Central Fever & Paroxysmal Sympathetic Hyperactivity
Two things the injured brain does that get mistaken for sepsis. Fever after acute brain injury independently worsens outcome, so normothermia is a treatment goal and not a comfort measure. But both diagnoses here are diagnoses of exclusion, and the expensive error runs in both directions: calling it central too early and missing ventriculitis, or calling it sepsis for three weeks and never treating the storm.
🔍 Approach
⚠ Fever is not a bystander in the injured brain, which is why this matters. After subarachnoid hemorrhage, traumatic brain injury and stroke, fever is independently associated with worse functional outcome and higher mortality, and the association holds after adjusting for injury severity. The mechanism is straightforward: a damaged brain has no metabolic reserve, and raising temperature raises cerebral metabolic demand, excitatory neurotransmitter release and intracranial pressure at exactly the moment perfusion is most precarious. So the target is normothermia, treated as actively as you would treat a blood pressure. The trap is the opposite error: fever is also the first sign of ventriculitis in a patient with an external ventricular drain, and "it is just central" is how that gets missed.
The Differential, Before You Blame the Brain
Both central fever and paroxysmal sympathetic hyperactivity are diagnoses of exclusion. Work through this list first, because most of it is treatable and some of it is lethal.
| Cause | What makes it likely, and how to exclude it |
|---|---|
| Infection always first | Ventriculitis or meningitis in anyone with an EVD, shunt, or skull base fracture -send CSF, do not just culture blood. Ventilator-associated pneumonia, urinary tract infection (usually catheter-associated), line infection, C. difficile, and sinusitis in a patient with a nasogastric or nasotracheal tube, which is easy to forget and easy to see on the CT you already have. |
| Venous thromboembolism | These patients are immobile, often with a contraindication to prophylaxis, so DVT and pulmonary embolism are common and are classic causes of unexplained fever. Low threshold for duplex ultrasound. |
| Drug fever | Phenytoin is the one to know in this population, and antibiotics are the other common culprit. It is diagnosed by stopping the drug, not by a test. |
| Blood in the wrong place | Subarachnoid and intraventricular blood are pyrogenic in their own right, and resorbing hematoma anywhere causes fever. This overlaps with central fever rather than being cleanly separate from it. |
| Other | Transfusion reaction, alcohol or benzodiazepine withdrawal (a common miss in trauma), seizure including non-convulsive status, acalculous cholecystitis, pressure injury, and pancreatitis. |
| ⚠ The hyperthermia syndromes | Neuroleptic malignant syndrome (these patients receive antipsychotics for agitation), serotonin syndrome, malignant hyperthermia, thyroid storm. Each is treated completely differently, which is why they belong on this list rather than being assumed away. |
⚠ Central fever and storming are not the same thing, and the difference is the sympathetic surge. Central fever is isolated hyperthermia: a disturbed thermoregulatory set point, without the rest of the autonomic picture. Paroxysmal sympathetic hyperactivity is fever PLUS a simultaneous surge of tachycardia, hypertension, tachypnea, drenching sweating and posturing, arriving in discrete paroxysms rather than as a plateau. If the temperature rises alone, think central fever. If it rises together with everything else and then subsides, think storming.
🌡️ Central Fever
When Is Fever Actually Central?
There is a validated way to answer this rather than guessing. Hocker, 2013 studied neurologic ICU patients with fever of undetermined cause and found that four features together predict central fever with a probability of 0.90:
- Negative cultures.
- No infiltrate on chest radiograph.
- A diagnosis of subarachnoid hemorrhage, intraventricular hemorrhage, or tumor.
- Fever onset within 72 hours of admission.
Why this is worth knowing rather than just interesting. The practical purpose of the model is permission to stop antibiotics in a patient who is still febrile. Without something like it, ongoing fever drives open-ended antimicrobial courses, repeated cultures and escalating agents in a patient who has no infection, with all the resistance and C. difficile risk that carries. A febrile patient is not automatically an infected patient.
What It Looks Like
- Higher temperatures than infection typically produces, often above 39°C, and sometimes strikingly high.
- A plateau pattern with loss of the normal diurnal variation -the temperature sits high rather than swinging, which is a useful bedside discriminator from an infectious pattern.
