Category: Cases

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We explore the world of thrombectomy for acute ischemic stroke with Justin F. Fraser (@doctorjfred), MD, FAANS, FAHA, Professor and Vice-Chair of Neurological Surgery and Director of Cerebrovascular Surgery and Neuro-interventional Radiology at University of Kentucky, where he specializes in cerebrovascular, endovascular, skull base, and endoscopic transsphenoidal surgery.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Takeaway lessons

1. In the opinion of Dr. Fraser, thrombectomy for qualifying patients with acute ischemic stroke is the current standard of care. Patients in non-thrombectomy centers should be transferred. Failure to do so is potentially negligent.
2. Dr. Fraser feels there are few true contraindications to thrombectomy (as long as the patient’s goals are concordant), and the current indications should probably be most strokes <24 hours with a large vessel occlusion on CTA – i.e. ICA (including with a tandem extracranial carotid occlusion), MCA, ACA, or basilar. He no longer feels most cases need perfusion imaging as even large or older infarcts seem to benefit.
3. The main current question is the utility of thrombectomy in “medium vessel occlusions,” such as M2 and more distal vessels.
4. Radial artery access is growing in popularity, similar to its growth in cardiovascular interventions, now that devices have shrunk enough to fit. The right wrist is preferred.
5. In general, qualifying patients should still receive systemic thrombolytics as soon as possible prior to performing thrombectomy, at least with the state of the evidence in 2025. This also helps manage any particles that embolize into more distal vessels during aspiration of a larger thrombus.
6. Generally, thrombectomy is merely a process of aspirating an embolus. However, if thrombosis also involved an intracranial atherosclerotic narrowing, there may still be unstable stenosis afterwards, so about a third of cases also require stent placement. (Carotid occlusions are a different story and usually need stenting, just as with elective endarterectomies.) When stents are placed for this reason or for a carotid lesion, dual antiplatelet inhibition is usually needed; this may be started during the procedure with an intra-arterial agent if DAPT is not already on board.
7. Thrombectomy can be performed under local anesthesia only, or under deeper sedation; the practice for Dr. Fraser’s group is to put everybody under general anesthesia. Anesthesia’s efforts are performed simultaneously to the interventional prep and should not delay it.
8. Post-procedure blood pressure targets are controversial. Fraser targets SBP <160 for 24 hours to limit reperfusion hemorrhage.
9. Post-procedure MRI is usually appropriate to delineate infarct size and to appreciate the degree of edema, potentially requiring decompressive craniectomy (large hemispheric or cerebellar stroke). If MRI is delayed, CT is appropriate, perhaps dual-energy CT to differentiate hemorrhage from contrast staining.
10. Expanding thrombectomy to more patients in smaller hospitals requires more trained neurointerventionalists, but this is not a completely simple matter, as it must be balanced against adequate volume to maintain proficiency for the proceduralists and their teams. Smaller centers also need a link to a larger center that can support them for scheduling gaps and complications.

Resources

Society of Neurointerventional Surgery

Get Ahead of Stroke

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We learn about the basics of fetal monitoring in the critically ill pregnant woman and how to integrate them into our ICU workflows, with Stephanie Martin, MFM obstetrician and host of the Critical Care Obstetrics podcast and teacher at the Critical Care Obstretrics Academy.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Takeaway lessons

