Pain drugs and cognition

Agent report for Kya, 4 September 2026. Evidence-based; guideline and trial figures are cited with URLs. Where something is clinical judgement rather than citation it is labelled as such. No prognosis or survival figures anywhere in this file. Not medical advice: it is a brief so her partner and a friend can ask the right questions of the ward, the pain team and the pharmacist. Breastfeeding: the patient is not breastfeeding, so it is not a constraint on any drug choice below.

The question this answers

Her partner's worry is "losing the patient" to the drugs: sedation, slowed thinking, confusion, word-finding trouble, flattening. That is a legitimate and well-described set of harms and it is the right thing to design the plan around. But the design has to include the other side of the ledger: her pain is uncontrolled, she is not sleeping, and she is taking unpredictable as-needed morphine on top of that. All three of those also damage thinking, and the evidence says they may be doing more damage right now than a steady, low, well-chosen drug regimen would. Section 2 sets that out honestly.

Pain phenotype (from her partner's account and the obs log): continuous right trigeminal-territory ache with flares, 12 to 14 months, daily 4 to 5 months, anti-inflammatories helped for a year, codeine did nothing, as-needed Oramorph barely touches it. New since about 1 September: involuntary neck jolts on waking, every 20 seconds fading over minutes, "cannot tell whether the jolt causes the pain or the pain causes the jolt", settled by ibuprofen. Right third nerve signs (ptosis, diplopia). Weight 51 kg, down from 58 to 60. Kidney and liver function assumed normal, bloods pending. Regular diclofenac started 4 Sep; morphine still as-needed only; dexamethasone held pending biopsy because of lymphoma on the differential.

1. Side-effect profiles by drug: thinking, sleepiness, mood

The figures below are from randomised trials in neuropathic pain populations (mostly diabetic neuropathy and post-herpetic neuralgia, sometimes trigeminal neuralgia or epilepsy), not from skull-base tumour patients. Treat them as the best available approximation. "Placebo" figures are given where the review reports them because they show how much of an effect is the drug and how much is the illness.

1a. Paracetamol

• Cognitive, sedative, mood effects: essentially none at therapeutic doses. It is the one analgesic with no meaningful central sedative or cognitive signal in trials or pharmacovigilance.
• Analgesic value here: modest on its own, but it is the base layer of every guideline ladder (WHO 2018, ESMO 2018) and it is opioid-sparing in combination. Whether it helps her nerve pain is uncertain; it costs nothing cognitively to find out.
• Dose at 51 kg: the BNF and MHRA advise dose reduction for adults under 50 kg. She is 1 kg over the line and losing weight, so ask the ward pharmacist whether 1 g four times daily or a weight-adjusted dose (around 15 mg/kg, roughly 750 mg) is right. This is a question for them, not a decision for the family.
• Interactions and pre-biopsy: no bleeding effect, no interaction with anything else on this list. Liver function should be normal, which the pending bloods will confirm.

1b. NSAIDs: diclofenac (started today), naproxen, ibuprofen, and the celecoxib option

• Cognitive and sedative effects: low. NSAIDs do have documented central effects (dizziness, headache, and in case series drowsiness, confusion and rarely psychosis, most often with indomethacin and mostly in older patients), but in a 34-year-old these are uncommon. Reviews: Hoppmann 1991 (Arch Intern Med) and Auriel 2014 (Handbook of Clinical Neurology), https://pubmed.ncbi.nlm.nih.gov/24365321/. For her, the cognitive cost of an NSAID is close to zero and the benefit is proven by a year of her own experience.
• Why they worked: an anti-inflammatory that works for a year on a skull-base lesion is a real signal. Bone and dura are prostaglandin-rich; NSAIDs also blunt the inflammatory sensitisation of the trigeminal nerve. Her report that ibuprofen settles the neck spasms fits.
• The three problems that matter for her specifically: 1. Bleeding before biopsy. Non-selective NSAIDs (diclofenac, naproxen, ibuprofen) inhibit platelet COX-1. Naproxen and diclofenac have a reversible platelet effect that clears within a few days of stopping; ibuprofen within about 24 hours. Neurosurgical and interventional practice commonly holds them 3 to 7 days before an intracranial procedure. Celecoxib (a COX-2 selective NSAID) does not impair platelet function even at 600 mg for 10 days in healthy adults (Leese 2000, J Clin Pharmacol, https://www.researchgate.net/publication/12650246) and does not need to be stopped for procedural bleeding risk. So the question for the team is: "If the NSAID is going to be stopped before the biopsy, could she be switched to celecoxib rather than to nothing, so she does not lose the anti-inflammatory benefit in the days before the procedure?" Clinical judgement: celecoxib has a similar analgesic effect to diclofenac and a lower GI bleeding rate, with a similar cardiovascular signal; in a 34-year-old with no cardiovascular history that signal is small. 2. Kidney risk at 51 kg with weight loss and a night-time BP of 92/50. NSAIDs reduce kidney blood flow. Risk factors for NSAID kidney injury are volume depletion, low blood pressure, existing kidney disease, and combination with ACE inhibitors or diuretics (Medsafe 2013, https://www.medsafe.govt.nz/profs/PUArticles/June2013NSAIDS.htm; general reviews above). She has two soft risk factors: low BP overnight and 12 to 15 percent weight loss with possible poor intake. Her creatinine from 3 Sep must be seen before regular diclofenac is continued more than a few days, and she should be drinking. Ask for: "urea, creatinine and eGFR from yesterday's bloods, and a repeat in 3 to 5 days while she is on a regular NSAID." 3. Cardiovascular. MHRA 2013: systemic diclofenac carries a small increase in arterial thrombotic events similar to COX-2 inhibitors; contraindicated in established heart or vascular disease (https://www.mims.co.uk/diclofenac-cardiovascular-warnings-contraindications-updated/musculoskeletal-disorders/article/1192344). Not relevant at 34 with no vascular history, but it is why naproxen is often preferred for longer courses. Naproxen worked for her for a year; if the team is open to it, "naproxen 500 mg twice daily with a proton pump inhibitor" is a reasonable alternative to diclofenac and has the best cardiovascular profile of the non-selective NSAIDs.
• GI protection: regular NSAID plus low weight plus stress plus a possible future steroid course is a GI-bleed set-up. A proton pump inhibitor (omeprazole or lansoprazole) alongside is standard; ask.
• Mood: no meaningful signal.