- Poor or absent response to antipyretics, discussed below, which is often the first clue.
- Highest risk after subarachnoid hemorrhage, particularly with intraventricular extension, and after tumor, traumatic brain injury and any lesion involving the hypothalamus or brainstem.
⚠ Why acetaminophen and NSAIDs barely work here, which is the single most useful piece of physiology on this page. Ordinary fever is prostaglandin-mediated: pyrogens drive PGE2 production, PGE2 raises the hypothalamic set point, and the body then defends the higher temperature. Antipyretics work by inhibiting cyclooxygenase and lowering that PGE2 signal. Central fever is not generated that way. It results from direct structural or inflammatory disruption of hypothalamic thermoregulation, so there is little prostaglandin signal for an antipyretic to interrupt. The clinical consequence: a fever that shrugs off acetaminophen should raise your suspicion of a central cause rather than prompting a bigger dose, and physical cooling becomes the mainstay because it lowers temperature without needing the set-point pathway to cooperate.
⚡ Storming (PSH)
The name was standardized because the literature had become unusable. Baguley, 2014 replaced more than 31 competing eponyms -sympathetic storming, dysautonomia, diencephalic seizures, autonomic dysfunction syndrome- and nine overlapping diagnostic criteria with one term and one instrument. The agreed definition is simultaneous, paroxysmal, transient increases in SYMPATHETIC activity (heart rate, blood pressure, respiratory rate, temperature, sweating) AND MOTOR activity (posturing) after severe acquired brain injury.
The PSH-AM: Two Halves
| Component | What it does |
|---|---|
| Clinical Feature Scale | Grades the severity of the six cardinal features: heart rate, respiratory rate, systolic blood pressure, temperature, sweating and posturing. This is what you score during an episode. |
| Diagnosis Likelihood Tool | Scores the context: whether features are paroxysmal, whether they are triggered by normally non-painful stimuli, whether they occur simultaneously, whether other causes have been excluded, and whether the picture follows a severe acquired brain injury. |
| Combined | The two subtotals together give an estimate of the probability that what you are watching is PSH. The value of using it is that it forces the diagnosis to be made deliberately rather than by pattern-recognition after weeks of failed sepsis workups. |
Who Gets It, and When
- Most often after severe traumatic brain injury, particularly diffuse axonal injury. Also hypoxic-ischemic brain injury, where it tends to be more severe, and after stroke, hydrocephalus and infection.
- It emerges days to weeks after the injury, and characteristically as sedation is weaned, which is precisely why it is so often misattributed to withdrawal, pain, agitation or a new infection.
- ⚠ Episodes are triggered by stimulation -turning, suctioning, physiotherapy, pain, a distended bladder, constipation. This is both the strongest diagnostic clue and, as below, a genuine treatment target.
⚠ What it gets mistaken for, and how to tell. Sepsis -but storming is paroxysmal rather than sustained, provoked by handling, and the workup stays negative. Seizure -posturing can look convulsive, but there is no EEG correlate, and if there is doubt the EEG settles it. Neuroleptic malignant syndrome -plausible because these patients receive antipsychotics, so review the drug chart before settling on PSH. Opioid or benzodiazepine withdrawal -the timing during sedation weaning is identical, and the two genuinely coexist. Also pulmonary embolism, serotonin syndrome, thyroid storm and pheochromocytoma.
⚠ Untreated storming is not merely uncomfortable to watch. Repeated surges drive enormous catabolism and weight loss, dehydration from sweating, intracranial pressure spikes, and contractures and heterotopic ossification from sustained posturing. It lengthens ICU and hospital stay and is associated with worse functional outcome. Treating it is rehabilitation, not sedation for the staff's benefit.
💊 Treatment
⚠ State the evidence honestly, because it changes how you prescribe. There are no randomized controlled trials guiding drug therapy for either condition. Everything below rests on case series, retrospective cohorts and consensus. That is not a reason to withhold treatment, but it is a reason to start one agent at a time, define what you expect it to do, and titrate to an observed effect rather than layering four drugs on in a night and never knowing which one worked.
