We discuss the medications typically used after organ transplant, their impact on critical illness, and how to manage them when these patients show up sick—with Olivia Philippart, transplant clinical pharmacist specializing in liver and kidney transplant at University of Kentucky HealthCare.
Most kidney transplants will end up on a calcineurin inhibitor like tacrolimus (or the older cyclosporine), an anti-proliferative like mycophenolate mofetil (Cellcept) or the older azathioprine, and possibly corticosteroids (e.g. prednisone). Formulations for some of these may need to be adjusted based on your formulary, so consult your pharmacist to get the equipotent dose.
How these patients present, their degree of immunosuppression, and risk of rejection, are all heavily dependent on the time since transplant. A patient <6 weeks from transplant is high risk for nosocomial infections (e.g. post-op complications). A patient years out is mainly at risk of the same infections as anybody else, in addition to opportunistic infections related to their immunosuppression.
Latent viral infections unmasked by immunosuppression or acquired from the transplant are usually not a surprise, as these are tested for as part of the initial workup.
The highest risk of organ rejection and hence the highest degree of immunosuppression is in organs with substantial amounts of lymphoid tissue transplanted. The highest is small bowel, then lung, then heart/kidney/pancreas, then the least in liver (liver transplant can actually overall support immune function). Some livers can be maintained on monotherapy, while lungs usually need triple therapy, and often dual therapy is used in the middle category.
Durations of therapy for identified infections may be longer in the immunosuppressed than for routine ICU care.
Mycophenolate is the first agent to consider dose reducing or holding in the setting of active bacterial infection. How to handle this depends on the severity of infection and degree of concern for rejection.
Both our calcineurin inhibitors (tacrolimus and cyclosporine) are primarily cleared in the liver and gut, so when there is liver impairment or bowel problems, dose decreases are often needed. Dietary intake also reduces drug absorption whereas NPO status may increase it. These drugs are heavily protein bound so albumin fluctuations (e.g. from malnutrition) may impact free levels.
Drug interactions are common as well; CYP3A4 or PGP inhibitors like diltiazem or verapamil, azole antifungals, amiodarone, macrolides (although not azithromycin), and paxlovid will tend to increase levels, while inducers like phenytoin or phenobarbital will tend to decrease them.
Overall, the therapeutic index of the calcineurin inhibitors is small, so have a low threshold for checking trough levels early and often.
After holding a dose, the serum levels will normalize within 3-5 half-lifes, but full return of immune function may take several weeks. However, the baseline level of immunosuppression is usually not so profound that the difference between “off” and “on” is huge and binary.
Organ rejection is possible but rare when drugs are acutely held (for days, maybe a week or two) in setting of severe infection, as this is already a relatively immunosuppressed state. However, this depends heavily on the time from transplant, and the organ transplanted.
Mycophenolate levels (or mercaptopurine levels for the older azathioprine) tend not to fluctuate as much; the metabolism (via glucuronidation) is not as sensitive to hepatic function, so monitoring levels is rarely needed.
Most of our immunosuppressants are not significantly renally cleared, so renal injury (even dialysis) usually require no dose adjustment. However, they can be nephrotoxic, so high levels may CAUSE renal injury, not vice versa.
Tacrolimus is available in either immediate release capsule (taken twice daily) or a long-acting form (taken once daily). The latter helps to decrease peaks and some of the neurotoxicity (seizure, tremors), but cannot be opened. There is an 80% conversion between formulations (multiply the long-acting dose by 1.2, then divide by half to get the short-acting BID equivalent). Levels checked should always be troughs.
Short-acting tacrolimus capsules should not be opened and put down tubes, but can be opened and given sublingually (50% dose reduction)—just dribbled under the tongue—although nurses need to take special precautions like gowning and double gloving. There is also a liquid tacrolimus formulation available.
IV tacro exists, but has substantially higher nephrotoxicity, and the dose conversion is tricky; other routes are preferred.
Cyclosporine is available in suspension which can go down a feeding tube, or via IV form (dose reduction needed).
IV mycophenolate is available (1:1 conversion), as well as a liquid suspension.
