather than adequately powered randomised trials. The oral precursor literature is the better-developed part of this field, and that part has repeatedly shown the same pattern: NAD+ levels rise reliably, and downstream clinical outcomes mostly do not follow. That pattern does not prohibit a different result for infusions, but it does make the burden of proof for the infusion claim higher rather than lower, because the most plausible mechanism by which a drip would work is the same one that has now been tested orally several times without a clear clinical payoff.

If you are considering an NAD+ infusion, the reasonable question to put to a clinic is which human trials support the specific outcome they are describing, delivered by the specific route they are offering.

The regulatory position is genuinely confusing for consumers and is worth setting out, because availability online is easily mistaken for approval.

NMN is treated as a novel food in the European Union, meaning it was not consumed to a significant degree in the EU before May 1997 and therefore requires pre-market authorisation before it can lawfully be placed on the market as a food or food supplement. That framework applies in Ireland, where the Food Safety Authority of Ireland administers it alongside the European Commission's process. A substance being purchasable from an overseas website says nothing about whether it is authorised for sale here.

The assessment has been progressing. The European Food Safety Authority published an opinion in 2026 on chemically synthesised beta-NMN, evaluating it as a novel food and as a source of niacin in food supplements. It concluded that, under the proposed conditions of use, beta-NMN is safe for adults at intakes up to 300 mg per day, excluding pregnant and lactating women, and that its identity, production process, composition and specifications did not raise safety concerns.

Two points follow. First, an EFSA safety opinion is a scientific assessment of safety, not a finding of efficacy, and not in itself an authorisation. Authorisation is a separate decision taken by the European Commission and member states, and the two should not be conflated. Second, the intake EFSA assessed as safe, up to 300 mg per day for adults, sits at the bottom of the range used in the trials in this meta-analysis, which ran as high as 2000 mg per day. Anyone taking NMN at the higher end of the studied range is above the intake that has been formally assessed as safe for supplement use in the EU, which is worth knowing.

Any injectable form of NMN sits in an entirely different and stricter category again. That is not a food supplement question but a medicines question, and no such product holds a marketing authorisation in Ireland.

If you take NMN, this evidence suggests you are unlikely to be harming yourself in the short term, and equally unlikely to be getting the metabolic benefits the category is marketed on. That is an unsatisfying conclusion, but it is what 15 randomised trials currently support, and it is more useful than either enthusiasm or dismissal.

The fair characterisation is that NMN is a well-tolerated compound whose clinical benefit in humans has not been demonstrated. It is not a debunked product, and a mostly null meta-analysis of small, short trials is not the same as showing something does not work. Longer and larger trials, particularly in older adults and people with early metabolic risk where the authors think a signal is most likely, could change this picture. The blood pressure finding is a legitimate reason to run those trials.

What would be unreasonable is to treat the current evidence as established support for buying it. The pattern across this field is now consistent enough to be a useful rule of thumb: NAD+ precursors reliably raise NAD+, and measured clinical outcomes have mostly failed to follow. Anyone selling you the first step should be asked for evidence of the second.

If your interest in NMN is driven by fatigue, weight that will not shift, or blood sugar concerns, the more productive route is a GP assessment and standard blood work, because those symptoms have common causes with established treatments. This article is general information and not medical advice, and it is worth discussing any supplement with your GP or pharmacist, particularly if you take prescribed medicines.