Episode 41: Preventing and managing complications (part 2) with Matt Siuba

Part two of our discussion with fan favorite Matt Siuba (@msiuba), Cleveland Clinic intensivist, on complications in critical care and how to prevent and manage them. Today we focus on respiratory failure after extubation, and unintentional self-extubation.

Takeaway lessons

  1. When considering extubation of borderline patients, extubating to high flow nasal cannula or CPAP/BiPAP is often a good compromise. This is probably at least a little better than waiting for them to struggle before applying the support, plus it’s easier to assess their course. They can always come off if they look stellar.
  2. Set up for extubation success by first optimizing volume status, sedation strategies, mobility, and other good liberation practices.
  3. If concerned about pulmonary edema, a trial of a “tube compensation” mode alone (versus pressure support with PEEP) may be a good “strict” trial, as compared to more primitive ZEEP or T-piece trials.
  4. Post-extubation stridor is not always predictable, although known airway trauma should raise suspicions. If severe, or even borderline, patients should be promptly reintubated. If more mild, a trial of a couple hours on medical therapies and NIPPV is reasonable. Try steroids (dexamethasone 10 mg IV or so).
  5. Cuff leak tests are not very predictive and as likely to mislead as help. Visual inspection of high-risk airways for laryngeal edema may be helpful, although remember that a large tube in a small airway may never have a leak (and always visually look tight), yet may not be at risk for narrowing after tube removal.
  6. Self-extubation should prompt emergent preparation to oxygenate and reintubate, although you can assess their stability before actually doing it. Remove the tube if still stuck in the mouth. Stop sedative drips that suppress breathing.
  7. If agitation precludes oxygenation, consider antipsychotics. Dexmedetomidine may be useful in this situation, but takes a good 30-60 minutes to get loaded, so you may need another agent as a bridge. Don’t use a loading dose of dex, but starting at a higher rate (>0.6) is smart.
  8. A patient intubated primarily because of agitation will usually do fine after extubation, whether intentional or accidental. The main problem is that agitation precludes a clear, easily-interpretable SBT.
  9. “Extubation hesitancy” is a common error in the ICU. Clinicians are overly hesitant about failed extubations but not worried enough about prolonged intubation courses from the failure to try. Accept that a 0% chance of reintubation means leaving people on the vent for too long; acknowledge risks, plan for fallbacks, and don’t take failure personally; optimize the circumstances; but in the end, try. Risk need not be zero, it should just be lower than the risk of continued mechanical ventilation. “Not everybody is going to be ready every day, but you should treat every day like it’s extubation day.”
  10. The immediacy of the psychological feedback when a patient self extubates gives it primacy and power in our minds. It’s easy to see its harms, while it’s harder to see the harms of the oversedation that prevents it. “Overcautious” is really “overmedicalizing” and is not a safer flavor of risk.

Episode 40: Making the diagnosis with André Mansoor

Dr. André Mansoor (@AndreMansoor), associate professor of medicine in Portland, Oregon, author of the excellent Frameworks for Internal Medicine, and contributor to Physical Diagnosis PDX, talks us through a complex case of encephalopathy and respiratory failure to illustrate some principles of diagnostic reasoning.

Takeaway lessons

  1. The hardest part of treating most diseases is making the diagnosis. Countless resources are available to assist with treatment pathways, but you won’t know which one to use if you don’t know what you’re treating. Empiric treatment is sometimes necessary in the critically ill, but it tends to obfuscate, not support diagnosis.
  2. Even diagnostic support tools (such as burgeoning field of artificial intelligence) will require clinicians to collect the contributory data points, such as history and physical exam findings; this still requires enough diagnostic acumen to guide the data-gathering process.
  3. The H&P is not “over” after the initial survey; an iterative back-and-forth process ensues between discovering diagnostic abnormalities and using them to formulate new hypotheses that direct additional, more focused questions and examination.
  4. Most hypoxemia is best approached by first calculating, or at least approximating, the A-a gradient. This routes you into completely different diagnostic pathways depending on the results.
  5. Guillain-Barré is best confirmed by lumbar puncture showing an elevated protein without pleocytosis. However, start with a physical exam suggestive of lower motor neuron disease.
  6. Start with a general history and exam, but after that, broad or “shotgun” testing, consultation, or empiric therapy is high in cost and complications, and just doesn’t tend to work. Formulate specific hypotheses and use your studies or consultations to test them.
  7. 90% of diagnoses are made from the history and physical alone. Lean on these as the cornerstone of your diagnostic process, not on high-tech tests.
  8. It’s reasonable to treat a patient who easily matches a standard disease script as if they have the common disease. But when they have features that appear a little different, that’s a good time to step back and work your diagnostic process methodically.

Episode 39: ECMO for COVID-19 with Kim Boswell

An overview of VV ECMO with a focus on COVID-19, with Dr. Kimberly A. Boswell (EM and CCM) of the University of Maryland, perhaps the busiest center in the country for COVID-related ECMO. We discuss evaluating for candidacy, induction, maintenance, weaning, and general approaches to the COVID patient.