1. A fetus is considered potentially viable at 23-24 weeks gestational age, with 22-23 weeks being occasionally viable in specific circumstances and highly specialized centers. “Potentially viable” does not mean guaranteed survival, as fetal mortality is still quite high. In other words, at 23 weeks and above, intervention to promote fetal survival make sense. Every additional day of gestation improves outcomes.
2. A conversation should occur preemptively between the mother, ICU, and obstetric teams to clarify what options will be considered—in some circumstances, early delivery (via C-section) is not desired due to the risk to the mother, and should not be assumed to be the contingency in all viable pregnancies. On the flip side, delivery of a non-viable fetus could still be appropriate for the mother’s health, such as in uterine infection or hemorrhage.
3. If a fetus will not be delivered early, there may be no role for fetal monitoring.
4. Fetal monitoring is therefore relevant at viable gestational ages. However, it is also more difficult for early pregnancies; the monitors can easily wander off a tiny fetus, and the strips are harder to interpret.
5. Fetal monitors essentially monitor 1. Fetal heartrate (via Doppler), and 2. Uterine contraction. Heartrate is monitored primarily to determine variability, i.e. how much the rate changes from its average baseline in response to stimulus, particularly uterine contraction (which causes fetal stress of sorts). Poor variability with markers like late decelerations can be a sign of fetal acidosis and ischemia, particularly to the brain, which can increase the risk of fetal demise or birth defects such as cerebral palsy. Prematurity creates particular vulnerability to this.
6. Maternal sedation leads to fetal sedation, which can make interpreting the heart rate more difficult.
7. Uterine contractions rarely turn into labor, but they provide a natural stress test to the fetus.
8. Much of the interpretation of “fetal distress” comes down to the context—for instance, maternal acidemia will always cause fetal acidemia, but in a rapidly reversible setting such as DKA, the best solution may simply be resuscitating the mother.
9. Fetal distress is often an early marker of shock and other systemic stress, as uterine perfusion is sacrificed fairly early by the body in favor of other organs. This often manifests as uterine contractions.
10. Any pregnant woman with a gravid uterus up to the umbilicus, or >20 weeks, who is critically ill, should not lie supine; the uterus will compress the great vessels and may cause shock. Elevate the head of the bed or tilt them laterally at all times. (During CPR, assign someone to manually displace the uterus to the left, as tilting the entire patient is challenging.)
11. There is relatively little role for ultrasound or other tools for fetal monitoring; the gold standard is fetal heart rate monitoring.
12. Paroxysms of vital sign changes (tachycardia, hypertension, etc) in a pregnant woman could be a subtle marker of contractions.
13. With regards to ionizing radiation, generally, do whatever test you would do in a non-pregnant woman. Birth defects are generally established by the end of the first trimester, so in a viable pregnancy, it should not be a concern at all. While appropriate attention should be paid to avoiding needless radiation, if an important diagnosis needs to be made, do the x-ray or CT scan (or even fluoroscopy, likely the highest risk).
14. In the post-partum patient, the longer it’s been since birth, the less likely a maternal illness is pregnancy-related. In the first week, assume it’s pregnancy related. In the first six weeks, consider it, especially hypertension complications. Cardiac problems (e.g. peripartum cardiomyopathy) can occur even later, especially as the diagnosis may be delayed. A common presentation is post-partum “asthma,” actually pulmonary edema, as the fluid bolus of delivery overloads a cardiomyopathic heart. The most hypercoagulable period in pregnancy is actually the first six weeks post-partum, so VTE is an important concern.

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We discuss the basics of EEG in the ICU, including when to do it, selecting the appropriate study, and the basics of bedside interpretation, with Carolina B Maciel, MD, MSCR, FAAN, triple boarded in neurology, neurocritical care, and critical care EEG.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Takeaway lessons