Central Fever
| Step | Detail and rationale |
|---|---|
| 1. Exclude infection properly | Cultures, chest radiograph, urine, and CSF if there is any drain, shunt or basal skull fracture. Then use the four Hocker features to decide whether you are entitled to stop antibiotics. Do this before reaching for cooling, because a cooled patient with ventriculitis simply looks better while getting worse. |
| 2. Antipyretics anyway | Acetaminophen is still reasonable to trial: it is cheap, safe at appropriate dosing, and many of these fevers are of mixed cause. Expect a limited response, and treat a poor response as diagnostic information rather than as a reason to escalate the dose. |
| 3. Physical cooling the mainstay | Surface or intravascular cooling to normothermia. This is the mainstay precisely because it bypasses the set-point pathway that antipyretics depend on. Automated feedback-controlled devices hold target far better than ice packs and blankets. |
| ⚠ 4. Manage shivering | Shivering undoes everything cooling was meant to achieve: it raises metabolic rate, oxygen consumption and intracranial pressure. Assess it formally with the Bedside Shivering Assessment Scale and treat in steps: Acetaminophen, Buspirone, magnesium, skin counter-warming (warm air to the face and hands while cooling the core, which is more effective than it sounds), then opioids and sedation, with neuromuscular blockade only as a last resort since it masks seizures and examination. |
| 5. Drugs, if refractory | The evidence is weak and mostly from case reports and small series. Bromocriptine (dopamine agonist), Propranolol, and Baclofen have the most published support, and a propranolol plus baclofen combination has been described as more effective than baclofen alone while also controlling heart rate. |
Paroxysmal Sympathetic Hyperactivity
Think in three separate jobs, because the drugs differ by job. Abort the episode happening now. Prevent the next one with background therapy. Remove the triggers that provoke them, which is the part most often skipped and the cheapest of the three.
| Goal | Agents and what they contribute |
|---|---|
| ABORT an episode | Morphine is the cornerstone and is often dramatically effective, acting both centrally and by blunting the afferent stimulus that provoked the surge. Intravenous benzodiazepines also terminate episodes and help the motor component. |
| PREVENT: beta-blockade | Propranolol is the preferred beta-blocker because it is lipophilic and crosses the blood-brain barrier, unlike the more cardioselective agents, and it also reduces the catabolism that makes these patients waste away. It has the best supporting evidence of the preventive agents. |
| PREVENT: gabapentin | Gabapentin has become increasingly favored, acting on neuronal excitability rather than blocking the end-organ response. It is particularly useful for the ongoing background tone between paroxysms and for allodynia, where beta-blockers do little. |
| PREVENT: alpha-2 agonists | Dexmedetomidine for the intubated or acutely storming patient and Clonidine for longer-term oral therapy. They reduce central sympathetic outflow, which is complementary to beta-blockade rather than duplicating it. Watch for bradycardia and hypotension. |
| PREVENT: the motor component | Baclofen, including intrathecal baclofen in refractory cases, targets the posturing and spasticity that drive contractures. Bromocriptine is used mainly for the temperature and dystonic components. |
| ⚠ REMOVE TRIGGERS the underrated half | Since episodes are provoked by stimulation, look for the provocation before adding a fifth drug. Treat pain properly. Decompress a distended bladder. Treat constipation and fecal loading. Cluster nursing care rather than disturbing the patient hourly, and reduce unnecessary noise and handling. This is genuinely therapeutic, not merely supportive. |
| Support the consequences | Feed them aggressively, since caloric requirements are enormous and weight loss is rapid. Replace the fluid and electrolytes lost through sweating. Involve physiotherapy early for contracture prevention, because the posturing is what produces the long-term disability. |
🎤 Rounds
Pimp Questions
- Why treat fever in a brain-injured patient more aggressively than in other patients? Fever independently worsens functional outcome and mortality after SAH, TBI and stroke. The injured brain has no metabolic reserve, so fever raises cerebral metabolic demand and intracranial pressure when perfusion is already marginal.
- Name the four features that predict central fever. Negative cultures, no infiltrate on chest radiograph, a diagnosis of SAH, IVH or tumor, and onset within 72 hours of admission. Together they give a probability of about 0.90.
- What is the practical use of that model? It gives you permission to stop antibiotics in a patient who is still febrile, which prevents open-ended antimicrobial courses in someone with no infection.