1c. Gabapentin

• What the Cochrane review found (Wiffen 2017, CD007938, 37 trials, 5914 participants, https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007938.pub4/full): at 1200 mg/day or more, dizziness 19% vs 6.6% placebo; somnolence, drowsiness or sedation 14% vs 5.2% placebo; peripheral oedema 7%; gait disturbance 14%. Withdrawal because of side effects about 11% vs 8% placebo. Serious adverse events no different from placebo.
• Cognitive effects specifically: in a randomised crossover study in healthy volunteers, Meador 2001 (Neurology, https://pubmed.ncbi.nlm.nih.gov/11422333/) found both carbamazepine and gabapentin produced mild cognitive effects compared with no drug, with gabapentin slightly better on some measures; the difference between the two was modest. The commonest patient reports are fogginess and slowed thinking during titration, dose-dependent.
• Time course: dizziness and drowsiness are worst in the first one to two weeks and during each dose increase, and settle in most people over two to four weeks. Slow titration reduces them. (FDA label for Neurontin and the Cochrane review; the "settles in 2 to 4 weeks" figure is consistent across product information and reviews but is not from a single trial.)
• Renal: gabapentin is cleared entirely by the kidney; dose must be reduced if eGFR is low. Another reason the creatinine result matters.
• Interaction that matters most for her: gabapentin plus opioids. MHRA Drug Safety Update, October 2017: gabapentin has been associated with a risk of severe respiratory depression, especially when combined with opioids and in people with respiratory or renal impairment, and doses should be adjusted accordingly (https://www.gov.uk/drug-safety-update/gabapentin-neurontin-risk-of-severe-respiratory-depression). In practice: if she starts gabapentin while on morphine, the sedation from both adds up and the team should start at the low end and watch her breathing and drowsiness. NG193 (2021) advises against starting gabapentinoids for chronic primary pain; that guidance does not apply here, because this is secondary pain with a structural cause, and CG173 (neuropathic pain) still lists gabapentin as a first-line choice.
• Mood: generally neutral; occasional low mood or irritability reported; suicidality warnings exist for all antiepileptics but the signal is small.

1d. Pregabalin

• Cochrane (Derry 2019, CD007076, 45 trials, 11,906 participants, https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007076.pub3/full): somnolence 11% at 300 mg/day and 15% at 600 mg/day vs 3.1% and 4.5% placebo; dizziness 13% at 300 mg and 22% at 600 mg vs 3.8% and 4.4% placebo. About 1 in 10 stopped because of side effects. Serious adverse events no different from placebo.
• Cognitive: similar profile to gabapentin (same mechanism, same class); dose-dependent, worst during titration, improves over weeks. Weight gain and peripheral oedema more prominent than with gabapentin; for someone losing weight that is not a problem.
• Practical differences from gabapentin: twice-daily dosing, more predictable absorption, faster to reach a useful dose (150 mg/day in a week rather than two to three weeks). Same MHRA warning about combination with opioids. Same renal clearance issue.
• Honest ranking against gabapentin: no head-to-head trial shows one is cognitively kinder; the choice is usually about speed of titration and dosing convenience. Sedation rates in the Cochrane reviews are of the same order.

1e. Carbamazepine

• The trigeminal neuralgia drug, and the one with the strongest evidence for stabbing, paroxysmal, trigger-evoked pain (NICE CG173 1.1.13: offer carbamazepine as initial treatment for trigeminal neuralgia; EAN 2019 Bendtsen: first choice, https://onlinelibrary.wiley.com/doi/10.1111/ene.13950; Cochrane Wiffen 2014, NNT about 2 for TN).
• It is also the drug on this list most likely to cause exactly what her partner fears. Cochrane (Wiffen 2014, CD005451, https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD005451.pub3/full): at least one adverse event in 66% on carbamazepine vs 27% on placebo, NNH 2.6. In the largest real-world tolerability series in trigeminal neuralgia using the Liverpool Adverse Events Profile (Besi 2015, J Headache Pain, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4558989/): tiredness 31%, memory problems 23%, sleepiness 18%, disturbed sleep 14%, difficulty concentrating 12%, unsteadiness 12%. Women reported significantly more side effects than men, and the same authors put the usual toxic threshold in women at about 800 mg/day of carbamazepine.
• Cognitive effects in the epilepsy literature (Meador's comparative studies; Loring review, https://med.emory.edu/departments/neurology/_documents/loring_aeds_np_review.pdf): modest psychomotor slowing with reduced attention and memory, dose-related, present at steady state and not simply a titration phenomenon. Slow titration reduces the drowsiness and unsteadiness but does not abolish the slowing.
• Hyponatraemia: reported in 4 to 40% of carbamazepine users across studies; a large Dutch cohort found sodium under 134 in 26% on carbamazepine and 46% on oxcarbazepine (Berghuis 2017, Epilepsia, https://discovery.ucl.ac.uk/1558510/). Low sodium itself causes confusion, nausea, headache and unsteadiness, which can be mistaken for drug sedation. Sodium must be checked before starting and at one to two weeks.
• Enzyme induction: carbamazepine is a strong CYP3A4 inducer. It lowers levels of many other drugs (hormonal contraception, many antiepileptics, some antiemetics, and steroids including dexamethasone, whose clearance it increases) and its own level changes over the first weeks as it induces its own metabolism. If she later needs dexamethasone or chemotherapy, carbamazepine complicates the dosing of both. This is a real reason to prefer oxcarbazepine or a non-inducer if a sodium-channel drug is wanted.
• Blood counts and liver: rare but serious marrow suppression and hepatitis; FBC and LFTs before starting and early in treatment. Serious skin reactions (Stevens-Johnson) are strongly associated with HLA-B*1502 in people of Han Chinese or Thai ancestry; screening applies to that group.
• BNF dosing for trigeminal neuralgia: initially 100 mg once or twice daily, increased gradually according to response; usual 200 mg three to four times daily; up to 1.6 g daily. NICE CKS advises the same slow stepwise increase and referral if it is not tolerated or not working.
• Mood: generally mood-stabilising rather than flattening; sedation and fatigue are the problems, not emotional blunting.