1. There is little to no role for a very short (<2 hour) EEG in the critically ill patient, who generally has “less of everything”; to determine the presence of seizure activity or other electrical disease, more data is usually necessary.
2. Long-term or continuous EEG is usually defined as >12 hours. 2-12 is a middle ground (both clinically and for billing purposes). In most ICU cases, a “middle” study of a few hours can be done, then the findings used to inform the need for a longer study; validated scores exist for this, such as 2HELPS2B.
3. Don’t forget the non-seizure diagnoses that can be made/supported from EEG, such as brain death, cefepime-induced encephalopathy, sudden clinical changes due to osmotic shifts, etc. In reality, EEG readers, particularly in the community, may or may not be making great efforts to appreciate these things. You will get better reads if you communicate your questions to the reader, and consulting neurologists/neurointensivists may be able to glean more from a non-specific EEG report as well. Critical care EEG folks like Carolina may be the most helpful, but there are very few training programs for this.
4. Basic filters on the EEG include the high and low pass filters (should be LFF of 1 hz, HFF ~7–8 hz), and potentially a notch filter for 60 hz (in the US) or 50 hz (in Europe) to filter out AC electrical noise.
5. Dark vertical lines on the strip occur every 1 second. With normal scale there should be about 3 centimeters (around your thumb’s length) between them.
6. Odd numbered leads are on the left side of the head. Even numbers are on the right. Z-numbered leads are in the sagittal midline.
7. Do you see intermittent bursts of something pointy, like it will hurt to sit on? These may be muscular artifact, which can be hard to distinguish (look at the patient to see if they’re moving/twitching), but if not, this may be an epileptic discharge; similar to a PVC, or someone coughing in the symphony audience, it’s an inappropriate interruption in brain activity. This may be focal or global (all leads), and focal may be higher risk. They may be repetitive, occurring somewhat regular intervals, which are also more concerning. Ultimately, the concern is always whether they are going to evolve/organize into full seizures, so if no evolution ever occurs, that is also more reassuring.
8. When to treat epileptogenic discharges on EEG is always a judgment call and must be put in context of the patient. More abundant discharges with a more malignant appearance are more concerning, but the clinical correlation matters too; EEG findings with no clinical correlate are less worrisome. Convulsive seizures are a medical emergency (especially with continuous tonicity), but non-convulsive electrical activity, even non-convulsive status, usually has room and time to weigh the risks versus the benefits of therapy. Talk to experts and make a thoughtful decision.
9. Carolina hates fosphenytoin due to the cardiotoxic effects. Lacosamide is quite benign.
10. The ictal-interictal spectrum is an electrical finding (not meeting the arbitrary definition of unequivocal status epilepticus) that may be important or not; you must consider the patient. If there is a clinical alteration in mental status that may be due to the activity, you should challenge them with a loading dose of a fast-acting, minimally sedating anti-seizure medicine and see if it helps their EEG and clinical status. Carolina dislikes benzos for this (sedating), thinks levetiracetam is only okay (too slow to reach peak brain concentration, ~90 minutes); brivaracetam (5 mins) and valproic acid (5-10 mins) are good.

References

Training modules for ACNS 2021 ICU EEG terminology

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

A roundup from members of the SCCM’s ICU Liberation committee, recorded at SCCM Congress 2025.

Included:

• Heidi Engel
• Kali Dayton
• Kristina Betters
• Stacey Williams
• Jessica Anderson
• Jenna Domann
• Sergio Zanotti
• Erika Setliff
• Brian Peach

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We discuss the basics of evaluation for tracheostomy placement, periprocedural care, and post-procedure complications with Vinciya Pandian, PhD, ACNP, FCCM, tracheostomy nurse practitioner and researcher.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We discuss assessment, monitoring, medical stabilization, and when to consider transplant of the patient with acute liver failure. We are joined by Dr. Sergio Navarrete, anesthesiologist and intensivist with fellowship training in transplant anesthesia.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Takeaway lessons