Steroids can be used in the ICU as usual (e.g. stress dosing), and indeed temporarily converting transplant patients to a pure steroid regimen is a reasonable approach during critical illness (remember: 20 mg hydrocortisone is equivalent to 5 mg prednisone).
It’s generally sound to touch base with someone who knows a patient’s transplant history, even years out (often just their normal nephrologist, pulmonologist, etc in that case, not necessarily the original transplant team), when these patients are admitted for critical illness.
Calcineurin inhibitors can cause headaches, seizures, even PRES, hyperkalemia and hypomagnesemia, and hypertension, hypercholesterolemia, hyperglycemia/diabetes. Attributing these effects to the drug is usually a diagnosis of exclusion.
We discuss assessing patients prior to intubation or other airway management, including both elective and emergent circumstances, with Dr. Jed Wolpaw, anesthesiologist and intensivist from Johns Hopkins, anesthesiology residency program director, and host of the ACCRAC podcast.
Edentulous (toothless) patients are usually easier to intubate, but harder to mask ventilate. Heavy beards are harder to mask (can you trim it, or cover it with a Tegaderm?), larger neck circumferences, and larger tongues likewise.
Consider the history, particularly involving the head and neck anatomy. Is there surgical history here? Jaw or oral surgery? Prior trachs or oral/neck radiation? Rheumatoid arthritis or Down syndrome (which can cause atlanto-occipital instability and may warrant trying to limit any forced neck extension)?
Start by looking into the patient’s mouth (mouth open, sitting up, no “aah”):
Mallampati score (do you see the entire uvula, part of it, soft palate, or hard palate only?)
How is the dentition? Remove dentures if present. Are there loose teeth?
Is there an excess of soft tissue in the mouth (large tongue, etc)?
Evaluate the thyromental distance (thyroid bump to chin); <3 cm (or fingerwidths) suggests a more “anterior” airway.
Evaluate neck flexion and extension (passively if necessary) to appreciate limitations in neck mobility.
If the patient is able, evaluate how well the jaw can protrude/prognath: ability to bite more of the upper lip with the lower teeth is a good thing. This is probably the single most predictive test for airway difficult, although it usually requires patient cooperation.
Review the chart (or ask the patient) for prior documentation of intubation or anesthesia to determine if they have a history of a difficult airway. This can require some interpretation of the context and who was intubating previously. Good practice when documenting: write exactly what you did, and if it was difficult, write why! If you used a technique like awake intubation, a bougie, etc for elective or training reasons, document that reason so they don’t earn a label of a difficult airway forever.
The STOP-BANG score is used to predict post-anesthesia airway obstruction (i.e. OSA), and probably has some association with faster deoxygenation and difficult mask ventilation, but is generally not super relevant for intubation.
A patient with any concern for difficult intubation warrants consideration for factors also contributing to difficult LMA placement or cricothyrotomy. LMAs are difficult to place when the mouth opening is very small (about 2 inches) or the oral-laryngeal anatomy is unusual, and crics are difficult when the neck anatomy is impossible (eg a superimposed tumor, goiter, or heavily distorted anatomy). A patient who cannot have a cric may warrant an awake intubation to avoid the risk of inducing a patient who cannot be rescued.
Obesity is not a predictor of anatomically difficult intubation. Mask ventilation may be a little harder if there is increased oropharyngeal soft tissue. It is a predictor of physiologic difficulty (faster desat), though.
For emergent intubations: confirm code status, briefly evaluate the head/neck/mouth, use video laryngoscopy. Use hemodynamically stable agents for induction and reduce the dose, and ensure the team knows to subsequently sedate any patient who received a long-acting paralytic. Have a vasopressor drip ready, or better yet, running. Always set up everything and be prepared for every eventuality before you take away a patient’s ability to breathe.
Either RSI with paralytics, or perform awake intubation. Otherwise, never RSI the critically ill without neuromuscular blockade, which will reliably reduce your chances of success. Short-acting paralytics (succinylcholine) are brief—i.e. not much longer than the apneic period of a short-acting sedative—and long-acting paralytics (eg rocuronium) can be reversed with suggamedex, in the rare situations where letting the patient wake up and resume breathing is a smart move.