1f. Oxcarbazepine

• Same mechanism and same indication as carbamazepine; the EAN 2019 guideline lists it as equal first choice for trigeminal neuralgia and says it "may possibly be associated with fewer side effects than carbamazepine but both drugs show reduced tolerability in females." A meta-analysis of the three comparative trials and a 2021 real-world series of 354 patients (Di Stefano, Eur J Pain, https://onlinelibrary.wiley.com/doi/abs/10.1002/ejp.1727) found side effects in roughly 30% on oxcarbazepine vs about 44% on carbamazepine with similar efficacy; tolerability rated good or excellent by 62% vs 48%.
• Cognitive: the same family of effects as carbamazepine (tiredness, slowed thinking, memory) at a somewhat lower rate. Not cognitively clean; cognitively less bad.
• Hyponatraemia is the price: rates are consistently higher than with carbamazepine (23 to 73% across studies; 46% in Berghuis 2017; severe sodium under 125 in about 2.5 to 3%). Sodium check before starting, at one to two weeks, and after any dose increase. Combination with an SNRI (duloxetine), which can also cause SIADH, adds to the risk.
• Enzyme induction is much weaker than carbamazepine, so it interferes far less with steroids, chemotherapy or other drugs. This is the main practical advantage for a patient who may need those things.
• Cochrane (Zhou 2017, CD007963, https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007963.pub3/full): more withdrawals for adverse events than placebo (RR 2.84 in radiculopathy trials). NICE CG173 says oxcarbazepine should not be started for neuropathic pain in non-specialist settings unless a specialist advises; on a ward with a pain team that condition is met.
• If a sodium-channel drug is needed for a paroxysmal component, clinical judgement is that oxcarbazepine is the better fit for her than carbamazepine because of interactions, provided sodium is monitored.

1g. Amitriptyline and nortriptyline

• Cochrane (Moore 2015, CD008242, https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD008242.pub3/full): at least one adverse event in 55% vs 36% placebo, NNH 5.2; serious adverse events rare; withdrawals not different from placebo. Finnerup 2015 (Lancet Neurol) gives tricyclics the best NNT of any neuropathic pain class (about 3.6).
• The characteristic side effects are dry mouth, morning grogginess, constipation, blurred vision, postural dizziness and, at higher doses, slowed thinking. Amitriptyline is sedating at all doses and that is partly why it is given at night: used well, the sedation is turned into sleep, which is one of the things she most needs. Weight gain is common, which for her is neutral or helpful.
• Cognitive: anticholinergic drugs impair memory and attention in a dose-related way; the effect is well documented in older people and is small at 10 to 25 mg in a 34-year-old but not nil. Morning hangover is the usual complaint and it improves over the first week or two.
• Nortriptyline: the same analgesic action with less sedation, less postural hypotension and fewer anticholinergic effects (Canadian Pain Society consensus, Moulin 2014; https://www.cfp.ca/content/64/9/634.3). The Cochrane review of nortriptyline (Derry 2015) found no good trial evidence either way for pain, so its use is by class extrapolation. Clinical judgement: if her partner's priority is minimal daytime fog and the team wants a tricyclic, nortriptyline 10 mg at night is the cognitively gentler version; if her sleep is the bigger problem, amitriptyline's sedation is a feature.
• BNF dosing (neuropathic pain, off-label): amitriptyline 10 mg at night, increasing if necessary to 75 mg at night; higher only under specialist supervision.
• Interactions: additive sedation with opioids and gabapentinoids; serotonergic (avoid tramadol and do not combine with duloxetine without specialist input); QT prolongation with some antiemetics (ondansetron, domperidone) at higher doses, so an ECG at baseline is reasonable.
• Mood: at analgesic doses it is not an antidepressant dose; emotional flattening is not a typical effect at 10 to 50 mg. Some people feel groggy and slowed, which can look like flattening.

1h. Duloxetine

• Cochrane (Lunn 2014, CD007115, https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007115.pub3/full): nausea 29% vs 10% placebo (NNH 6.3), somnolence about 14% (NNH 11), constipation (NNH 11), reduced appetite (NNH 18); 16% stopped because of side effects; serious adverse events 2 to 3% and no different from placebo. Analgesic separation from placebo starts at about one week, maximum at about four weeks.
• Cognitive: the least sedating of the antidepressant options and, in trials, the most cognitively neutral; the main complaints are nausea (usually first one to two weeks), sweating, insomnia in some, and a discontinuation syndrome if stopped abruptly.
• The concerns for her: reduced appetite in someone who has lost 12 to 15 percent of body weight; hyponatraemia (SIADH) especially in combination with oxcarbazepine or carbamazepine; increased bleeding risk when combined with an NSAID (serotonergic drugs impair platelet serotonin uptake; the combination raises upper GI bleeding risk, another reason for a PPI); serotonin syndrome if combined with tramadol, other antidepressants or some antiemetics (metoclopramide, ondansetron at high dose, and pethidine, fentanyl or tramadol).
• Dosing (BNF): 60 mg once daily; many clinicians start at 30 mg for a week to reduce nausea.
• Mood: it is an antidepressant at these doses. For someone in a frightening situation with a year of untreated pain that may be a benefit. Emotional blunting is reported with SSRIs and SNRIs by a minority; in trials of duloxetine for pain it is not a prominent finding, but it is the one drug on the list where "feeling less like herself emotionally" is a plausible effect to watch for.

