A whole paper,
answered in full.

The therapeutic effect at induction is loss of consciousness. Three groups of factors determine whether, how fast and how deeply it occurs: pharmacokinetic (is the drug available at the effect site?), pharmacodynamic (what does it do once there?) and physiological (how is it delivered to and cleared from that site?).

Availability at the effect site. Dose and the rate of administration — a bolus produces a high peak plasma concentration and a steep gradient into brain, whereas a slow injection or infusion does not. Propofol is highly lipid-soluble with a pK_a of about 11, so at physiological pH the free drug is almost entirely unionised and crosses the blood–brain barrier readily. About 98% is protein-bound, so hypoalbuminaemia raises the free fraction. Onset is one arm–brain circulation time; offset after a single dose is redistribution, not metabolism. Clearance is high and exceeds hepatic blood flow, because metabolism is both hepatic and extra-hepatic, principally in the lung.

Action at the effect site. Propofol potentiates the GABA_A receptor, a ligand-gated chloride channel: chloride influx hyperpolarises the neurone and increases inhibitory tone. Co-administered opioids and benzodiazepines act at different sites and shift the dose–response curve leftward, so less propofol is needed — synergy, not simple addition.

Delivery and clearance. Cardiac output determines how much of a bolus reaches the brain and how fast: a low cardiac output means a smaller initial volume of distribution and a higher peak concentration in a slower-arriving bolus, so the same dose produces a deeper effect. Age matters in both directions — children need proportionally larger doses, the elderly smaller ones, through reduced initial distribution volume, reduced clearance and increased brain sensitivity.

Elimination half-life is the time for the plasma concentration to fall by half during the terminal elimination phase. It is a property of the drug in a given patient — t_½ = 0.693 × V_d / Cl — and it is a single number that does not change with how long the drug has been given. Propofol’s terminal half-life is long, of the order of several hours, because a small amount returns slowly from deep, poorly perfused tissue.

Context-sensitive half-time is the time for the concentration to fall by half after an infusion is stopped, where the “context” is the duration of that infusion. It is not a single number: it is a function, and it has to be read off a curve.

They differ because half-life describes a one-compartment idealisation, whereas a real drug fills peripheral compartments during an infusion. When the infusion stops, plasma concentration falls both by clearance and by continued distribution — but the peripheral compartments now return drug to the plasma, opposing the fall. The longer the infusion, the fuller those compartments and the slower the decline. For propofol the CSHT rises only modestly and then flattens, because clearance is high and exceeds liver blood flow; for fentanyl it climbs steeply and keeps climbing. That contrast is the reason propofol suits total intravenous anaesthesia and fentanyl by infusion does not.

A ceiling effect is a dose above which further increase produces no additional therapeutic effect, while adverse effects continue to increase. For intrathecal fentanyl the analgesic response plateaus at around 15 to 25 µg; larger doses do not improve the quality or duration of analgesia but do increase pruritus, nausea, sedation and respiratory depression.

The mechanism is saturation of a finite receptor pool. Fentanyl acts on µ-opioid receptors in the substantia gelatinosa of the dorsal horn, reducing neurotransmitter release from primary afferent terminals and hyperpolarising second-order neurones. Once those receptors are occupied, additional drug cannot produce additional spinal effect. Because fentanyl is highly lipid-soluble, the excess does not linger in cerebrospinal fluid — it is rapidly taken up into the cord and into epidural fat and systemic circulation, so the extra dose behaves increasingly like an intravenous one and produces systemic side effects instead.

On a log dose–response diagram this appears as two curves with different plateaus. The analgesic curve reaches an E_max that is not the maximum possible antinociception and flattens; the side-effect curve continues to climb past that point. The vertical gap between them narrows and then reverses, which is the observation the question is asking to be illustrated.

Both are potent synthetic phenylpiperidine µ agonists with rapid onset and high lipid solubility, used intraoperatively to blunt the sympathetic and nociceptive response and to reduce anaesthetic requirement. The differences all follow from one property: how they are metabolised.

Fentanyl is cleared by hepatic CYP3A4, so its offset depends on liver blood flow and function, and it distributes into a large peripheral reservoir. Its context-sensitive half-time rises steeply with infusion duration — about 20 minutes at one hour, but several hours after prolonged infusion. Remifentanil has an ester linkage hydrolysed by non-specific plasma and tissue esterases — not plasma cholinesterase, so it is unaffected by atypical cholinesterase. Clearance is organ-independent and very high, giving a context-sensitive half-time of 3 to 4 minutes regardless of infusion duration, unchanged in hepatic or renal failure.