The one exception might be a ketamine-only intubation, which generally keeps the patient breathing, allowing you to either proceed to paralyzing or not depending on what you see, or maybe allow them to wake up.
While it’s nice if an emergent intubation has been NPO, it probably won’t change your technique; changes in gut motility in the critically ill mean almost anybody can have stomach contents. Treat most ICU patients as if they have a full stomach, i.e. RSI. The one exception: the PREVENT trial showed that mask ventilation during induction (usually a no-no for RSI) of critically ill patients does not increase aspiration risk and does reduce hypoxemia, so should probably usually be done.
In the highest aspiration risk patients like SBO or upper GI bleeding, keep the head of bed elevated, ensure ample/multiple suctions catheters, and be ready/willing to intubate the esophagus intentionally with your ETT and place it to suction to divert the stomach contents while you use a fresh ETT to intubate the glottis. Placing an NG beforehand to decompress the stomach is hit or miss as it can induce vomiting; it works better in a fully awake patient (who can manage any vomiting).
We should probably still learn and teach direct laryngoscopy, but do so using a video scope with regular-geometry blade.
We discuss head and neck surgery with Dr. Alexandra Kejner, otolaryngologist at the Medical University of South Carolina specializing in transoral robotic surgery, reconstructive surgery including microvascular free tissue transfer, salivary neoplasms, and sialoendoscopic procedures.
Robotics has enabled much less invasive approaches to many head and neck procedures.
Major airway procedures create edema, and there is always risk for bleeding, so patients often remain intubated overnight.
The other common ICU indication is a free flap, a portion of tissue (potentially including skin, subcutaneous tissue, muscle, even bone) removed from a remote site and transplanted into the head and neck area, with vessels anastomosed. These are at risk of failure and require close monitoring.
Most of these procedures will involve placing a tracheostomy, and potentially a PEG (or NG). This facilitates both surgical access and early recovery.
Tumors are superficially resected with adequate margins, then reconstruction begins. Meanwhile, exposure of deeper structures and deeper resection occur, which may involve a jig to guide the removal (prepared in advance from imaging), and a matching cut to prepare the flap tissue. Lymph nodes are removed en bloc. Then the flap is transplanted and vessels anastomosed (at least one robust artery and vein), using microsurgery and teeny sutures (often 8-0 nylon).
As a supplement to the clinical exam, an implantable Doppler monitor is occasionally left in place to augment post-op monitoring of perfusion, as well as sometimes a Vioptix near-infrared spectroscopy device which performs real-time tissue oximetry.
On POD 0-1, hourly nursing monitoring of the flap is usually needed, with periodic provider checks. Changes in the exam (swelling, turgor, cap refill, color), signal, or bleeding may require return to the OR for revision. A single ICU night is the norm, although comorbidities are common and may require a longer stay if the stress of surgery unmasks other problems.
Laryngectomy may be performed, involving removal of the larynx (voice box), leaving a blind pouch; the lungs no longer connect to the upper airway in this case, and the entire team should be aware of this anatomy, as the patient cannot be intubated or their airway otherwise managed from above.
Most flaps will be on a baby aspirin and enoxaparin, but occasionally may use a heparin drip.
Most will receive three doses of dexamethasone, both to reduce edema and to treat any adrenal insufficiency.
Chlorhexadine or salt water oral rinses are performed to keep the operative site clean.
Multimodal pain management is needed for both the oral site and the donor flap site.
A drop in the Vioptix signal from the initial post-op reader, neck swelling, or difficulty breathing (dyspnea, hypoxia, etc) all warrant immediate involvement of the surgical team for danger to the airway or the flap. Flaps might also turn purple from venous congestion, sometimes a little later, also a surgical emergency.
A questioned flap might be scratched to see if it bleeds (which is good).
A patient in shock might need vasopressors, fluid, or to be hypotensive, none of which are great for a flap. A balanced approach is probably best. A low-dose phenylephrine drip may be the most appealing pressor, and vasopressin might be the riskiest. MAP >65 is a minimum, some prefer higher (>80).
