The 400 mg capsule did nothing in a week. Is gabapentin a bust?
NNT is the honest column on this blotter
Substantial relief in PHN trials needed seven people treated so one extra person got there versus placebo. That is the number this page refuses to hide.
Number needed to treat is how you tell a patient the truth without a brochure. For postherpetic neuralgia, Cochrane's 2018-era pool (Wiffen, Derry, Bell, Rice and colleagues, CD007938) put substantial benefit - at least fifty percent pain relief or 'very much improved' - at 32 percent on gabapentin 1200 mg or more versus 17 percent on placebo. NNT 6.7 (5.4 to 8.7). Eight studies, 2260 people. Moderate-quality evidence.
Moderate benefit - thirty percent relief or 'much improved' - looked better: NNT 4.8. Still not a coin flip in your favor. Around three or four in ten on drug got the big win. One or two in ten got it on placebo. Over half will not have worthwhile relief and may still feel dizzy.
A 400 mg Neurontin capsule is how many US titration schedules begin or step. It is not the dose those NNT rows were built on. Quoting 400 mg as if it were the trial target is how clinics under-dose for a month and then declare the class useless.
This is not a GABA agonist story. The molecule binds the alpha-2-delta auxiliary subunit of voltage-gated calcium channels. The trade name is a historical tease. Keep that straight when a patient asks why they are not 'on a benzo.'
Dizziness, edema, and the opioid stack
Somnolence and dizziness are the titration tax. Peripheral edema is dose-related and easy to blame on heart failure if you do not look. Weight and ankles belong in the week-two note.
The 2019 FDA respiratory warning sits on gabapentinoids plus opioids, and on COPD. That is a pharmacodynamic stack, not a theoretical. Discharge summaries that add oxycodone to a fresh 400 mg start without naming the breathing risk are incomplete.
Withdrawal after chronic use is real. Taper. Abrupt stop after months at trial-range doses can produce anxiety, insomnia, and sweating that look like the original pain returning. That is not a reason to stay forever. It is a reason to come down on a schedule.
Misuse and diversion reports grew as opioid desks looked for 'safer' substitutes. A 400 mg capsule is still a CNS drug. Count it. Do not leave a bottle in a shared bathroom and call it harmless because it is not an opiate.
Cochrane's 6.7 is a pool, not a promise
Thirty-seven studies and 5914 participants fed the adult neuropathic-pain Cochrane update. Most used 1200 mg or more. Duration was typically four to twelve weeks. Small crossovers and messy withdrawal handling were the main bias problems. Conclusions did not flip from the prior update.
Painful diabetic neuropathy sits in that pool with a thinner, noisier file than PHN. Other neuropathic syndromes - radiculopathy, HIV, mixed 'neuropathy' labels - barely have a row. Do not borrow a PHN NNT for a back-pain clinic.
Fifty percent pain reduction is the patient-valued cut because it tracks sleep, fatigue, and function better than a two-point mean that nobody felt. Mean-change forest plots hide the split: some people drop hard, most do not.
If a consultant quotes 'gabapentin works' without an NNT or a syndrome, ask which pool. The Neurontin formulary line will not invent a diabetic-foot guarantee the trials never printed.
Patient global impression of change was a co-traveler in Rice and in the Cochrane substantial-benefit definition. A two-point mean drop that nobody would call 'very much improved' is not the same stamp. Ask which endpoint the clinic is celebrating.
Who the enrolment desks quietly left out
PHN trial ages ran old - Rice's mean was 73 - but the protocols still excluded unstable cardiac disease, severe renal failure, and people already deep on other sedatives. Your clinic panel is messier.
Women of childbearing potential, active substance-use disorder, and mixed nociceptive-neuropathic back pain were not the enrolment core. Quoting a 6.7 NNT into those charts is extrapolation.
Placebo responses of 17 percent for substantial relief are not 'nothing happened.' They are why you need a time-limited trial with a written stop date. Open-ended 400 mg 'to see' becomes a forever drug.