1i. Lamotrigine

• Cochrane (Wiffen 2013, CD006044, https://www.cochrane.org/evidence/CD006044_lamotrigine-antiepileptic-drug-chronic-neuropathic-pain-or-fibromyalgia): no convincing evidence of benefit in neuropathic pain at 200 to 400 mg/day; rash in about 1 in 27 (nearly 10% in some trials); serious skin reactions estimated at about 1 in 1000.
• For trigeminal neuralgia specifically, the EAN 2019 guideline lists it as an add-on or alternative that "may possibly be associated with fewer side effects than carbamazepine and oxcarbazepine", on the basis of one small crossover trial.
• Cognitive: the cleanest of the antiepileptics on cognition (it is used in epilepsy partly for that reason); it can cause insomnia rather than sedation.
• The problem: it must be titrated over about six weeks to avoid rash, so it cannot control pain now. It is a later option if a sodium-channel drug is needed long term and the others are not tolerated.

1j. Topical and interventional options

• Lidocaine 5% plaster (Versatis). 12 hours on, 12 hours off, applied over the painful skin. Minimal systemic absorption, no sedation, no interactions. Evidence in trigeminal neuralgia: the PATCH randomised withdrawal trial found a reduction in weekly mean pain severity versus vehicle (https://pmc.ncbi.nlm.nih.gov/articles/PMC8330571/); a retrospective series of 103 patients found about 1 in 5 responders (https://pubmed.ncbi.nlm.nih.gov/36039508/). Modest but the cognitive cost is zero, so it is worth a trial on the cheek and temple. Cannot be applied over the eye or on broken skin.
• Intranasal lidocaine (10 mg per spray, two sprays to the affected nostril) is the RCS 2021 bridge recommendation for trigeminal neuralgia flares. No cognitive effect.
• Sphenopalatine ganglion block. Through the nose, local anaesthetic on a cotton applicator or via a nasal applicator, repeatable, low risk. Series in head and neck cancer pain: pain scores reduced 38 to 80% with average relief about 23 days (https://www.ncbi.nlm.nih.gov/books/NBK557751/); trigeminal neuralgia pilot 2025 (Cureus). Best for maxillary (cheek, upper jaw) pain. No cognitive effect. This is an acute pain team question and is not a problem before biopsy because it does not touch the lesion.
• Gasserian ganglion procedures (radiofrequency, glycerol, balloon compression). Not appropriate before diagnosis: the ganglion sits in Meckel's cave beside the cavernous sinus, in or next to the mass, and any procedure there is an MDT and skull-base decision, not a pain-team one. Say this plainly if anyone raises it.
• Greater occipital nerve block. Clinical judgement: for the neck and back-of-head component it is low-risk, cognitively free and done at the bedside by many pain teams; worth asking about.

1k. Opioids: as-needed versus low regular background dose

• What the literature says about opioids and thinking:
• Kurita 2009 systematic review of controlled studies in cancer pain (Support Care Cancer, https://pubmed.ncbi.nlm.nih.gov/18762994/): six of ten studies found poorer reaction time, attention, balance or memory with opioids; deficits were associated with dose increases and with the use of supplemental (as-needed) doses; clinical relevance uncertain.
• Kamboj 2005 (Pain, https://journals.lww.com/pain/abstract/2005/10000/the_effects_of_immediate_release_morphine_on.16.aspx): in palliative care patients already on modified-release opioids, an immediate-release morphine dose on top produced measurable anterograde and retrograde memory impairment and slower complex tracking for the hours after the dose, while subjective sedation was not much changed. This is the direct evidence that the peaks of as-needed dosing are what cost memory.
• Tassain 2003 (Pain, https://pubmed.ncbi.nlm.nih.gov/12855350/): 12 months of oral sustained-release morphine in chronic non-cancer pain did not impair cognition and moderately improved some measures, attributed to pain relief and better mood.
• Dhingra 2015 review (Pain Medicine, https://academic.oup.com/painmedicine/article-abstract/16/suppl_1/S37/2472485) and FPM Opioids Aware (https://www.fpm.ac.uk/opioids-aware-clinical-use-opioids/side-effects-opioids): cognitive effects and sedation are most concerning in the first few days after starting or increasing, and tolerance to sedation usually develops within days; constipation does not.
• Reading of that for the patient: a steady low background (modified-release morphine or oxycodone, twice daily) with occasional rescue is more predictable for thinking than hourly as-needed Oramorph, and stable levels allow tolerance to the sedation to develop. NICE CG140 1.1.4 and 1.1.5: regular oral morphine with rescue doses; typical starting total of 20 to 30 mg per day in an opioid-naive patient; review frequently during titration; laxative for everyone.
• Opioid-sparing: yes, achievable, and the whole point of the rest of this document. Finnerup 2015 (Lancet Neurol) puts strong opioids third line for neuropathic pain (NNT 4.3, NNH 11.7) behind tricyclics and gabapentinoids; the RCS 2021 trigeminal neuralgia guideline says there is no evidence for opioids in that condition. Every effective adjuvant (NSAID, neuropathic agent, topical lidocaine, nerve block) lowers the opioid requirement and therefore the cognitive cost.
• Morphine vs oxycodone: no consistent difference in cognitive effects; oxycodone is sometimes better tolerated for nausea and hallucinations, morphine is the NICE first choice. If she has renal impairment on the pending bloods, morphine's active metabolites accumulate and oxycodone is safer. Codeine failing tells you nothing about morphine: about 7 to 10 percent of Europeans cannot convert codeine into morphine at all.
• Opioid-induced myoclonus is a recognised effect of morphine at any dose and is worth raising given the neck jolts; if the jolts are worse on the nights she has had more Oramorph, that pattern matters.

1l. Ketamine and lidocaine infusions (inpatient rescue)

• Intravenous lidocaine: meta-analysis (Lee 2019, J Palliat Med, https://www.researchgate.net/publication/330207312) found 4 to 5 mg/kg over 30 to 80 minutes superior to placebo for greater than 50% reduction in cancer pain; data scarce. Side effects during infusion: light-headedness, perioral tingling, drowsiness, transiently; requires cardiac monitoring; no lasting cognitive effect. It is a legitimate acute pain team rescue for a refractory neuropathic flare and can buy days while oral drugs load.
• Ketamine: Cochrane (Bell 2017, CD003351) found insufficient evidence for benefit as an opioid adjuvant in cancer pain. Its side effects are exactly what her partner fears: dissociation, vivid dreams, hallucinations, confusion, in a substantial minority even at low doses. Clinical judgement: not a first choice for her; a last resort in a monitored setting.