Intro-operative feeding has been used in some centers due to the prolonged procedure times.
Flap failure historically was most often from a venous clot, but this has reduced over time; nowadays it’s often late failures due to a salivary fistula contaminating the area and creating a region of digestion, clot, and breakdown.
Surgeons will occasionally request deeper sedation (or even forcing the patient to maintain a specific neck position) to avoid dislodging monitors, disrupting a very delicate anastomosis, etc.
A swollen or firm anterior tongue, ooziness in the mouth, or a difficult airway on the initial intubation may lead a surgeon to request delaying extubation.
The immediate post-op appearance usually heals into a better eventual aesthetic result. Occasionally measures like prosthetics can be used.
We learn about liver transplant with Dr. Meera Gupta, transplant surgeon at the University of Kentucky Healthcare Transplant Center, and surgical director of the Kidney and Pancreas Transplant Program. We discuss eligibility, triage, the peri-operative course, and important post-op complications.
Liver transplant eligibility is based on need, not time on the list. The MELD score (MELD 3 now, including albumin) is used for this, with MELD >9 (historically >15) considered the cutoff for transplant potentially exceeding the risk of not transplanting.
Livers can now be placed on warm perfusion pumps, allowing continued viability for much longer. This is mainly used in donors who died from cardiac death, those with high BMI or similar risks for primary non-function (i.e. the transplanted liver never starts working), and longer transport distances or expected operative times.
Incision is a large right subcostal incision, extended as needed. The liver hilum is dissected, preserving the feeding vessels. Caval clamping may be tested, then the liver is removed. This anhepatic phase in minimized to <60 minutes, preferably <45 minutes. The new liver is then anastomosed to the portal veins, vena cava, hepatic artery, and the bile duct. Some instability can occur during reperfusion, such as right heart strain, electrolyte abnormalities, or volume shifts.
Patients will usually remain intubated post-op, lines in place. Sedation ideally is limited so the patient can rouse and confirm the absence of encephalopathy. Systolic BP is closely watched (goal >90), as diastolic BP tends to be low in most liver failure patients. Hepatopulmonary patients can rest on the vent a little longer and are expected to remain on oxygen for the time being. Patients can be fed once extubated and stable.
High-dose steroids are loaded up front and then tapered, and oral immunosuppression initiated soon after.
Some AKI is common. Colloid like albumin is favored early.
Chronic thrombocytopenia is common and is monitored to determine when DVT prophylaxis can be started. Platelets >20k are targeted.
If INR >2, vitamin K is given empirically. FFP is usually not given prophylactically. Bleeding is usually considered a little preferable to clotting, in terms of ease of treatment.
A liver duplex is performed in the first 24 hours to ensure the new vascular supply is patent.
We learn about pancreaticoduodenectomy (the Whipple) with Michael Cavnar (@DrMikeCavnar), surgical oncologist at University of Kentucky, with a fellowship in Complex General Surgical Oncology from Sloan Kettering. He specializes in GI surgical oncology (liver, pancreas, stomach, etc), with ongoing research in GI stromal tumors and hepatic artery infusion pump chemotherapy.
The Whipple involves an aggressive resection and reconstruction of pancreatic head tumors. Along with the head of the pancreas, the entire duodenum, the bile duct (up to near the entry of the cystic duct), the gallbladder, and usually the distal third of the stomach, along with the nearby lymph nodes, are all removed. There are then anastomoses at the small intestine, the bile duct, and the stomach.
The pylorus of the stomach is generally removed, but can be left in a pylorus-sparing Whipple. The benefit of this is not well-established.
It is almost always done for malignancy (or occasionally for other conditions like pre-malignant changes or pancreatitis with stricture). Mortality in high-volume centers is a few percent, and usually involves deaths in the first 90 days due to various complications more than death in the OR.
Hypotension in the first 24 hours is a poor sign, as it may lead to bowel ischemia, portal vein thrombosis, anastomotic ischemia, or other injuries to vulnerable areas. If getting behind on hemodynamics, consider holding an epidural if present.