If the pain is radicular, post-surgical, or fibromyalgia-labeled, say so on the blotter. Those rows were not the PHN pool. Honesty about the missing file is part of the consent.
Care-home enrolment was thin. A 400 mg bedtime start in a ninety-year-old with a walker and a sleeping pill already on the list is not Rice's protocol. Fall risk is an NNH the PHN pool only partly captured.
Diabetic nerve pain is a thinner file
Backonja-era painful diabetic neuropathy trials showed benefit, but the certainty and the NNT are less tidy than the PHN stack. Cochrane still allows the class a role at 1800 to 3600 mg for some people with DPN. It does not let you promise a foot will quiet.
Glucose, shoes, and ulcer care remain the disease-modifying column. Gabapentin is a symptom stamp. Mixing those columns is how a 400 mg bedtime capsule becomes a substitute for podiatry.
If the same patient also sits on metformin, keep the outcome file and the pain file separate. The Glucophage 1000 outcomes study is about events and legacy, not burning feet.
Pregabalin comparisons exist and are not this page. Different titration, different cost, overlapping adverse events. Do not swap brands mid-climb without resetting the NNT conversation.
A 400 mg capsule is a rung
Saturable intestinal transport means three 400 mg capsules swallowed at once do not equal a clean 1200 mg exposure. Split dosing is pharmacology. The 400 mg strength exists so you can climb without jumping from 100 to 600 in one clumsy step.
Renal clearance owns the ladder. eGFR cuts drop the ceiling and the step size. A 400 mg three-times-daily plan in a seventy-year-old with a creatinine of 2.0 is not 'conservative' - it may already be above the labeled renal row.
Trial titration was forced and time-limited. Clinic titration is often timid and endless. Patients live for six weeks at 400 mg, collect dizziness, and never reach the 1800 mg band where PHN NNT was measured. That is a protocol failure, not a drug failure.
Stop rules belong on the same card as the rung: if function and sleep have not moved after a fair climb to a tolerated trial-range dose, come down. Escalating to 3600 mg because 'the label allows it' is how edema and fog accumulate without a responder.
Gralise and gabapentin enacarbil are not three 400 mg immediate-release capsules in a trench coat. Once-daily gastro-retentive and prodrug rows have their own titration and their own trial doses. Swapping brands mid-climb without resetting the schedule is how people under-expose and then blame the NNT.
Rowbotham and Rice wrote the PHN bar
Rowbotham's 1998 JAMA trial put gabapentin on the PHN map with a forced titration and a pain-diary primary. Rice's later multicentre study (Pain, 2001) randomised 334 adults - mean age 73 - to 1800 mg, 2400 mg, or placebo in three divided doses over seven weeks.
Rice's final pain-score drop versus baseline was about 34.5 percent on 1800 mg and 34.4 percent on 2400 mg, against 15.7 percent on placebo. The extra 600 mg did not buy a second miracle. Sleep-interference diaries moved in the same direction. Dizziness and somnolence clustered in the titration week, not as a late surprise.
Those two designs still sit under every 'start 400 mg at night' card. The endpoint was average daily pain, not a neuropathic branding exercise. If your clinic stops at 600 mg because the patient is 'a bit better,' you never entered the dose band the bar was set on.
Later 1800 mg meta-analyses (Yin et al. and others) kept the pain-intensity and PGIC signal and also kept the adverse-event tax: somnolence, dizziness, edema, withdrawals. Efficacy and harm share the same week of the protocol.
Keep the NNT where the patient can see it
PHN is the cleanest gabapentin neuropathy file: NNT about 6.7 for substantial relief at 1200 mg or more. DPN is allowed but thinner. Most other nerve-pain labels are a shrug.
Neurontin 400 mg is a rung. The trial bar sat at 1800 to 3600 mg for many protocols, split across the day, with renal math. Under-dosing for a month is not a fair test.
Name dizziness, edema, and the opioid breathing stack on day one. Taper after a failed climb. Do not convert a non-responder into a chronic sedative.
Formulary holds stay on the Neurontin page. This study file only keeps the NNT ink from washing off in clinic talk.