1. Transaminases rising into the many hundreds or thousands (especially with pre-existing liver disease), or a MELD in the low teens (from baseline normal) should raise concern for a concerning degree of liver injury, usually due to shock liver, congestion, or infection. This should also prompt consideration for transplant evaluation, and usually a phone call to your transplant center.
2. Reversible causes, such as acetaminophen toxicity or portal vein thrombosis, must be ruled out.
3. Optimization of perfusion should include not only the left-sided systemic circulation, but also the right-sided system and venous congestion; congestive hepatopathy (from volume overload or RV failure) can absolutely cause severe liver injury. Echo, potentially with tools like VEXUS scoring, can be a great help here.
4. N-acetylcysteine has a clear indication for treating acetaminophen poisoning, but not much data for other causes of liver failure. However, many clinicians believe it may provide some benefit, and there is probably no harm—other than administering a fair amount of volume.
5. Hypoglycemia and hypothermia are both relatively late and ominous findings in the ALF patient (put them on a dextrose infusion and hourly glucose checks). Transaminase levels reflect hepatocyte injury but not liver function. Synthetic function as measured by INR or fibrinogen are helpful. Bilirubin is usually too slow and non-specific to be actionable. Trend this stuff every 6 hours or so.
6. Mental status is a key monitoring tool as a marker of cerebral edema. The clinical exam, ammonia level, potentially serial CT scans, and maybe invasive ICP monitoring (Sergio prefers a bolt over EVD) may all be needed in high-risk cases.
7. The highest risk patients for cerebral edema are those with truly acute/hyperacute liver failure. Trend ammonia, which has some correlation with herniation risk, but the neuro exam is more useful. Neurosonography could be used as well.
8. Lactulose should be used, and in extremis hyperosmolar therapy considered, although data for this is less clear than in other neurologic emergencies.
9. Liver ischemia and death will reliably cause a systemic inflammatory state with resulting distributive shock; this can persist even after transplant, due to persistent elements of the dying liver. Treat this like any SIRS/distributive shock state.
10. Bleeding and clotting can both occur; numbers usually suggest coagulopathy, but hemostatic rebalancing is often present, at least until something perturbs the balance (e.g. a procedure). Labs like the INR are a marker of disease severity, not bleeding risk. Fibrinogen is a little better, but TEG is probably the most useful marker of bleeding status, as many of these people are actually hypercoagulable.
11. Some would use CRRT relatively early in a liver failure patient; Sergio would not. However, he would consider it in the volume overloaded patient to manage congestion (if diuresis proved inadequate).
12. Liver-specific extracorporeal organ support using various devices (MARS, “liver dialysis,” albumin dialysis, etc) are interesting/promising therapies that largely have not shown convincing benefit in studies. They tend to be sporadically available and highly institution-specific.
13. In all cases, earlier consultation to liver transplant specialists is better than later (this may involve an interfacility phone call or transfer). Several days are usually needed for transplant evaluation, many aspects of which are not directly medical, such as assessment of social support, insurance, pre-transplant workup, etc. Waiting too long may mean a patient dies before the process can be completed.
14. All truly acute liver failure should be referred for transplant evaluation.
15. Typical rule-outs for transplant include uncontrolled metastatic malignancy, age (often >75; every center has a different cutoff), and severe unrepairable cardiac dysfunction. Infections such as active bacteremia are a concern. Much of this is a judgment call and up to the transplant team, and their culture and policies.
16. Alcohol use is not necessarily a rule-out for transplant; some (not all) centers will consider these patients. The social milieu is more important. It is not unreasonable to refer a patient to a more distant center that has broader eligibility criteria than a nearer one that rules them out.
17. Some critically ill patients may be transplant candidates, particularly if most of their problems are deemed secondary to their liver failure and hence potentially reversible. Liver transplant is a procedure that can and often should be performed in the setting of multi-organ failure, shock, respiratory failure, etc. But each center has its own risk tolerance.

References

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We talk about the nitty-gritty details of a well-run cardiac arrest, with Scott Weingart of Emcrit (@emcrit), ED intensivist.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Takeaway lessons