The trade-off is therefore titratability against residual analgesia. Remifentanil allows profound intraoperative opioid effect with predictable, immediate emergence — valuable for neurosurgery, for a shared airway, or where rapid neurological assessment is needed. But it leaves no analgesia at all when stopped, and is associated with acute tolerance and opioid-induced hyperalgesia, so a longer-acting analgesic must be given before emergence. Fentanyl’s accumulation is a liability in a long case and an advantage at the end of a short one.

Pharmacodynamic. Monoamine oxidase inhibitors with pethidine — excitatory serotonin syndrome with hyperthermia, rigidity and cardiovascular instability; the same risk applies to tramadol and fentanyl to a lesser degree. Additive central depression with benzodiazepines, propofol, volatile agents, alcohol and gabapentinoids — the benzodiazepine–opioid combination is the classic respiratory-depression pairing. Synergy with propofol at induction. Serotonergic drugs — SSRIs, SNRIs, linezolid — with tramadol or pethidine. Buprenorphine, a partial agonist with very high affinity, blunting the effect of a subsequently given full agonist; naloxone and naltrexone reversing it.

Pharmacokinetic. CYP3A4 inhibitors — erythromycin, clarithromycin, ritonavir, grapefruit juice — prolonging fentanyl and alfentanil; inducers such as rifampicin and carbamazepine shortening them. CYP2D6 governs activation of the prodrugs codeine and tramadol, so a poor metaboliser gets no analgesia and an ultra-rapid metaboliser is at risk of toxicity; fluoxetine and paroxetine inhibit it. Reduced hepatic blood flow from any cause prolongs hepatically cleared opioids.

Nitrous oxide is a weak, insoluble, analgesic anaesthetic — MAC about 104%, blood:gas partition coefficient 0.47 — which cannot produce anaesthesia alone at safe inspired concentrations but is valuable alongside a potent volatile for four reasons.

1. It is MAC-sparing. The MAC values of inhaled agents are additive, so 66% nitrous oxide contributes roughly 0.6 MAC and the volatile can be reduced by that amount. Since the cardiovascular depression of volatiles is dose-dependent, this buys haemodynamic stability — less vasodilatation, less myocardial depression — at the same overall depth.

2. It speeds induction. Two related phenomena. The concentration effect: because nitrous oxide is given in high concentration and taken up rapidly into blood, the remaining alveolar gas is concentrated and drawn in by the resulting volume loss, so its own alveolar partial pressure rises faster than uptake alone predicts. The second gas effect: that same uptake concentrates the volatile in the residual alveolar gas and augments its inspired flow, so the volatile’s alveolar partial pressure also rises faster. Nitrous oxide therefore accelerates the agent given with it.

3. It provides analgesia. Volatile agents are poor analgesics; nitrous oxide is genuinely analgesic, through NMDA receptor antagonism and activation of descending noradrenergic and opioidergic pathways. This reduces intraoperative opioid requirement.

4. It speeds emergence, because its low blood:gas coefficient means rapid elimination once discontinued.

The limits on its use are what determine when the adjuvant is inappropriate: diffusion hypoxia on discontinuation, expansion of closed gas spaces, inhibition of methionine synthase, a high emetogenic potential, and the fact that a fixed 66% ceiling both caps the benefit and restricts the maximum inspired oxygen to about 33%.

Efficacy is whether the agent reliably produces and maintains anaesthesia, and how well it can be controlled.

• Potency, expressed as MAC — the alveolar concentration at which 50% of subjects do not move to a standard stimulus. MAC correlates inversely with the oil:gas partition coefficient (the Meyer–Overton correlation), so lipid solubility predicts potency.
• Speed of onset and offset, governed principally by the blood:gas partition coefficient: a low value means the agent saturates blood quickly, alveolar partial pressure rises fast, and control is tight. Desflurane 0.42 and sevoflurane 0.65 are faster than isoflurane 1.4.
• Delivery factors — inspired concentration, alveolar ventilation, functional residual capacity, cardiac output (a high output slows the rise), the concentration and second gas effects, and the presence of shunt.
• MAC-modifying factors in the patient: age, temperature, pregnancy, opioids, other sedatives, chronic alcohol.

Safety is the margin between the effective and the harmful.

• Therapeutic index — the gap between MAC and the concentration producing unacceptable cardiovascular depression. All volatiles cause dose-dependent vasodilatation and myocardial depression, so this margin is narrow.
• Metabolism, because toxicity generally comes from metabolites, not the parent. Halothane about 20% metabolised, sevoflurane 3–5%, isoflurane 0.2%, desflurane 0.02%. Hence halothane hepatitis, and fluoride from sevoflurane.
• Degradation in the circuit — sevoflurane with strong bases yields compound A; desiccated absorbent with desflurane yields carbon monoxide.
• Organ-specific effects — cerebral vasodilatation raising intracranial pressure above about 1 MAC; airway irritability (desflurane); uterine relaxation; all are malignant hyperthermia triggers.
• Physical properties — stability without preservative, non-flammability, and a boiling point and saturated vapour pressure compatible with a safe vaporiser. Desflurane’s boiling point of 22.8 °C is why it needs a heated, pressurised vaporiser.