1m. Dexamethasone

• The lymphoma conflict: steroids before biopsy roughly triple the chance of a non-diagnostic sample (Scheichel 2021 pooled OR 3.3; Tosefsky 2024, Neurosurgery, https://pubmed.ncbi.nlm.nih.gov/38865324/; EHA-ESMO 2024 CNS lymphoma guideline). The team is right to hold it. The right response is to expedite the biopsy and maximise non-steroid analgesia, not to argue for steroids.
• Its own psychological effects are large and often forgotten. Classic figures: severe psychiatric reactions in about 6% and mild to moderate reactions in about 28% of steroid-treated patients (Lewis and Smith 1983; reviewed in Warrington and Bostwick 2006, Mayo Clin Proc, https://www.mayoclinicproceedings.org/article/S0025-6196(11)61160-9/pdf); dose-related in the Boston Collaborative Drug Surveillance data (1.3% under 40 mg prednisolone equivalent, 4.6% at 41 to 80 mg, 18.4% above 80 mg). Short courses tend to cause euphoria, agitation and insomnia; longer courses low mood. Dexamethasone disrupts sleep in a dose-dependent way (PLoS One 2024, https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0296028). Onset within days.
• So if steroids do come in after the biopsy, expect sleep loss, restlessness and mood change, give the dose in the morning, and know that these are the steroid and not the patient.

1n. Interactions summary (the ones that matter for her)

• NSAID plus biopsy: hold non-selective NSAIDs for the pre-procedure window the surgeons specify; celecoxib is the platelet-neutral alternative.
• NSAID plus low BP, low weight, poor intake: kidney risk; check creatinine; drink; PPI cover.
• NSAID plus duloxetine (or any SSRI/SNRI): higher GI bleeding risk; PPI.
• Gabapentin or pregabalin plus opioid: additive sedation and respiratory depression (MHRA 2017); start low.
• Carbamazepine: induces metabolism of dexamethasone, hormonal contraception, many chemotherapy and antiemetic drugs; oxcarbazepine much less so.
• Carbamazepine or oxcarbazepine plus duloxetine: additive hyponatraemia risk.
• Serotonergic stacking: duloxetine or amitriptyline with tramadol, pethidine, fentanyl, metoclopramide, ondansetron, linezolid or St John's wort raises serotonin syndrome risk. Tramadol should simply be avoided in this patient.
• Amitriptyline plus opioid plus gabapentinoid: three sedatives; add one at a time.
• Breastfeeding: not applicable.

2. Which drugs are least likely to affect thinking and communication

Ranked from least to most likely to change how she thinks and speaks, at the doses that would be used here. Honest uncertainty: the ranking within each band is judgement; the bands themselves are supported by the trial figures in section 1.

Band A: essentially no cognitive cost 1. Paracetamol. 2. Lidocaine plaster, intranasal lidocaine. 3. Sphenopalatine ganglion or occipital nerve block. 4. NSAIDs (diclofenac, naproxen, ibuprofen, celecoxib). Cognitive cost near zero; the costs are kidney, gut and platelets, all monitorable.

Band B: small, mostly at the start, mostly avoidable by going slowly 5. Duloxetine. Least sedating of the neuropathic agents; nausea and appetite are the issues; the one to watch for emotional change. 6. Nortriptyline at 10 to 25 mg at night. Gentler than amitriptyline on daytime fog. 7. Amitriptyline at 10 to 25 mg at night. Morning grogginess, settles over a week or two; turns into sleep if timed right. 8. Low steady background opioid (modified-release morphine or oxycodone, 10 to 15 mg twice daily) with a laxative. Sedation in the first few days, then tolerance; steady levels are kinder than peaks.

Band C: real, dose-related, usually improves over weeks but does not fully vanish 9. Gabapentin or pregabalin. Somnolence 11 to 15%, dizziness 13 to 22% at the doses that work; fog during titration; adds to opioid sedation. 10. Oxcarbazepine. Fewer side effects than carbamazepine but the same family; sodium is the hidden cause of confusion.

Band D: the ones most likely to produce what her partner is describing 11. Carbamazepine. Two thirds report a side effect; memory and concentration complaints in a fifth to a quarter; women affected more; enzyme induction complicates everything that comes after. 12. Hourly as-needed Oramorph. Not because morphine is the worst drug, but because the shape of the dosing produces repeated peaks, each of which measurably impairs memory for hours (Kamboj 2005), with no chance for tolerance to develop. 13. Ketamine. Dissociation and confusion are its signature. 14. Dexamethasone. Not a pain drug in the usual sense; a third of people have a psychological reaction, most commonly insomnia and agitation.

The counter-case, which has to be said plainly

• Uncontrolled pain impairs thinking. Chronic pain is associated with measurable deficits in attention, memory, processing speed and executive function (Moriarty 2011, Prog Neurobiol; systematic review 2021, https://pmc.ncbi.nlm.nih.gov/articles/PMC8492915/). Pain and attention compete for the same resources. A patient in continuous severe pain is not at her cognitive baseline before a single drug is given.
• Sleep loss impairs thinking and increases pain. One night of sleep deprivation produces generalised hyperalgesia, and sleep disturbance predicts next-day pain more strongly than pain predicts next-day sleep (Finan 2013, J Pain; Krause 2019, J Neurosci, https://www.jneurosci.org/content/39/12/2291). Two-hourly obs, pain, and pacing the corridor for Oramorph are costing her sleep every night, and sleep loss is itself a cause of the fog and the word-finding trouble her partner is afraid of.
• As-needed opioid cycling is often worse for clarity than a steady low dose. The direct evidence is Kamboj 2005 (immediate-release doses impair memory for hours) against Tassain 2003 (steady sustained-release dosing over a year did not impair and modestly improved cognition once pain was controlled). A comparison of under-treated pain and opioid medication on cognitive impairment (2022, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8824639/) found both contribute; the answer is not "no opioid" but "the least opioid that controls the pain, given steadily".
• The neck jolts on waking are more likely to be a product of poor sleep, pain and possibly the Oramorph than of any adjuvant she has not yet started.