NG tubes are often placed to around 55 cm. They should not be advanced or replaced by the ICU staff, as the stomach has been shortened, and advancing the tube may traumatize the anastomosis. Bilious gastric drainage is normal in anyone with post-Whipple anatomy. Patients will generally remain NPO for several days.
Patients will emerge with 1-2 surgical drains. Output should be serosanguinous (sparsely bloody at most), less than 200ml/hr or so. It may occasionally be lymphatic (clear to lightly serosanguinous), which can be somewhat higher volume. Output should not be bilious or feculent.
Early bleeding requiring surgical take-back is uncommon and usually obvious in the drains, unless they clot, which can occur. A pancreatic leak can also dribble onto the stump of the gastroduodenal artery, causing erosion, and subsequent bleeding usually tracks back up into the bowel lumen and hence GI bleeding.
Respiratory distress can be treated with oxygen or high-flow nasal cannula, but positive pressure (eg. BiPAP) should be used with caution and consultation with the surgical team, particularly within the first two weeks post-op, as aerophagy can apply pressure to the bowel anastomosis.
A leaking pancreatic anastomosis causing fistula will tend to marinate the gastroduodenal artery’s anastomosis in pancreatic juices, creating a pseudoaneurysm; this can be managed early before it turns into massive hemorrhage. Any streak of fresh blood in the drainage should be considered a sentinel marker of this and immediately evaluated, usually involving CTA or pancreas-protocol CT. The treatment of choice is IR embolization and stent, not open repair.
Glucose control in diabetics will usually be worse post-operatively, both due to stress and due to removing a portion of the pancreas. SGLT inhibitors can cause strange metabolic effects as well if not fully washed out.
Exocrine pancreatic insufficiency can be discovered once feeding begins, usually manifesting as diarrhea (steatorrhea), and can be treated with pancreatic enzyme supplements.
A proton pump inhibitor should generally be used post-operatively.
Type B dissections do not involve the heart or coronaries, but Type A vs B nomenclature is falling out of favor versus more anatomically specific labeling; this system helps characterize the gray area between the innominate and the left subclavian.
The main sequelae of concern in type B dissection in end organ ischemia. This may be dynamic, due to movement of the flap to obstruct the feeding artery, or static, due to occlusion by thrombosis.
Hypotension is unusual in type B dissection and should be a red flag for another factor, such as involvement of the heart (coronary dissection, tamponade), or rupture.
Rupture is not a common event in dissection (as compared to aortic aneurysm), but can occur.
Medical management of type B dissection involves controlling the impulse against the dissection flap by reducing heart rate and blood pressure. SBP <120 and HR <80 are reasonable standard goals, but should be customized somewhat to the patient; allowing higher goals in a pain-free patient, particularly one who is experiencing sequelae of relative hypotension may be reasonable.
During initial presentation, impulse control may prevent dissection from extending over a period of hours. Later, once it has thrombosed and scarred, risk may be somewhat less.
Dissection involving the renal arteries can be explored using doppler ultrasound in skilled hands.
Focal neurologic deficit should prompt concern for both stroke, and (in the lower extremities) thrombosis.
First line is usually an IV beta blocker for heart rate and either IV beta or calcium channel blocker for BP. Esmolol is a classic beta blocker, although involves a large volume of infusate, and is not always very effective. Labetalol and nicardipine are nice choices. Nitroprusside is usually a rescue.
Favor the right radial artery for an arterial catheter, as the left arm will sometimes be needed for the repair.
Transition to oral agents as they stabilize. A repeat CTA 5-7 days from admission (often prior to discharge) is usually appropriate.
The most common indication for repair is aneurysmal degeneration at the dissection site. Extension of the dissection, in the setting of appropriate medical management, is less common although possible, and may also indicate the need for repair.
The primary goal of repair is to cover the entry to the dissection, and potentially stenting to expand the true lumen. When there is involvement of the iliac arteries, stenting is usually needed there. Malperfusion to visceral vessels is often corrected with these maneuvers, but they can be specifically stented or thrombectomy performed if needed.
Open repair of type B dissection has become vanishingly rare due to high morbidity and rare indication.
