Gabapentin before starting it

Prepared 5 September 2026 by Kya for her partner and the patient, because you asked for a lot more information before trialling it. Every figure was checked today against the manufacturer's product information (the Neurontin Summary of Product Characteristics), the MHRA safety notices, the NHS pages and the trials named, all listed in research thread 14. Where something is judgement rather than evidence, it says so. No prognosis content. This is to help you decide and to help you run the trial safely; the prescriber sets the doses.

The short answer

• A trial is reasonable. Gabapentin is one of four NICE first-line drugs for nerve pain, and head-to-head trials show none of the four beats the others. It is the only one with a placebo-controlled trial in people already on opioids for nerve pain from cancer, where it helped within ten days at modest doses. It acts where several irritated nerves converge rather than on any one nerve, which fits the registrar's reading of the patient's pain. It has no liver or drug-level interactions with anything she may need later. The weight gain most people dislike would suit her.
• The costs are real and specific. At working doses about one in five people are dizzy and about one in seven drowsy, against roughly one in fifteen and one in twenty on placebo. About one in ten stop because of side effects. Over half get no worthwhile relief. Blurred or double vision is a common effect, in a woman who already has double vision. With morphine the sedation adds up and breathing can slow, which is the MHRA warning.
• The conditions that make it safe enough: kidney function known before the dose climbs; a slow, night-first schedule; one step at a time, only from a plateau; a written description of her eye movements before the first dose; someone watching her breathing for the first three nights and after each increase; the diary kept; and the stop rule and the swap agreed in advance and written down.
• What would make you stop: fog or drowsiness that does not lift after a step down and a week's wait; no pain signal at all after two weeks at 1,800 mg a day; any change in mood that is not her; worsening double vision that the doctors attribute to the drug; or any breathing, rash or allergic event, which stops it the same hour.
• What it will not do: treat the tumour, work in a day, or remove the need for the anti-inflammatory and a steady low morphine dose while it builds. It is one layer, added slowly, and the diary is what makes it reversible within two weeks if it costs her more than it gives.

What it is and how it works

Despite the name, gabapentin does not act on GABA receptors. It binds a subunit of the calcium channels on nerve endings and in the spinal cord, and by doing so damps the release of the chemicals that overexcited nerves use to signal pain. That is why it works on the nerve component of pain and does nothing for the inflammatory or bony component; it is added to the anti-inflammatory and the opioid, not swapped for them.

The evidence

• Cochrane review (Wiffen 2017): at 1,200 mg a day or more, substantial relief in 32 to 38 percent of people against 17 to 21 percent on placebo. That is a number needed to treat of six to seven. The review's own words: over half will not have worthwhile relief.
• NICE CG173: one of four first-line choices with amitriptyline, duloxetine and pregabalin; if the first does not work or is not tolerated, switch to another.
• Cancer nerve pain on opioids (Caraceni 2004): the only placebo-controlled trial in this exact situation. At 600 to 1,800 mg a day, added to opioids, pain improved within ten days, and the first thing to change was the burning, electric quality of the pain, before the score moved. Dropouts equalled placebo.
• With morphine (Gilron 2005): morphine plus gabapentin relieved pain better than either alone, at lower doses of each. That is the argument for a steady low morphine dose alongside it rather than more of either.
• Face and trigeminal pain: small studies only; one suggests a nerve block plus low-dose gabapentin beats gabapentin alone, which is a reason to ask for both rather than either.
• Against the alternatives: pregabalin is the same mechanism with more predictable absorption and similar side effects; duloxetine is the least sedating; nortriptyline or amitriptyline help sleep. None is proven more effective; they differ in side effects, which is why the swap exists.

Dosing and the slow schedule

The label schedule is 300 mg on day one, twice on day two, three times on day three. That is too fast for a 51 kg woman on morphine whose clarity matters. The schedule to ask for, in writing (judgement, built from the product information):

• 100 to 300 mg at night for the first three nights.
• Then a step every three to seven days, and only from a plateau: 300 mg twice a day, then three times a day (900 mg), then upward if needed.
• Expect a first signal at about 900 mg a day and a fair verdict at 1,800 mg after two weeks. Most responders in trials needed 1,800 to 3,600 mg, but in people on opioids the tolerated maximum is lower, and the evidence favours adding a second agent over pushing the dose.
• Kidney function sets the ceiling: the product information has a table by eGFR, and her result from today's bloods is needed before the climb.
• Take it with or without food; not within two hours of an antacid, which cuts absorption.

Side effects, with rates and what to expect

• Dizziness about one in five, drowsiness about one in seven, unsteadiness about one in seven, ankle swelling about one in fourteen, at working doses. About one in ten stop because of side effects, against one in twelve on placebo. These are worst in the first weeks and at each step up, and mostly settle at a steady dose; that is why the steps are slow.
• Thinking: concentration, word-finding and memory complaints occur, usually early and dose-related. In a twelve-week randomised study in healthy volunteers, gabapentin affected one of twenty-four measures of thinking, which is a smaller footprint than carbamazepine's. Her clarity is the thing to score daily.
• Eyes: blurred vision, double vision and eye flicker are all listed as common. Because she already has double vision from the tumour, a written baseline of her eye movements before the first dose is essential, so a drug effect can be told from a tumour effect.
• Weight gain and appetite increase: common, and in her case welcome.
• Mood: a class-wide warning about low mood and suicidal thoughts applies; any change that is not her is a stop signal.
• Rare and serious: a severe skin reaction with fever and swollen glands, allergic swelling, and breathing suppression, particularly with opioids. Each stops the drug the same hour and needs a doctor.