1. In any sudden loss of pulse/consciousness, particularly in a known cardiac patient, the presumption should be for a shockable arrhythmia and rapid defibrillation should be prioritized above all else.
2. Anterior-posterior pad placement may or may not be superior, but tends to be logistically helpful, as it allows rolling the patient a single time then never again; a second set of pads can be added for double sequential defibrillation without moving them, and a mechanical compression device can be applied at the same time as the pads.
3. The primary or highest-trained provider should not be the sole “code runner,” but ideally offering high-level leadership, thinking about reversible causes and judgment calls, and performing procedures, while another leader (often a nurse) runs the standard activities of ACLS such as timing, coordinating rhythm checks, assigning jobs, quality assurance, and directing the room. That frees your cognitive bandwidth by handling all your logistics, and they can act as the one-stop-shop for passing needs and issues up and down the chain.
4. IOs are probably the go-to for immediate access, if no IVs are present. But Scott likes to always place central access, usually femoral. He does ECPR, so the access may be needed, and even if not, it maintains the skill for next time. He also likes an arterial line, so it’s easy to place venous alongside it. He would generally not place it fully sterile (gowns, drapes, etc), but will use sterile gloves and prep the skin, assuming that any femoral line placed in the ED is going to be replaced within 24 hours.
5. Scott loves an arterial line. It eliminates the “pulse check,” allowing instant confirmation of pulsatility, while also allowing a very sophisticated assessment of coronary perfusion.
6. A diastolic BP above 35–40 mmHg, measured from the arterial line during cardiac arrest, suggests adequate coronary perfusion. This must be measured manually, as the automated number will falsely measure the wrong spot in the waveform during the “suction” of chest recoil (see link below); the true point of measurement is just before the upstroke of systole begins. If you’re above this DBP, just skip epinephrine, which will probably merely be toxic (ie promoting arrhythmias).
7. A low DBP should be used as a general marker of poor perfusion, and prompt other changes. Try modifying the point of compressions on the chest to avoid obstructing the LVOT (TEE is even better for this, but not available most places). Swap out compressors to ensure the most vigorous compressions, even if they still “look okay” or claim to be. Look for a reversible cause, such as hemorrhage or obstruction. Finally, if it’s truly just vasoplegia, consider other moves, such as adding vasopressin/steroids (an evidence-based practice) or high-dose epinephrine (5 mg epinephrine).
8. ETCO2 should be used in all arrests, to confirm airways, prognosticate, and provide a marker of perfusion much like the arterial DBP.
9. Scott thinks we should stick to 30:20 mask ventilation when an airway is not in place; breaths don’t really go in during compressions, and bagging during the upstroke is very tough. But he prefers to just insert a supraglottic airway quickly and use that, a skill anyone attending cardiac arrests should have. If using the BVM, you should use capnography to confirm breaths are actually going in.
10. Intubation should be done with video if available. Hyperangulated or regular geometry are both fine. Use a bougie if you have regular geometry (and are good with it). There should never be an intentional pause in compressions for the airway, however; just intubate during compressions, not so hard with video and a bougie. Position the patient optimally, just as in any situation; don’t rush.
11. Never perform “pulse checks,” only rhythm checks. If the rhythm is non-perfusing, resume compressions. If it’s organized and potentially perfusing, only then check for a pulse (or preferably your arterial line).
12. POCUS is essential: look for pericardial effusions, a dilated RV (although this is usually present), signs of hemorrhage, and pneumothorax. Maybe even more importantly, use it for pulse checks rather than your fingers. Scott will start with this, and if a “sonographic pulse” (visual pulsatility of the vessel) is found, he’ll then apply his fingers to see if it’s strong enough to feel. At this moment in time, he thinks palpability is a reasonable cutoff for when to call flow PEA vs hypotension.