Stenting of the aorta creates risk for spinal cord ischemia, so keep BP higher. Extremity neuro changes should prompt driving the MAP >90, naloxone, and IV steroids.
Lumber drain placement probably reduces this risk, and can be placed either reactively or proactively. Neurosurgery and/or anesthesiology or interventional radiology may do this.
Shorter ischemic time to organs or extremities, and baseline vasculopathy (which gives time for the body to develop a tolerance to it), portend better recovery after revascularization. Prolonged ischemia to extremities may require amputation or at least fasciotomies to prevent compartment syndrome.
Aspirin and perhaps clopidogrel (with or without a load) will usually be needed post-operatively.
Infection of long-standing grafts are not common but can occur. Contrast imaging and perhaps tagged WBC scans (nuclear scintigraphy) can identify these. Surgical removal may or may not be possible and tends to be morbid.
We discuss the nuts and bolts of urinary infection with an obstructing stone with Ashley Winter (@AshleyGWinter), board certified urologist with a fellowship in male and female sexual medicine, and chief medical officer of Odela Health.
A patient with UTI (or even just undifferentiated sepsis) and a non-trivial ureteral stone generally needs decompression of the affected kidney, whether or not there is significant hydronephrosis on imaging. Hydro is sensitive to other factors, such as dehydration, but its absence does not rule out sepsis secondary to urinary obstruction. CT is more sensitive here than ultrasound, which is mostly useful for ruling in hydronephrosis. (Such patients will usually need a stent, not a nephrostomy, as the latter is difficult when there is little hydronephrosis.) From a urology perspective, the size and position of the stone is probably more important than the hydronephrosis.
That being said, be attuned to the possibility of a patient with another source of sepsis, and an incidental bacteriuria and kidney stone. Anesthesia and a urology procedure won’t help these people. A cleaner urinary sample (e.g. straight cath or Foley, if the initial sample was a “clean” catch) can sometimes help here.
Consider also that a completely obstructing stone may be hiding pyelonephritis because the bacteria and leukocytes cannot pass the stone. This is not a very common scenario, but can lead to a “clean” urinalysis, so consider it in a patient with an obstructing stone and septic picture.
Try to get a urine sample before giving antibiotics.
Intra-renal stones will usually not cause obstruction, but occasionally in the setting of abnormal anatomy they may, such as a stone in a caliceal diverticulum causing a local/segmental hydronephrosis.
Obstructing stone + UTI + unstable with sepsis = emergent decompression within hours. Overnight cases should generally be drained overnight. Stable patients can potentially wait longer.
Option #1 for decompression is a ureteral stent, which stretches from the intra-renal pelvis to the bladder, traversing the area of the stone, and is deployed via cystoscopy. Urine drains around the stent, not necessarily through it. Stents can usually only be left for a maximum of 3 months and should be removed when no longer needed (i.e. when serial imaging shows passage of the stone, or a procedure has been performed to remove the stone). Long-term stent requirements involve serial stent replacements. They are placed in the OR under some level of sedation. Very distorted anatomy, such as in oncology cases, may make it difficult to find the ureteral orifice or to traverse the ureter.
Option #2 is a percutaneous nephrostomy. These are placed by Interventional Radiology. The patient is proned (not possible in all patients), and imaging (usually ultrasound) is used to guide a needle to the renal pelvis, then a pigtail catheter using a Seldinger technique. This can often be done with local anesthesia only. Lack of significant pelvic dilation or large body habitus make these more difficult. The result is a nephrostomy tube and drainage bag, which can be aesthetically unappealing to many patients. Anticoagulation may be a contraindication since you’re puncturing the renal parenchyma. They are usually not intended to be permanent, but can be left long-term in some cases.
Stents tend to be more uncomfortable, sometimes creating flank pain or a sensation of the need to void even with an empty bladder. Urine can even reflux up the stent into the renal pelvis during voiding.
Urostomies can sometimes make the procedure to remove a massive intrarenal stone like a staghorn calculus, since percutaneous nephrolithotomy can use the pigtail for access. Smaller stones can be removed via ureteroscopy.