With morphine: the interaction that matters

Morphine raises gabapentin levels in the blood by nearly half. The MHRA warns that gabapentin with any opioid increases the risk of slowed or suppressed breathing, and a large Canadian study found a measurable increase in opioid-related deaths when the two were co-prescribed. The population risk is real; the individual risk at her doses, under supervision, is small, and it is the reason for three rules: the morphine stays modest while gabapentin climbs, someone watches her breathing for the first seventy-two hours and after each increase, and no alcohol or other sedatives. If she is hard to rouse, or her breathing is slow, shallow or pausing, both drugs stop and you call.

Dependence and stopping

Gabapentin became a Class C controlled drug in April 2019 because of misuse in people with a history of it. Every published addiction case had prior substance misuse; the risk to the patient is low, and the classification changes the paperwork, not the medicine. It must never be stopped suddenly after weeks of use: the product information requires a taper over at least a week, and the withdrawal from an abrupt stop is unpleasant.

The diary and the rules

Two lines a day, next to the pain diary already running:

• Morning: hours slept; worst pain overnight out of 10; every dose since the last entry with times.
• Evening: average pain today out of 10; clarity today out of 10, where 10 is entirely herself, scored by the patient and by her partner separately if you differ; tick any of drowsy, dizzy, unsteady, double vision worse, ankle swelling, nausea, mood change, word-finding trouble; breathing normal overnight, yes or no.

Rules to agree with the prescriber and have written on the letter:

• Hold a step if, on the day of a planned increase, she is drowsier, dizzier or foggier than two days earlier, has had a fall or near-fall, her double vision is worse, or she has needed more Oramorph than usual. Stay put another three to seven days. Holding is free; stepping up early is what makes the fog.
• Drop back a step if drowsiness, dizziness or fog persist beyond seven days at a steady dose; if clarity stays two or more points below her baseline for a week; if ankle swelling troubles her; or if she sleeps by day and wakes by night.
• Stop, taper and swap if there is no pain benefit at all after two weeks at 1,800 mg or her highest tolerated dose; if fog does not lift after a step down; if her mood is not her; or if the doctors attribute worsening double vision to the drug. Swap to duloxetine, the least sedating, or nortriptyline, which helps sleep.
• Stop the same hour and call for slow, shallow or paused breathing, being unable to rouse her normally, rash with fever or swollen glands, lip or throat swelling, severe persistent stomach pain, hallucinations, or thoughts of self-harm.
• The honest signal it is working: not "pain gone". A fall of two points or more in the average daily score, or thirty percent, held for a week; fewer Oramorph doses; sleeping through more of the night; and often, first, a change in the quality of the pain, less burning and less electric. If the score does not move but she is using half the morphine, that is also a response.

Ten questions for the prescriber

1. What starting dose and what schedule, in writing? We would like 100 to 300 mg at night for three nights, then a step every three to seven days, not the day one, two, three label schedule, because she is 51 kg and on morphine.
2. What is her eGFR, and has the dose been checked against the renal table using her actual weight?
3. What is the target dose for a fair trial, and at what point do we call it a failure? Our suggestion: a verdict at 1,800 mg after two weeks.
4. Will breathing rate and a sedation score be recorded for the first seventy-two hours and after each increase? At home, what breathing signs make us stop both drugs and call?
5. Can a baseline description of her eye movements and double vision be written down before the first dose?
6. Does the morphine plan change? Should the Oramorph become a low regular dose rather than as needed, and should the total stay where it is while gabapentin climbs?
7. What is the stop rule for fog or drowsiness, and what is the swap, duloxetine or nortriptyline? Will that be written so the GP can act without another referral?
8. Can a sphenopalatine ganglion or occipital nerve block be done alongside, so she may need less gabapentin?
9. Who prescribes the ongoing supply, given the 28-day controlled-drug rule, and who reviews her at two weeks?
10. How will we taper if it does not suit her?

Sources

Neurontin Summary of Product Characteristics (emc product 3195). MHRA Drug Safety Updates 2017 (gabapentinoids and opioids: respiratory depression) and 2019 (Class C reclassification). NHS gabapentin pages. Wiffen et al 2017, Cochrane (PMID 28597471). Finnerup et al 2015, Lancet Neurology. Caraceni et al 2004, Journal of Clinical Oncology. Gilron et al 2005 and 2009, New England Journal of Medicine and Lancet. Gomes et al 2017, PLoS Medicine (opioid and gabapentin co-prescription). Eckhardt et al 2000 (morphine raises gabapentin exposure). Salinsky et al 2005 (cognition in volunteers). Lemos et al 2008 (block plus gabapentin). Evoy et al 2021 (misuse). Bockbrader et al 2010 (pregabalin and gabapentin kinetics). Full list with links: research thread 14, em-research/14-gabapentin.md. NICE CG173 wording is carried from research thread 08 because the NICE site could not be fetched today.