The two work in opposite directions on the same competition. A non-depolarising block is competitive antagonism between the relaxant and acetylcholine at the postjunctional nicotinic receptor. Neostigmine increases the competitor; sugammadex removes the antagonist.

Neostigmine is a quaternary amine that reversibly inhibits acetylcholinesterase by forming a carbamylated enzyme complex, which hydrolyses far more slowly than the acetylated complex formed by acetylcholine itself. Acetylcholine therefore accumulates — but at every cholinergic synapse, not only the junction, so muscarinic effects (bradycardia, salivation, bronchospasm, increased gut motility) oblige co-administration of an antimuscarinic: glycopyrronium for its matched onset, or atropine. Because the enzyme pool is finite, neostigmine has a ceiling: once acetylcholinesterase is fully inhibited no further acetylcholine can be recruited, so it cannot reverse a deep block and needs some spontaneous recovery — at least two twitches of the train-of-four — before it will work.

Sugammadex is a modified γ-cyclodextrin: a hollow, doughnut-shaped molecule of eight oligosaccharide units with a lipophilic core and a hydrophilic exterior. It encapsulates the lipophilic steroidal relaxant — rocuronium best, vecuronium well, pancuronium less so — held by van der Waals forces and electrostatic interaction. The resulting complex is inert, so free plasma relaxant falls abruptly, the concentration gradient reverses, and relaxant leaves the junction. It has no ceiling, no cholinergic effect and needs no antimuscarinic, and it reverses a block of any depth if dosed to that depth.

The concern. Suxamethonium produces a transient rise in intracranial pressure, by two mechanisms: fasciculation raising intrathoracic and intra-abdominal pressure and so impeding cerebral venous drainage, and increased afferent activity from muscle spindles raising cerebral blood flow and metabolic rate. The rise is modest and short-lived, of the order of a few mmHg for a few minutes, and has not been demonstrated consistently.

The counter-argument, and it is the stronger one. The things that raise intracranial pressure most in this patient are hypoxaemia, hypercapnia, coughing, straining and an inadequately obtunded laryngoscopy response — all of which are worsened by a slow or failed intubation. Suxamethonium provides the fastest reliable intubating conditions (onset 30–60 s, duration 5–10 min) and the shortest time to a secured, ventilated airway. Its ICP effect is also readily attenuated: an adequate induction dose of propofol or thiopentone, a defasciculating dose of a non-depolariser, intravenous lidocaine, opioid pretreatment, head-up positioning and prompt hyperventilation all blunt it.

What should actually change the decision is rarely the intracranial pressure. It is the other contraindications, which this population frequently has: hyperkalaemia from upregulated extrajunctional acetylcholine receptors after denervation, stroke, spinal cord injury, prolonged immobility or burns — a risk that begins around 24 to 72 hours after the insult and persists for months.

Position. In a patient needing rapid securing of the airway, the transient ICP rise is not a sufficient reason to avoid suxamethonium, provided the response is obtunded and hypoxia and hypercapnia are prevented. Where the insult is more than about 24 hours old, or there is denervation, immobility or burn, avoid it for the hyperkalaemia risk. Rocuronium 1.2 mg/kg with sugammadex available is the reasonable alternative and, where sugammadex is stocked, increasingly the default.

Both classes are compared against the minimum inhibitory concentration (MIC): the lowest concentration preventing visible growth of the organism in vitro. What differs is which feature of the concentration–time curve predicts killing.

Concentration-dependent agents kill faster the higher the peak. The predictive indices are peak:MIC (target roughly 8–10 for aminoglycosides) and AUC:MIC. Examples: aminoglycosides (gentamicin, amikacin), fluoroquinolones (ciprofloxacin, levofloxacin), metronidazole, daptomycin. They also have a prolonged post-antibiotic effect — suppression of growth persisting after the concentration falls below MIC — so a trough below MIC is tolerable and even desirable. Dosing is therefore large and infrequent: once-daily gentamicin gives a high peak for efficacy and a low trough, which matters because aminoglycoside nephrotoxicity and ototoxicity are trough-related. Once-daily dosing is thus simultaneously more effective and less toxic.

Time-dependent agents kill at a rate that saturates a little above MIC; raising the peak further achieves nothing. The predictive index is the percentage of the dosing interval spent above MIC (%T>MIC) — roughly 40–50% for penicillins and carbapenems, 60–70% for cephalosporins. Examples: beta-lactams (penicillins, cephalosporins, carbapenems), vancomycin (strictly AUC:MIC-driven and time-dependent in behaviour), clindamycin, macrolides, linezolid. They have little or no post-antibiotic effect against Gram-negatives, so the concentration must be kept above MIC. Dosing is therefore smaller and more frequent, by extended infusion, or continuous.