- Why does acetaminophen work poorly in central fever? Ordinary fever is prostaglandin-mediated and antipyretics lower the PGE2-driven set point. Central fever comes from direct disruption of hypothalamic thermoregulation, so there is little prostaglandin signal to interrupt.
- So what actually lowers the temperature? Physical cooling, because it works regardless of the set-point pathway.
- What is the main hazard of cooling, and how do you manage it? Shivering, which raises metabolic rate, oxygen consumption and ICP. Score it with the Bedside Shivering Assessment Scale, then acetaminophen, buspirone, magnesium and skin counter-warming, escalating to opioids and sedation, with paralysis last because it masks seizures.
- Distinguish central fever from paroxysmal sympathetic hyperactivity. Central fever is isolated hyperthermia. PSH is fever occurring simultaneously with tachycardia, hypertension, tachypnea, sweating and posturing, in discrete paroxysms.
- Give the six cardinal features scored in the PSH-AM. Heart rate, respiratory rate, systolic blood pressure, temperature, sweating and posturing.
- Which injury most commonly causes storming, and when does it appear? Severe TBI, especially diffuse axonal injury. It emerges days to weeks later, characteristically as sedation is weaned, which is why it is misread as withdrawal or agitation.
- What triggers an episode, and why does that matter twice? Stimulation: turning, suctioning, pain, bladder distension, constipation. It is the strongest diagnostic clue and also a treatment target, since removing triggers reduces episodes.
- How do you separate storming from a seizure? No EEG correlate. If there is genuine doubt, the EEG answers it.
- Which drug aborts an episode? Morphine, often dramatically, acting centrally and by blunting the afferent stimulus. Benzodiazepines also work.
- Why propranolol rather than metoprolol? Propranolol is lipophilic and crosses the blood-brain barrier, and it also reduces the catabolism that causes rapid weight loss.
- What does gabapentin add? It works on neuronal excitability rather than the end-organ response, so it helps the background tone between paroxysms and allodynia, where beta-blockers do little.
- What is the quality of the evidence for all of this? Low. There are no randomized trials; recommendations rest on case series and retrospective cohorts. Start one agent at a time and titrate to effect.
- What happens if storming goes untreated? Massive catabolism and weight loss, dehydration, ICP spikes, and contractures with heterotopic ossification from sustained posturing, plus longer stay and worse outcome.
📣 Sample Presentation
"Mr. R is a 24-year-old man on day 11 after a severe traumatic brain injury with diffuse axonal injury, and over the last 48 hours he has had recurrent episodes of temperature to 39.4, heart rate to 150, systolic pressures in the 190s, tachypnea, drenching sweating and extensor posturing. Each episode lasts about 20 minutes and resolves. The overnight team has cultured him three times and restarted broad-spectrum antibiotics twice. What makes me think this is paroxysmal sympathetic hyperactivity rather than sepsis is the pattern: it is paroxysmal rather than sustained, all six features rise together, and the nurses have noticed it reliably follows turning and suctioning. His cultures are negative, his chest radiograph is clear, and his EEG showed no correlate during a captured episode, which also excludes seizure. I have reviewed the chart for haloperidol given the overlap with neuroleptic malignant syndrome, and he has had none, and the timing does fit sedation weaning so I have considered withdrawal, which may be contributing. My plan is morphine to abort episodes, propranolol as background therapy because it crosses the blood-brain barrier and will also reduce his catabolism, and I would add gabapentin for the ongoing tone between episodes. Just as importantly, I want the triggers addressed: his bladder scan shows retention, he has not had a bowel movement in four days, and we should cluster his cares rather than disturbing him hourly. He has lost 6 kg, so I have asked nutrition to increase his caloric target, and physiotherapy is seeing him today for contracture prevention."
Ward Checklist
- ⚠ Infection excluded properly -cultures, chest radiograph, urine, and CSF if there is an EVD, shunt or basal skull fracture.
- VTE considered, with a low threshold for duplex ultrasound in an immobile patient.
- Drug chart reviewed for phenytoin, antibiotics and antipsychotics.
- The four Hocker features applied before continuing antibiotics in a persistently febrile patient.