So the honest position is: the plan that best protects the patient's thinking is the one that controls the pain and lets her sleep with the smallest total sedative load, added one drug at a time so that any change can be attributed and reversed. "No drugs" is not the cognitively safe option; it is the one she has now.

3. A proposed plan: questions and requests for the clinicians

Framed as things to ask for. The ward doctors, acute pain team and pharmacist decide; the family's job is to make sure the questions are asked, the monitoring happens and the stop rules are agreed in advance. One new central drug at a time, 48 to 72 hours between additions, so any change in the patient can be pinned to its cause.

Step 0 (today): the foundations, no cognitive cost

Step 1 (today or tomorrow): reshape the opioid so it is steady, low and small

Ask: "Her pain is continuous. Could she have a low regular background opioid rather than hourly as-needed Oramorph, and could the as-needed dose be kept for genuine breakthrough?"

The goal is a background dose small enough that she is clear-headed and rescue doses are rare. If the neck jolts get worse as the opioid becomes regular, say so: opioid myoclonus is a reason to switch opioid (oxycodone) or reduce.

Step 2 (within 48 hours): the first neuropathic agent, chosen for cognition

Ask the team to characterise the pain first: "Is there a stabbing, electric, touch- or chewing-triggered component, or is it all continuous ache with flares?" That decides which of these comes first.

Option 2A, if the pain is mostly continuous ache (which is what her partner describes): duloxetine first, as the least sedating agent with first-line NICE status.

Option 2B, if sleep is the dominant problem and the team prefers a night-time drug: nortriptyline (or amitriptyline) at night.

Option 2C, if there is a clear paroxysmal, trigger-evoked component (the neck jolts might be this, or might be spasm; the team should decide): oxcarbazepine rather than carbamazepine, because of interactions with future treatment, with a sodium check.

Carbamazepine only if the team has a specific reason to prefer it; if so, same monitoring, expect more side effects, and ask how it will affect dexamethasone and any later drugs.

Step 3 (if Step 2 is insufficient at a tolerated dose, or not tolerated): gabapentinoid

Step 4 (any time pain is severe or a flare is uncontrolled): acute pain team interventions

• Sphenopalatine ganglion block (bedside, through the nose). Ask on day one; repeatable.
• Greater occipital nerve block for the neck and occipital component.
• Intravenous lidocaine infusion as a monitored rescue.
• Patient-controlled analgesia if she is in a flare and needs many rescue doses: it removes the wait and the pacing, and the total dose is usually lower than nurse-administered as-needed.
• Ketamine: only if everything else has failed, and with the family told what it does to perception.

Non-drug measures with evidence or sound rationale

• Sleep protection. Ask for obs to be spaced overnight if her NEWS2 allows (the junior doctor was already asked; ask the consultant to write it in the plan), the lights and noise reduced, and the night-time doses timed so she is not woken for them. Sleep deprivation raises pain sensitivity and predicts next-day pain (Krause 2019; Finan 2013). This is the single cheapest cognitive protection available.
• Heat. Warm compress or wheat bag to the neck and jaw for the spasm component. Evidence is low-quality for neck pain generally but it is safe, immediate and she can control it. Clinical judgement.
• Positioning for the neck. Sleeping with the head supported in neutral, a single firm pillow, side-lying on the left with a pillow between the knees, avoiding lying flat on the back with the head turned. Judgement, not citation.
• Physiotherapy referral for the neck: gentle range-of-movement and relaxation for a guarding spasm. Low risk. Ask.
• Eye patch or occlusion of one eye for the diplopia. Double vision is exhausting and a source of headache; an orthoptist can provide a patch or a Fresnel prism. This is a comfort measure, not a pain drug, and it costs nothing cognitively.
• A pain diary kept by her partner: time, score 0 to 10, what was given, sleep hours, what the patient was like (clear, foggy, flat, herself). Two lines a day. It is the evidence that lets the team adjust with confidence and it is how you tell drug effect from pain effect.
• Dietitian: weight loss plus low appetite plus possible duloxetine or opioid nausea; feeding and hydration protect the kidneys under an NSAID.

What to ask the acute pain service and when to ask for supportive care

• Acute pain team, today: the Faculty of Pain Medicine core standard is assessment of complex inpatient pain within 24 hours. Ask them: "What is the regular plan, which neuropathic agent, what is the stop rule for each drug, can she have a sphenopalatine ganglion block, and who reviews her tomorrow?"
• Specialist palliative care / supportive care team: ask for them now, not later. The words that unlock it: "We understand this is for symptom control, not end of life." NICE CG140 1.1.1 tells staff to address that fear directly. They are the people in the hospital who manage exactly this combination (a mass, neuropathic pain, an opioid, a family worried about sedation) every day, and the FPM standard says referral for cancer pain that is not improving quickly is expected. They also write the discharge plan and hold it in the community.
• Pharmacist: a medicines review for interactions, renal dosing and the paracetamol dose at 51 kg.
• Named consultant: to write the pain plan and the stop rules into the notes so that night staff can act without a review being "declined".

The one-paragraph version for the ward

"the patient's pain is continuous, right trigeminal territory, with an inflammatory component that responded to NSAIDs for a year and a nerve component that codeine and as-needed morphine do not touch. Her family's overriding concern is preserving her thinking and communication. Please could she have: regular paracetamol and NSAID with a PPI and a creatinine check; a low regular modified-release opioid with rescue doses rather than hourly as-needed Oramorph; one neuropathic agent started now, chosen for lowest sedation (duloxetine or nortriptyline for the continuous ache; oxcarbazepine with sodium monitoring if there is a paroxysmal component); lidocaine plaster and nasal spray; a same-week sphenopalatine ganglion block; acute pain team and supportive care team referrals today; overnight obs spaced to protect sleep; and a written stop rule for each drug so that any drowsiness or confusion is acted on the same day."

4. Red flags for the family, by drug class

Escalate to the nurse in charge the same hour, and if unhappy with the response, the registrar, then the consultant.