Some stones are impacted, which may be difficult to navigate across with a stent. Technical maneuvers can be attempted, but occasionally it can’t be done and nephrostomy needs to be done as a rescue.
Ultimately, stent vs nephrostomy often comes down to institutional and logistical considerations, such as availability of urology compared to IR. Many centers have policies on who to call first.
A common phenomenon is clinical deterioration after decompression. Some of this may be iatrogenic; both stenting and nephrostomy involve pressurizing the renal pelvis by injecting contrast, which can force out some bacteria into the circulation. Reducing the volume or rate of contrast injection may help with this.
Antibiotic coverage can be as routine for sepsis, but if there is complex urological history, remember to check prior cultures (including stone cultures, if available), which may reveal a history of resistant organisms.
Stented patients who fail to improve in the acute to subacute period may be experiencing stent migration. Check position with a plain x-ray (KUB); if the proximal portion is not curled, further imaging may be needed as it suggests it’s not in the kidney.
Stent removal can sometimes precipitate instability as well if there is some degree of infection present.
We explore the cutting edge practice of point-of-care ultrasound of the brain, including optic nerve sheath measurement, transcranial doppler, assessing midline shift, and more, with Aarti Sarwal, neurologist and neurointensivist, director of the neurocritical care unit at Wake Forest, and director of their neurovascular lab and ultrasound courses.
POCUS can potentially be used to identify elevated ICP by optic nerve sheath ultrasound or pupillary assessment (in patients with difficult-to-assess pupils due to edema or other factors). Midline shift can be seen and quantified via the temporal window, and hemorrhagic masses can potentially be visualized. Finally, spectral doppler of the cranial vessels can show changes in intracranial compliance, similar to that seen in formal TCDs during vasospasm.
Learning curve for these studies is probably around 50-100 studies until competence, but may be creeping closer to 30–50 and eventually lower due to improving education, and increasing awareness and skills with the general concepts being applied.
The linear probe can be used over a closed eyelid to visualize the iris, allowing assessment of pupil response when light in shined in the opposite eye; this can be useful when the lid cannot be opened, such as from edema or trauma. M-mode can even be used for quantitative pupillometry.
An increased diameter of the optic nerve sheath measured 3 mm from the globe (using the linear probe in a transverse, ear-to-ear axis) correlates with increased intracranial pressure, as the sheath is a continuation of the cranial space and tends to swell with higher ICP. Papilledema can also be seen here as bulging of the optic disc. Use the orbital or ocular preset, which reduces power (mechanical and thermal indexes) delivered to the eye.
A cutoff reflecting elevated ICP is usually somewhere in the 5-7 mm range. However, normal values vary a lot, and very acute ICP crisis can choke off the continuity and cause normal diameters, so simple measurement can imperfect (analogy: IVC measurement). Trending can be more useful if you can establish a baseline, and papilledema is somewhat more specific. In pediatrics, adjusting for head circumference can help.
Slower increases in ICP tend to be associated with larger optic nerve sheath diameters, whereas rapid increases may actually be associated with normal sheath diameter, due to edema at the basal cistern level choking off communication with the cranial vault.
Midline structures like the pineal gland, third ventricle, or septum pellucidum can be seen from the temporal ultrasound window; 85-95% of the population will have adequate windows here, at least on one side, with some decrease in old age. Males tend to have worse windows, windows worsen over time, and there is some ethnic variation.
Research is early, but distance of the midline structures from the probe can be compared with their distance from the opposite skull (i.e. in the deep field). Differences between the two can help diagnose and quantify midline shift. Caveats: it’s difficult to establish exactly the same angle when insonating opposing sides, and identical angles can be impossible due to limitations in the windows, so don’t compare that way. The region of edema may mean different structures are shifted while others are normal, or even that there is no shift (e.g. herniation is not lateral); ideally, pick a midline structure that makes sense for where their pathology is found. This is probably more useful for serial comparisons than absolute values, since your angle that penetrates the small temporal window will usually not be perfectly flat, but will be reproducible.