13. Once he’s ruled out reversible causes, he tries not to look at the heart with ultrasound, since it tends to detract from compressions (without TEE); sonographic pulse implies organized cardiac activity. An arterial line obviates all of this, although it’s not clear what BP is adequate; Scott still uses a DBP 35-40 but would accept a MAP of 40 as a reason to defer compressions, if rapid efforts are undertaken to increase it.
14. Scott always likes mechanical compression devices when available (he likes the Lucas), which ensures good quality, provides a backboard, and reduces the energy in the room, even if it doesn’t improve outcomes. Buy one for your hospital’s code team and bring it to the arrests. If not available, he likes a backboard.
15. The impedence threshold device (ITD), potentially in combination with active compression-decompression devices, is interesting but the initial promising data has not been replicated; he would not consider this ready for use.
16. Heads-up CPR is also interesting but not yet proven.
17. When defibrillating, always max out the current on the machine. It creates no meaningful injury and maximizes your chance of conversion.
18. When a shockable rhythm is seen, he resumes compressions while charging, and in fact often performs hands-on defibrillation (shocking during compressions, using some kind of standoff between hands and chest, such as a towel, or even just gloves); mechanical compressions make this easiest.
19. Pre-charging before the rhythm check is wise, and the nurse code leader can coordinate this; do it every time.
20. Amiodarone or lidocaine are equally reasonable first line antiarrhythmics. If they’ve had one and you’re still in electrical storm, try the other.
21. If storm persists, these are excellent ECPR cases. Otherwise, you can try esmolol (bolus 500 mcg, usually no drip), then double sequential defibrillation.
22. DSD: don’t let pads touch; shock as simultaneously as possible (used to be intentionally separated, but some data now suggests closer together is better). There is a very small but real risk of damaging defibrillators doing this (and the damage may actually not be obvious, i.e. the machine will still pass a later self-check).
23. How long to go? Depends on baseline functional status, rhythms seen, and other factors. Past 40 minutes of low-flow time, arrest is probably not survivable without ECPR (for which the cutoff is probably 90 minutes), unless there has been stuttering flow (intermittent ROSC), which tends to reset that timer. ETCO2 persistently <10 is very poor. No cardiac motion seen on echo in PEA is poor.
24. One exception might be if your interventional cardiologists are willing to cath intra-arrest during mechanical compressions; in that case you might go longer to bridge to this.
25. In a young or baseline well patient, Scott would almost never stop before 40–45 minutes.
26. Scott always runs a norepinephrine drip at 50 mcg during the arrest, making it easy to transition to the drip after ROSC and avoiding any delays.
27. After ROSC, a STEMI or high-risk patient should go to the cath lab. Everyone else should have a pan-CT, including head, chest, abdomen/pelvis. This ideally is gated/timed to triple rule out PE, coronary occlusion, and aortic dissection. It also identifies important post-CPR trauma.
28. In 2025, Scott’s take on TTM is reactive: place an invasive temp probe (esophageal or Foley, not rectal, which is too slow and inaccurate) and monitor for fever; if it occurs, then cool actively to normothermia. There are probably some patients who benefit from more cooling, but nobody knows who (longer downtimes?). This method is as good and cheaper than empirically applying a cooling device to maintain normothermia before fever occurs, and that might cause problems, such as if it aggressively cools for a trivially increased temperature and induces shivering. Put the cooling device at the bedside without opening the pads if you really want to be ready.
29. If having trouble inserting your floppy esophageal temp probe, use an esophageal stethoscope from the OR, or use a split 8.0 ET tube to introduce a lubricated probe into the esophagus.
30. For out-of-unit codes, it’s all the more important to have a nurse code leader. A code team should bring the specialized equipment (maybe make a cart) to the bedside, such as mechanical compression devices, ultrasound, capnography, etc.