• Any opioid (morphine, oxycodone, Oramorph): hard to wake, breathing slow (under 10 to 12 a minute) or shallow, pinpoint pupils, confusion or hallucinations, new or worse jerking (myoclonus), no bowel movement for 3 days. These are toxicity signs, not "getting used to it".
• NSAIDs (diclofenac, naproxen, ibuprofen, celecoxib): black or tarry stool, vomiting blood or coffee-grounds, stomach pain, passing much less urine, ankle swelling, sudden rise in blood pressure, wheeze. Not drinking for a day while on an NSAID is itself a flag.
• Carbamazepine, oxcarbazepine, lamotrigine: any rash, mouth ulcers, fever or sore throat (serious skin or blood reactions: stop and tell the doctor immediately); headache with nausea, confusion, unsteadiness or new drowsiness (low sodium: needs a blood test that day); yellowing of skin or eyes; unusual bruising or bleeding.
• Gabapentin, pregabalin: excessive drowsiness, slow or shallow breathing especially with an opioid, unsteadiness leading to a fall, sudden mood change or suicidal thoughts (rare but listed).
• Amitriptyline, nortriptyline: fast or irregular heartbeat, fainting on standing, inability to pass urine, severe constipation, marked confusion, blurred vision that is new (she has diplopia already; the flag is a change).
• Duloxetine: agitation, sweating, fever, tremor, twitching, racing heart, diarrhoea (serotonin syndrome, especially if any other serotonergic drug has been added; tramadol, metoclopramide, ondansetron are the usual culprits); confusion or headache with nausea (low sodium); black stool with the NSAID; a sudden emotional change she or her partner notices; do not stop abruptly.
• Dexamethasone, if it is started after biopsy: sleeplessness, agitation, euphoria or low mood, paranoia, raised blood sugar (thirst, passing more urine), stomach pain, weakness of the thighs. Morning dosing reduces the sleep loss.
• Lidocaine plaster or spray: skin reaction only; systemic toxicity is not a realistic risk at plaster doses.
• Ketamine, if ever used: vivid dreams, feeling detached, hallucinations, high blood pressure. Expected during the infusion; should not persist after it.
• Not drug-related but overriding (from the existing plan): any change in sight, worsening double vision or eye movement, new facial numbness or weakness, sudden severe headache with vomiting or drowsiness, fever with neck stiffness, new limb weakness.

5. Sources

Guidelines and standards - NICE CG173 Neuropathic pain in adults: pharmacological management (2013, updated 2020). https://www.nice.org.uk/guidance/cg173 - NICE CG140 Palliative care for adults: strong opioids for pain relief (2012, updated 2016). https://www.nice.org.uk/guidance/cg140 - NICE NG193 Chronic pain (primary and secondary) in over 16s (2021). https://www.nice.org.uk/guidance/ng193/chapter/recommendations - NICE CKS Trigeminal neuralgia (page not fetchable by this agent; dosing quoted from the BNF). https://cks.nice.org.uk/topics/trigeminal-neuralgia/ - BNF entries: paracetamol, diclofenac, naproxen, celecoxib, gabapentin, pregabalin, carbamazepine, oxcarbazepine, amitriptyline, nortriptyline, duloxetine, lamotrigine, morphine, oxycodone, lidocaine plaster. https://bnf.nice.org.uk/ - Bendtsen L et al. European Academy of Neurology guideline on trigeminal neuralgia. Eur J Neurol 2019. https://onlinelibrary.wiley.com/doi/10.1111/ene.13950 - Royal College of Surgeons (FDS) Guidelines for the management of trigeminal neuralgia 2021. https://www.rcseng.ac.uk/-/media/files/rcs/fds/guidelines/trigemina-neuralgia-guidelines_2021_v4.pdf - Fallon M et al. ESMO Clinical Practice Guidelines: management of cancer pain in adult patients. Ann Oncol 2018. https://pubmed.ncbi.nlm.nih.gov/30052758/ - WHO Guidelines for the pharmacological and radiotherapeutic management of cancer pain (2018). https://www.who.int/publications/i/item/9789241550390 - Faculty of Pain Medicine, Opioids Aware: management of opioid-related side effects. https://www.fpm.ac.uk/opioids-aware-clinical-use-opioids/side-effects-opioids - Faculty of Pain Medicine, Core Standards for Pain Management Services in the UK (2021). https://fpm.ac.uk/standards-guidelines/core-standards - MHRA Drug Safety Update, October 2017: gabapentin and risk of severe respiratory depression. https://www.gov.uk/drug-safety-update/gabapentin-neurontin-risk-of-severe-respiratory-depression - MHRA / EMA 2013: diclofenac cardiovascular contraindications. https://www.mims.co.uk/diclofenac-cardiovascular-warnings-contraindications-updated/musculoskeletal-disorders/article/1192344 - Medsafe 2013: NSAIDs and acute kidney injury. https://www.medsafe.govt.nz/profs/PUArticles/June2013NSAIDS.htm - Moulin D et al. Canadian Pain Society consensus statement on neuropathic pain, 2014 (nortriptyline vs amitriptyline tolerability). Summarised at https://www.cfp.ca/content/64/9/634.3

