Low-Dose Naltrexone Real-World Response Rate: What Patients Actually Experience

Is there a real-world response rate for LDN?

The available trials used different populations, doses, outcomes and responder definitions, so none provides a response rate that can be applied to everyone using LDN [6][1][3][4][5][10]. One small fibromyalgia trial reported 32% of women meeting its responder definition on LDN versus 11% on placebo [1]. Beyond that, what exists is a group of small trials in different conditions, and the results disagree with each other [6][10]. A 2025 chart review followed 20 patients prescribed LDN for refractory cancer pain outside a trial [2].

LDN is not an approved product and has no standard dose. It means naltrexone at roughly 0.5 to 6 mg a day, most often 4.5 mg, used off-label and supplied mainly through compounding pharmacies [6]. A 2026 narrative review calls the 4.5 mg dose empirically derived rather than evidence-based [6].

The fibromyalgia trial that counted responders

The clearest responder number comes from a 2013 crossover trial in Arthritis and Rheumatism. Thirty-one women with fibromyalgia took 4.5 mg naltrexone daily and placebo in random order, double-blind. Daily pain fell 28.8% on LDN versus 18.0% on placebo (P=0.016). Thirty-two percent met the trial's responder criteria, which required a significant pain reduction plus a significant reduction in either fatigue or sleep problems, versus 11% on placebo (P=0.05). Fatigue and sleep were not improved. LDN was rated as tolerable as placebo, with no serious side effects [1].

It was a small trial, and its result needed replication. That replication is where the story changes.

What happened when larger trials repeated it

Three later placebo-controlled fibromyalgia trials were bigger and did not reproduce the result.

• 58 patients, crossover (Pain Reports, 2023). LDN 4.5 mg daily versus placebo. The median difference in Fibromyalgia Impact Questionnaire scores was -1.65 (P=0.3) and in summed pain ratings -0.33 (P=0.4). The authors concluded the data did not show clinically relevant analgesic efficacy [4].
• 99 women, 12 weeks (Lancet Rheumatology, 2024). LDN 6 mg daily. Pain on an 11-point scale fell 1.3 points with LDN and 0.9 with placebo, a between-group difference of -0.34 points (95% CI -0.95 to 0.27, p=0.27) [3].
• 98 women, 12 months (INNOVA, European Journal of Pain, 2026). LDN 4.5 mg daily. At 3 months pain changed -0.33 points on LDN versus -0.64 on placebo, an adjusted difference of 0.49 (p=0.236). The trial reported low and similar responder rates in both groups (p=0.219 to 0.954). Adverse events were reported by 68.8% on LDN and 72% on placebo [5].

The pooled view

A 2024 meta-analysis combined four randomized fibromyalgia trials, 222 patients in total. Pain scores were lower with LDN, with a mean difference of -0.86 points (95% CI -1.20 to -0.51). The revised Fibromyalgia Impact Questionnaire score did not differ significantly between LDN and placebo. Vivid dreams and nausea were more common with LDN; serious adverse events were not significantly different [10].

The 2026 narrative review screened 105 studies, including 15 randomized trials across seven clinical areas. It found that early positive results from uncontrolled studies were rarely confirmed in placebo-controlled trials. It concluded that current evidence does not support routine clinical use, while describing LDN as low-risk but unproven [6].

Crohn's disease: one small positive trial

A 2011 randomized trial in Digestive Diseases and Sciences gave 40 adults with active Crohn's disease 4.5 mg naltrexone or placebo daily for 12 weeks. A fall of at least 70 points in the Crohn's Disease Activity Index occurred in 88% on naltrexone versus 40% on placebo (p=0.009). Endoscopic response occurred in 78% versus 28% (p=0.008), and 33% versus 8% reached endoscopic remission [7].

The effect was large, but it comes from a single small trial. An earlier open-label study of 17 adults reported a response in 89% and remission in 67% [11], and a pediatric pilot of 14 children used a weight-based dose and reported improvement without serious adverse events [12].

Multiple sclerosis: quality of life, not disease outcomes

A 2010 pilot in Annals of Neurology gave 4.5 mg nightly for 8 weeks in a double-masked crossover. Eighty people enrolled and 60 completed. LDN improved several self-reported mental health measures, including a 3.3-point gain on the SF-36 Mental Component Summary (p=0.04) and a 6-point gain on the Mental Health Inventory (p<0.01), and was well tolerated with no serious adverse events. High dropout and data errors substantially reduced the trial's statistical power [13].

Long COVID: uncontrolled data only

A 2026 systematic review in BMJ Open searched through 5 May 2026 and identified no randomized trials of LDN for long COVID. It pooled four observational before-and-after studies from the USA and Ireland, 155 people in total, at doses of 1 to 6 mg a day. The pooled changes suggested moderate improvement in fatigue, brain fog and sleep and large improvement in pain and daily functioning, but uncontrolled before-and-after data cannot separate a drug effect from placebo or from natural recovery [8].

What a real-world chart review looks like

A 2025 case series in the Journal of Pain and Symptom Management reviewed 20 cancer patients prescribed LDN in 2022 and 2023. Eighty percent (16 of 20) were counted as responders at the first follow-up, a mean of 65.9 days after starting LDN, and 76.9% (10 of 13) were responders at the second follow-up, a mean of 104.8 days in. The most common highest dose was 3.0 mg daily, range 1.5 to 4.5 mg. Three of the 20 had stopped by the second visit, two for inadequate pain relief and one for diarrhea [2].

Patients were classed as responders if they attributed improvement in their main pain complaint to LDN. Because responders reported their own improvement and the review had no comparison group, these figures cannot show how much of the improvement was due to LDN [2].

Why the numbers differ so much

The trials above used 4.5 mg, 6 mg and weight-based doses, ran for 8 weeks to 12 months, and measured different things: daily pain, impact scores, disease activity indices, quality of life. The 2026 review's overall finding was that positive early findings rarely held up in the placebo-controlled trials that followed [6].

Side effects patients report

The 2026 review describes LDN trial side effects as mostly mild and transient, most often insomnia, vivid dreams, nausea and dizziness [6]. The 2024 meta-analysis found vivid dreams and nausea more often with LDN than placebo, with no significant difference in serious adverse events [10]. In the 12-month INNOVA trial, adverse events were reported by slightly fewer people on LDN than on placebo [5].

There is no FDA-approved label specifically for LDN. The existing naltrexone labels cover 50 mg tablets and a 380 mg monthly injection and give no guidance on low doses [9][14][6].

The opioid interaction

The naltrexone labels contraindicate the drug in anyone receiving opioid analgesics or currently dependent on opioids, and the contraindication is written without a low-dose exception [9][14]. The 50 mg tablet label says that when withdrawal is precipitated in an opioid-dependent person it can be severe enough to require hospitalization, with symptoms usually appearing within five minutes of ingestion and lasting up to 48 hours; the label gives no corresponding timing for LDN [9][6]. The labels recommend a minimum of 7 to 10 opioid-free days, and the tablet label notes that people switching from buprenorphine or methadone may remain vulnerable to precipitated withdrawal for as long as 2 weeks [9][14].

What compounding means for the product itself

No FDA-approved low-dose naltrexone product exists [6]. FDA states that compounded drugs are not FDA-approved and that the agency does not verify their safety, effectiveness or quality before they are marketed [15]. Naltrexone is not named on FDA's list of bulk drug substances that may present significant safety risks for compounding [16], and the 2026 review notes that inconsistencies between compounded formulations add uncertainty to any real-world result [6].