Resources

1. PO Berve at Emcrit on correctly measuring arterial line diastolic BP during cardiac arrest
2. Emcrit on ultrasound in cardiac arrest
3. Cardiopulmonary Resuscitation Quality: Improving Cardiac Resuscitation Outcomes Both Inside and Outside the Hospital: A Consensus Statement From the American Heart Association

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

Our approach to common problems and troubleshooting:

• Difficulty feeding guidewires
• No flash on arterial lines
• Pneumothorax during subclavian lines
• Difficulty inserting ET tubes during hyperangulated laryngoscopy
• No response to vasopressors
• High gastric residuals during tube feeds

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We talk about diagnosis, treatment, and subsequent care of the patient with bacterial meningitis, with Emory neurointensivist Casey Albin, MD (@caseyalbin).

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!

Takeaway lessons

1. Many septic patients have altered mental status, but suspicion should be raised for CNS infection when there is also: headache, photophobia, vomiting, or any possibility of seizure activity.
2. Meningitis and encephalitis are separate entities usually involving different organisms, different imaging findings, and with different prognostic implications and downstream complications. However, at the early diagnostic stage, they can be largely lumped together.
3. Empiric antimicrobials must consider CNS penetration. Piperacillin/tazobactam (ie Zosyn) has very little. Ceftriaxone is better. Cefepime is fine, although the prospect of cefepime neurotoxicity may make neurologists leery; ceftazidime is fine too. Add vancomycin (not necessarily for MRSA but for resistant Strep pneumo), acyclovir (for HSV), and a liberal approach to adding ampicillin for Listenia for anybody older, immunocompromised, or in the midst of an outbreak.
4. Dexamethasone has been shown to reduce hearing loss after Strep pneumo meningitis. If suspicion for meningitis is strong early, it’s reasonable to give early (before or concurrent with antibiotics). It’s probably not worth giving >24 hours later.
5. The main benefit of lumbar puncture is to allow stopping or narrowing antimicrobials without treating with the entire empiric cocktail for a full two weeks. (There is also the chance of identifying a resistance organism.)
6. Ideally, LP is done before antimicrobials. However, if non-culture-based diagnostics are available such as PCR panels, successful diagnosis can often occur even after antibiotic administration. It’s worth doing the LP even if late and no PCR is available, as the signature of protein, glucose, etc will often still be useful. (At least, up front in a patient who might have CNS infection, avoid creating new obstacles like loading them with anticoagulation, antiplatelets, low molecular weight heparin, etc.)
7. Most patients will already have a CT head performed before LP is considered, making the question of whether this is necessary (to assess risk of downward herniation) fairly moot. However, if not, it should probably be done prior to LP in anyone with an altered level of consciousness.
8. Order from all CSF: Gram stain and culture, cell counts (first and last tubes), glucose, protein, and HSV PCR. (VZV generally does not cause clinical meningitis per se, usually causing a meningitis vasculitis, e.g. in someone with small-vessel strokes.) If available, order PCR arrays too, although some centers may not run it unless the CSF WBC count is elevated (e.g. >5). In a patient with any immunocompromise, test for cryptococcus as well. Other immunosuppressed testing is case-specific.
9. Always measure opening pressure. This is not accurate in a patient sitting up. While technically possible to puncture a patient sitting up, then rotate them with assistance to lay flat, it’s not easy or elegant. In a sick patient, just do the LP laying down.
10. Remember that opening pressure is measured at the bedside in centimeters of water, but should be converted to millimeters of mercury to be clinically applicable.
11. Draw at least 20 cc of CSF in all cases. If opening pressure is high (and CT not concerning), fill the four tubes (~36 cc) and measure the closing pressure. Few patients are harmed by draining <40 cc. Draining >40-50 can create some risk for herniation or hemorrhage (eg small subdural hemorrhage) and should not be done thoughtlessly.
12. Meaningfully elevated CSF protein should not just be “high,” but should exceed the patient’s age.
13. Any meningitis patient with an altered mental status should at least have a spot EEG, and possibly long-term EEG depending on the findings.
14. Any meningitis patient with a high opening pressure on LP, who is sufficiently obtunded to be intubated, should be considered for invasive ICP monitoring (e.g. EVD), if available. Otherwise, close monitoring for ICP crisis with neuro and pupil checks and serial CT scans.
15. Treating high ICP in meningitis with EVD or lumbar drain is often appropriate.
16. Any neurologic deterioration after antibiotics and other initial care is very likely either seizure or ICP crisis. These are fixable patients; diagnose and treat these complications aggressively.
17. Transcranial doppler may be a useful non-invasive screen for elevated ICP, by revealing a high-resistance waveform (high pulsatility index) as ICP increases.

Resources

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | TuneIn | RSS

We discuss the practical barriers to implementing the A-F ICU liberation bundle, with Kali Dayton, ACNP-BC (@daytonicu), host of the Walking Home from the ICU podcast, and consultant to ICUs working on these issues.

Learn more at the Intensive Care Academy!

Find us on Patreon here!

Buy your merch here!