Cochrane reviews and meta-analyses - Wiffen PJ et al. Gabapentin for chronic neuropathic pain in adults. Cochrane 2017, CD007938. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007938.pub4/full - Derry S et al. Pregabalin for neuropathic pain in adults. Cochrane 2019, CD007076. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007076.pub3/full - Wiffen PJ et al. Carbamazepine for chronic neuropathic pain and fibromyalgia in adults. Cochrane 2014, CD005451. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD005451.pub3/full - Zhou M et al. Oxcarbazepine for neuropathic pain. Cochrane 2017, CD007963. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007963.pub3/full - Moore RA et al. Amitriptyline for neuropathic pain in adults. Cochrane 2015, CD008242. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD008242.pub3/full - Derry S et al. Nortriptyline for neuropathic pain in adults. Cochrane 2015, CD011209. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD011209.pub2/full - Lunn MPT et al. Duloxetine for treating painful neuropathy, chronic pain or fibromyalgia. Cochrane 2014, CD007115. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007115.pub3/full - Wiffen PJ et al. Lamotrigine for chronic neuropathic pain and fibromyalgia in adults. Cochrane 2013, CD006044. https://www.cochrane.org/evidence/CD006044_lamotrigine-antiepileptic-drug-chronic-neuropathic-pain-or-fibromyalgia - Bell RF et al. Ketamine as an adjuvant to opioids for cancer pain. Cochrane 2017, CD003351. https://researchportal.bath.ac.uk/en/publications/ketamine-as-an-adjuvant-to-opioids-for-cancer-pain-cochrane-revie/ - Finnerup NB et al. Pharmacotherapy for neuropathic pain in adults: systematic review and meta-analysis. Lancet Neurol 2015. - Lee JT et al. Lidocaine for cancer pain in adults: systematic review and meta-analysis. J Palliat Med 2019. https://www.researchgate.net/publication/330207312 - Tosefsky K et al. Steroids and non-diagnostic biopsy in CNS lymphoma, meta-analysis. Neurosurgery 2024. https://pubmed.ncbi.nlm.nih.gov/38865324/

Trials and cohort studies - Kamboj SK et al. The effects of immediate-release morphine on cognitive functioning in patients receiving chronic opioid therapy in palliative care. Pain 2005. https://journals.lww.com/pain/abstract/2005/10000/the_effects_of_immediate_release_morphine_on.16.aspx - Tassain V et al. Long term effects of oral sustained release morphine on neuropsychological performance in patients with chronic non-cancer pain. Pain 2003. https://pubmed.ncbi.nlm.nih.gov/12855350/ - Kurita GP et al. The cognitive effects of opioids in cancer: a systematic review. Support Care Cancer 2009. https://pubmed.ncbi.nlm.nih.gov/18762994/ - Dhingra L et al. Cognitive effects and sedation. Pain Med 2015. https://academic.oup.com/painmedicine/article-abstract/16/suppl_1/S37/2472485 - Comparison of the impacts of under-treated pain and opioid pain medication on cognitive impairment, 2022. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8824639/ - Meador KJ et al. Comparative cognitive effects of carbamazepine and gabapentin in healthy senior adults. Neurology 2001. https://pubmed.ncbi.nlm.nih.gov/11422333/ - Loring DW. Neuropsychological and behavioral effects of antiepilepsy drugs (review). https://med.emory.edu/departments/neurology/_documents/loring_aeds_np_review.pdf - Besi E et al. Comparison of tolerability and adverse symptoms in oxcarbazepine and carbamazepine in trigeminal neuralgia using the Liverpool Adverse Events Profile. J Headache Pain 2015. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4558989/ - Di Stefano G et al. Real-world effectiveness and tolerability of carbamazepine and oxcarbazepine in 354 patients with trigeminal neuralgia. Eur J Pain 2021. https://onlinelibrary.wiley.com/doi/abs/10.1002/ejp.1727 - Berghuis B et al. Carbamazepine- and oxcarbazepine-induced hyponatremia in people with epilepsy. Epilepsia 2017. https://discovery.ucl.ac.uk/1558510/ - Dong X et al. Hyponatremia from oxcarbazepine and carbamazepine. Neurology 2005. https://www.neurology.org/doi/10.1212/01.wnl.0000188819.45330.90 - Leese PT et al. Effects of celecoxib on platelet function in healthy adults: randomized controlled trial. J Clin Pharmacol 2000. https://www.researchgate.net/publication/12650246 - Hoppmann RA et al. Central nervous system side effects of NSAIDs. Arch Intern Med 1991. https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/615266 - Auriel E et al. NSAID exposure and the central nervous system. Handb Clin Neurol 2014. https://pubmed.ncbi.nlm.nih.gov/24365321/ - PATCH trial: 5% lidocaine-medicated plaster for trigeminal neuralgia. Headache 2021. https://pmc.ncbi.nlm.nih.gov/articles/PMC8330571/ - Effectiveness, safety and predictors of response to 5% lidocaine plaster in trigeminal neuralgia, 103 patients. Ann Pharmacother 2022. https://pubmed.ncbi.nlm.nih.gov/36039508/ - Sphenopalatine ganglion block, StatPearls (head and neck cancer series: 38 to 80% reduction, about 23 days). https://www.ncbi.nlm.nih.gov/books/NBK557751/ - Ho KWD et al. Sphenopalatine ganglion: block, radiofrequency ablation and neurostimulation, systematic review. J Headache Pain 2017. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5745368/ - Warrington TP, Bostwick JM. Psychiatric adverse effects of corticosteroids. Mayo Clin Proc 2006. https://www.mayoclinicproceedings.org/article/S0025-6196(11)61160-9/pdf - Dexamethasone induces sleep disturbance in a dose-dependent manner. PLoS One 2024. https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0296028 - Moriarty O et al. The effect of pain on cognitive function: a review of clinical and preclinical research. Prog Neurobiol 2011. https://www.researchgate.net/publication/49738338 - Evidence for cognitive decline in chronic pain: systematic review. 2021. https://pmc.ncbi.nlm.nih.gov/articles/PMC8492915/ - Krause AJ et al. The pain of sleep loss: a brain characterization in humans. J Neurosci 2019. https://www.jneurosci.org/content/39/12/2291 - Finan PH et al. The association of sleep and pain: an update and a path forward. J Pain 2013.

Caveats - Trial populations are diabetic neuropathy, post-herpetic neuralgia, trigeminal neuralgia and epilepsy, not skull-base tumour pain. Rates are indicative. - NICE CKS trigeminal neuralgia page returned 403 to this agent; its content was not verified directly this session. Dosing figures are from the BNF. - Finnerup 2015 and Finan 2013 are cited from prior threads and general knowledge of the papers; figures match the published abstracts but the full texts were not re-fetched today. - Neck jolt interpretation (spasm vs paroxysm vs opioid myoclonus vs sleep-deprivation hypnic jerks) is clinical judgement; the ward needs to see one.