Global edema may not be seen in midline shift, but may be seen in the TCD waveforms. Spectral doppler of vessels like the MCAs should show low resistance waveforms in a normal brain (low systolic peaks, a long runoff and high diastolic), while in a tighter brain with higher ICP, resistance gets worse, with a higher systolic, quicker drop, and lower diastolic pressure, all the way up until diastolic pressures become less than zero and flow is oscillating (e.g. back-and-forth during the cardiac cycle, which reflects no overall flow and is consistent with brain death). TCD measurements can be directly extrapolated to ICP using a number of published formulas.
Research is early, but transtemporal B-mode seems to have good sensitivity (>95%) for detecting parenchymal hemorrhage in the brain, as long as it is large and fresh; new blood shows up as a hyperechoic lesion with shadowing.
We dive into when to initiate renal replacement therapy, the modalities, settings, and physics involved, troubleshooting problems, and more, with Dr. Paul Adams, a dual-trained nephrologist and intensivist at the University of Kentucky.
One of the better indications for early dialysis in the ICU patient is to control volume, which in an oliguric patient you know is likely to keep accumulating.
Help determine who is likely to eventually need dialysis (and hence deciding early vs late, not early vs maybe never) with a furosemide stress test: give 1-1.5 mg/kg of furosemide (160 mg is often about right), then if they don’t make about 1 ml/kg/hr of urine for a few hours, they’re likely to end up needing renal replacement therapy.
Realistically, most true indications for acute dialysis in the ICU are hyperkalemia, volume overload, or occasional toxicology.
CRRT is generally more effective at volume management, particularly preemptive volume management, because it continues throughout the day and can more easily keep up with inputs. It is also more hemodynamically stable.
CRRT can be done via CVVH (using convective flow to drag out fluid and solutes via pressure across a filter), CVVHD (using diffusion gradients to clear solute and fluid), or CVVHDF (using both). Which modality of CRRT is used tends to come down to institution and practitioner practice, although there are some clinical differences in amount of solute clearance and such.
Effluent is the balanced electrolyte fluid which is used for therapy, and can be run into the blood before reaching the filter (diluting it and improving filter life, but decreasing efficiency), after reaching the filter (purely to replace what was lost), and on the other side of the filter (creating a dialysis effect). Total effluent rate gets divided among these sites as you like.
UF (ultrafiltration) is essentially whatever fluid is lost that you’re not replacing.
About 25–30 ml/kg/hr is usually about the right effluent rate. A higher rate helps make up for interruptions during the day.
150–250 ml/hr bloodflow is about right; it generally has relatively little effect on clearance in CRRT (unlike in intermittent HD, where it directly impacts clearance).
Circuit life can be prolonged with anticoagulation. Heparin can be used either systemically or regionally (infused at the start of the circuit, then reversed at the end using protamine), or citrate can be used regionally (replaced with calcium at the end), although it requires close monitoring of ionized calcium levels (really the ratio between total and ionized calcium, since citrate-bound calcium still registers on total calcium assays; a total calcium more than 2-2.5x higher than ionized levels suggests citrate toxicity).
16–18 hours of CRRT is usually needed before you start to see an impact on serum solute levels. For critical levels like severe hyperkalemia, start with IHD instead to get a quick correction.
Pressure problems at the dialysis access are almost always due to anatomic issues like catheter placement. Try adjusting the line, such as placing it deeper. Reducing bloodflow may help, using a different site, or rarely pharmacologically paralyzing the patient.
Pressure problems at the filter (“transmembrane pressure” or TMP) are usually from clotting. Consider anticoagulation if not already being used, or pre-filter fluid. Inflammatory patients like in sepsis can have very dirty, clotty blood.
If a patient starts making 600-1000ml of urine daily, consider weaning of renal replacement. That is not common in the critically ill, even if they eventually have later renal recovery; transition to IHD is more common.
If volume inputs are still ample (many liters a day), it’ll be hard to keep up using IHD, since UF rates top out around a liter per hour. Stick with CRRT in that case.
Rhabdomyolysis “disproportionately” increases BUN and creatinine, since those are products of muscle breakdown; they may have adequate renal function (demonstrated by robust urine output) despite high numbers.