- Normothermia targeted actively, with a formal shivering score if cooling.
- Episodes documented as paroxysmal or sustained, and whether all features rise together, since that is what separates storming from sepsis.
- Triggers hunted -bladder, bowels, pain, handling- before another drug is added.
- Calories and fluids escalated to match the catabolic load, with weight tracked.
- Physiotherapy involved early for contracture and heterotopic ossification prevention.
📋 Summary
At a Glance
| Point | Detail |
|---|---|
| Why it matters | Fever independently worsens outcome after SAH, TBI and stroke. The injured brain has no metabolic reserve, so normothermia is a treatment goal, not a comfort measure. |
| ⚠ Both are diagnoses of exclusion | Exclude ventriculitis (any EVD or shunt), VAP, UTI, line infection, C. difficile, sinusitis from an NG tube, VTE, drug fever (phenytoin), withdrawal, seizure, and the hyperthermia syndromes (NMS, serotonin syndrome, malignant hyperthermia). |
| Central fever: the four features | Negative cultures + no infiltrate on chest radiograph + SAH, IVH or tumor + onset within 72 hours predict central fever with probability 0.90. Its practical use is permission to stop antibiotics in a still-febrile patient. |
| How it looks | Often above 39°C, a plateau with loss of diurnal variation, and poor response to antipyretics. Highest risk after SAH with intraventricular extension. |
| ⚠ Why antipyretics fail | Ordinary fever is prostaglandin-mediated and antipyretics lower the PGE2-driven set point. Central fever is direct disruption of hypothalamic thermoregulation, so there is little prostaglandin signal to block. A fever that ignores acetaminophen is a clue, not a dosing problem. |
| Cooling, and its hazard | Physical cooling is the mainstay because it bypasses the set-point pathway. ⚠ Shivering undoes it, raising metabolic rate, oxygen use and ICP. Score with the Bedside Shivering Assessment Scale; acetaminophen, buspirone, magnesium, skin counter-warming, then sedation, with paralysis last since it masks seizures. |
| ⚠ Central fever vs storming | Central fever is isolated hyperthermia. PSH is fever PLUS simultaneous tachycardia, hypertension, tachypnea, sweating and posturing, in discrete paroxysms rather than a plateau. |
| PSH-AM | Clinical Feature Scale grades severity of six features (HR, RR, SBP, temperature, sweating, posturing); Diagnosis Likelihood Tool grades context. Combined, they estimate the probability of PSH. |
| Who and when | Most often severe TBI with diffuse axonal injury, also hypoxic-ischemic injury. Appears days to weeks later, typically as sedation is weaned, which is why it is misread as withdrawal or agitation. |
| ⚠ Triggered by stimulation | Turning, suctioning, pain, bladder distension, constipation. This is the strongest diagnostic clue and a treatment target. |
| Mimics | Sepsis (but storming is paroxysmal and provoked by handling), seizure (no EEG correlate), NMS (check for antipsychotics), withdrawal (same timing, and they coexist), PE, serotonin syndrome, thyroid storm. |
| Treatment: three jobs | ABORT with morphine (cornerstone) or benzodiazepines. PREVENT with propranolol (lipophilic, crosses the blood-brain barrier, reduces catabolism), gabapentin (background tone), clonidine or dexmedetomidine, baclofen for posturing. REMOVE TRIGGERS -bladder, bowels, pain, cluster care. |
| ⚠ The evidence | No randomized trials exist for either condition. Everything rests on case series and retrospective cohorts, so add one agent at a time and titrate to an observed effect. |
| Cost of missing it | Massive catabolism and weight loss, dehydration, ICP spikes, contractures and heterotopic ossification, longer stay, worse outcome. Feed aggressively and involve physiotherapy early. |
The Three Things to Remember
- Exclude infection before you blame the brain, and exclude the brain before you commit to three weeks of antibiotics. The error runs in both directions, and the Hocker features tell you which way to go.
- A fever that ignores acetaminophen is telling you something. Central fever is not prostaglandin-mediated, so cool it physically rather than escalating the dose.
- If the temperature rises alone it is central; if everything rises together in paroxysms it is storming. And before adding a fifth drug for storming, check the bladder and the bowels.