Talk with your own clinician or pharmacist before you change a tablet or a dose. Open the ledger disclaimer.
Last Updated
Reader mail
Reader questions on this article
Answered by Dr. Priya Nair, PharmD · Clinical pharmacology and drug safety
People hear 'nerve pill' and expect silence. These answers keep PHN trial NNT on the desk, including what a 400 mg capsule is actually for.
A week at 400 mg is a first rung, not a trial. PHN studies that produced the NNT of about 6.7 used 1200 mg or more, often 1800 to 2400, split through the day, after a forced climb. If dizziness already owns you at 400 mg, that is useful information - you may not tolerate the band where benefit was measured. If you are clear-headed, the honest next step is a planned climb with a stop date, not a declaration that the class failed. Details on saturable absorption sit on the Neurontin 400 mg line.
What does NNT 6.7 actually mean for my shingles pain?
It means you treat about seven people with gabapentin at a trial-range dose so one extra person gets substantial relief - half their pain or 'very much improved' - compared with placebo. In the Cochrane PHN pool, 32 percent on drug hit that mark versus 17 percent on placebo. You might be the one. You might be in the majority who are not. I would rather you hear that before week six than after a foggy winter on 400 mg with no written plan.
Can I take three 400 mg capsules at breakfast to 'hit 1200'?
You can swallow them. You will not get a clean 1200 mg exposure. Intestinal transport saturates. Split the total. Rice's PHN protocol used three divided doses. Morning-loading a day's worth is how people collect peak dizziness and still under-expose the evening pain. If renal function is reduced, even a 'small' 400 mg three times daily may already be too much - check the labeled eGFR row before you invent a schedule.
I have diabetic burning feet, not shingles. Same NNT?
No. Painful diabetic neuropathy has a thinner, noisier file. Cochrane still allows gabapentin a role at 1800 to 3600 mg for some people with DPN, but you should not borrow the PHN 6.7 as if it were a foot guarantee. Shoes, glucose, and ulcer care remain the disease column. This capsule is a symptom stamp. If you also take metformin, keep that outcome story on its own page - the UKPDS outcomes file does not treat burning soles.
My clinic added oxycodone to the 400 mg. Is that stacked risk?
Yes. The 2019 FDA warning on gabapentinoids plus opioids is about breathing, especially in COPD and in the elderly. It is not a theoretical. If both drugs are truly needed, the chart should name the stack and the monitoring plan. If the oxycodone is 'just in case,' that is a bad reason to combine two respiratory depressants. Bring the full list, including sleep aids, to the next visit.
Why do I have swollen ankles on a 'small' dose?
Peripheral edema is a known, dose-related gabapentin effect. It can show up before you ever reach 1800 mg. Weigh yourself, look at the shins, and do not assume new heart failure without an exam. Sometimes a step down clears the swelling. Sometimes the drug is the wrong stamp for you even if pain moved a little. Report it instead of buying compression socks and staying silent.
Is this basically a benzo? The name sounds like GABA.
The name is a tease. Gabapentin binds alpha-2-delta calcium-channel subunits. It is not a GABA-A agonist and it is not a benzodiazepine. You can still get sedation, and you can still have withdrawal after chronic use, which is why we taper. Do not mix it with leftover diazepam to 'help sleep' on the climb. That is a second CNS stack the PHN trials were not designed to bless.
How long before we call it a fail and come off?
After a fair climb to a tolerated dose in the trial band - for many adults with adequate kidneys, that means weeks at 1800 mg-class totals, not days at 400 - if sleep and function have not moved, come down. Write the stop date when you write the start. Open-ended 'see how you go' is how 400 mg becomes a forever sedative. Taper after months. Do not slam the door after a long course.
Should I read depression drug reviews the same way?
Same habit, different file. Pooled effect sizes and onset windows in the Prozac 10 mg meta file are also not personal guarantees. Do not merge HAM-D weeks with PHN pain diaries. If you are on both, each desk needs its own stop rule. I will not treat your mood trial as proof your feet should quiet, or the reverse.
General education from a clinician, not personal medical advice. Bring your own history to your own prescriber.