The statement is too absolute, and I would not accept it as written. It conflates several distinct situations that carry very different risks. There are, however, real pharmacological and procedural reasons for caution.

Pharmacological objections. Local anaesthetics are weak bases; only the unionised fraction crosses the nerve membrane. Infected tissue is acidic, which shifts the equilibrium towards the ionised form and reduces block quality — the familiar failure of infiltration into an abscess. Increased local blood flow in inflamed tissue also removes drug faster. Systemically, acidosis increases the risk of toxicity: it favours ion trapping within cells and reduces the seizure threshold, and it worsens the myocardial depression of a given plasma concentration. Hepatic hypoperfusionreduces clearance of amide agents, so infusions accumulate.

Procedural objections. Needling through infected tissue risks seeding, and neuraxial puncture during untreated bacteraemia carries a risk of meningitis or epidural abscess. Sepsis-associated coagulopathy and thrombocytopenia raise the risk of vertebral canal haematoma. The sympathectomy of a central neuraxial block is poorly tolerated by a vasoplegic, preload-dependent patient.

The counter-case. Regional techniques reduce opioid requirement, preserve respiratory function, and provide analgesia that is often better than systemic. A peripheral block at a clean site, remote from the infection — an interscalene block for a shoulder in a patient septic from a foot ulcer — carries none of the neuraxial risks.

Position. Not a prohibition, a set of conditions. Avoid neuraxial blockade in untreated bacteraemia, at an infected site, or with deranged coagulation. Peripheral blocks at clean sites are acceptable, ideally after antibiotics have been started, with reduced doses, awareness that acidosis lowers the toxicity threshold, and lipid emulsion available.

Both raise systemic vascular resistance by contracting vascular smooth muscle through a G_q-coupled receptor, phospholipase C and IP₃-mediated calcium release — but through different receptors, and that independence is the clinically important point.

Noradrenaline acts at the α₁ adrenoceptor, with useful β₁ activity at the heart and negligible β₂ effect. It therefore raises SVR and blood pressure while maintaining or modestly increasing cardiac output; heart rate often falls slightly through baroreceptor reflex despite the β₁ effect. It is titratable over a wide range and is the first-line vasopressor in septic and most vasodilatory shock. Its weakness is that its receptor fails when it is most needed: adrenoceptors are downregulated and desensitised in prolonged sepsis, and acidosis blunts the response, so escalating doses achieve progressively less while worsening tachyarrhythmia and regional ischaemia.

Vasopressin acts at the V₁ receptor, a pathway independent of the adrenoceptor and relatively preserved in acidosis and in adrenoceptor downregulation. In septic shock there is a relative vasopressin deficiency — endogenous stores are depleted after initial release — so replacing it at physiological concentration restores a missing mechanism rather than over-driving an exhausted one. It has no inotropic activity, and by raising afterload without supporting the ventricle it may reduce cardiac output in a failing heart. It also acts at renal V₂ receptors, causing water retention.

In practice they are complementary rather than alternatives: noradrenaline is titrated first, and vasopressin is added at a fixed, low, non-titrated dose to spare noradrenaline and to recruit a second pathway.

Start from the mechanism, because it decides the timing. Aspirin irreversibly acetylates cyclo-oxygenase-1, abolishing thromboxane A₂ production; clopidogrel is a prodrug whose active metabolite irreversibly binds the P2Y₁₂ ADP receptor. The platelet is anucleate and cannot resynthesise either protein, so the effect lasts the life of the platelet, not the life of the drug. Aspirin’s plasma half-life is about 20 minutes and is entirely irrelevant to the answer.

Recovery is therefore platelet turnover. Roughly 10–12% of the pool is replaced each day from a lifespan of about 7–10 days, and around 50% functional platelets are generally adequate for surgical haemostasis. That gives the numbers: clopidogrel stopped 5 to 7 days before surgery, and aspirin 7 days if it must be stopped at all.

But the timing is the smaller half of the question. For most surgery, aspirin should be continued: withdrawal produces a rebound prothrombotic state and the excess bleeding is usually modest and manageable. Stop it only where bleeding into a closed space would be catastrophic — intracranial, posterior chamber of the eye, spinal canal — or where the indication is weak primary prevention.

For clopidogrel the decisive question is why it is prescribed. On dual antiplatelet therapy after a coronary stent, premature cessation risks stent thrombosis, which carries high mortality. Elective surgery should be deferred — conventionally at least 1 month after a bare-metal stent and 6 months after a drug-eluting stent, with 3 months acceptable if surgery cannot wait. Where surgery must proceed, continue aspirin throughout and stop only the P2Y₁₂ inhibitor, in discussion with cardiology.