Takeaway lessons

  1. The limited amount of ECMO resources has led to narrowing of criteria. Maryland has reduced their standard upper age limit from <65 to <55, BMI of <40, mechanical ventilation duration <7 days (formerly <10). Also consider other organ failures, as well as duration of symptoms—not just intubation—as a prolonged pre-intubation course suggests a late, potentially fibroproliferative phase of disease which may not be responsive to ECMO.
  2. Almost all COVID cases at Maryland have been VV ECMO; they have very rarely considered VA ECMO. The most obvious indication for the latter would be right heart failure, but in most cases, they would be more likely to use VV ECMO (or other medical therapies, such as inhaled vasodilators or diuresis) to unload the right heart, or else to consider severe cardiogenic shock to actually be a contraindication to ECMO (as it suggests a late stage of disease less likely to respond to aggressive care).
  3. There is no obvious timeframe which is “too early,” but patients already
    at ECMO-ready centers might reasonably wait longer to go on bypass, as it can be done quickly and safely when necessary without requiring interfacility transport.
  4. Cannulation can be done by whomever is skilled and trained, such as cardiac surgery, trauma surgery, trained intensivists, etc.
  5. For VV ECMO, Maryland likes to cannulate the right IJ and right femoral veins, or perhaps the left femoral if needed. They prefer not to cannulate bilateral femorals, and prefers not to use dual-lumen IJ catheters (the Avalon bi-caval catheter), as flow is often not adequate.
  6. Anticoagulate most patients on VV ECMO with heparin to a PTT of 45-55. VA ECMO can go to 60-80. ECMO without anticoagulation can be done if there are bleeding issues, however.
  7. Maryland generally does not titrate FiO2 on the sweep gas. After induction, titrate the sweep; the goal is usually to correct hypercarbia over 6–8 hours, not all at once.
  8. Flow rates at least 4 L/min, unless more is required for hypoxia. RPM <4000 is usually the starting goal.
  9. Prone even while on pump for lung-protective reasons. Chest PT is good too. Prone first for 6-8 hours to ensure tolerance and skin integrity, then do around 4 more sessions of 16 hours each, as a starting goal.
  10. Ventilator settings on VV ECMO can be walked back after induction. Historically they used PEEP 10, PIP 10, RR 10. In heavily consolidated COVID patients, some need more pressure to maintain some degree of recruitment, such as PEEP 15 and PIP 10.
  11. Inhaled vasodilators can be continued or weaned depending on right heart function. Diurese until you develop flow problems (suction events) on the pump, a useful indicator of low intravascular volume.
  12. Have a low threshold to deeply sedate and/or paralyze while on pump to optimize synchrony and facilitate proning. However, Maryland likes to perform “partial paralysis,” with just enough NMB to achieve goals; respiratory rates below 20 or so are considered acceptable.
  13. Early tracheostomy is reasonable, but persistently high requirements for ventilator pressures often pushes it back.
  14. Hypoxemia can occur in VV ECMO patients due to too much flow through the native circulation and shunted lungs. In such cases, beta blockade may actually improve systemic oxygenation.
  15. Plasma free hemoglobin levels may be a useful marker that changing your oxygenator could improve gas exchange.
  16. Decannulate at the bedside when ready, watch them for 24 hours, then boot them out of the ICU; they’re ready.
  17. 65%+ of COVID ECMO patients at Maryland are surviving. Data remains slim, but there seems to be decent results in a well-selected population.
  18. In rare cases, patients who neither die nor recover may become candidates for lung transplant.

Episode 38: GI bleeding with Elliot Tapper

Back with returning guest Dr. Elliot Tapper (@ebtapper), gastroenterologist, transplant hepatologist, and director of the cirrhosis program at the University of Michigan in Ann Arbor, to talk about critical GI bleeding.

Takeaway lessons

  1. Consider the Glasgow-Blatchford score to stratify risk and need for admission, GI consultation, etc.
  2. Octreotide (or terlipressin) is indicated in every cirrhotic with GI bleeding, i.e. patients with confirmed or probable varices.
  3. Proton pump inhibitors are appropriate for bleeding ulcers. Note they are not needed in variceal bleeding, and are not needed if octreotide is also being given; octreotide reduces gastric pH just as much as a PPI.
  4. Bleeding cirrhotics should receive antibiotics. They have a high risk for inpatient infections, whether from bacterial translocation, instrumentation, etc.
  5. By and large, twice-daily PPIs are as good as PPI drips. The latter is mostly an evidence-free Hail Mary addition.
  6. As a general rule, colonoscopy for lower GI bleeding rarely needs to be done urgently; at most, early colonoscopy (within 24-48 hours) may reduce length of stay, but the yield of finding intervenable findings (particularly in unprepped bowel) is extremely low.
  7. In very unstable patients, it is not very common that you would need to place a gastric tube and perform lavage to prove an upper GI bleeding source; just do the EGD. In less obvious cases it can be quite useful, though. Don’t be misled by trace amounts of bleeding, which can occur (due to stress) even in lower GI bleeds.
  8. NG/OG placement in the setting of varices is safe, unless there’s been recent banding performed.
  9. Early CTA is a good approach for severe lower GI bleeding. It is basically never the first line approach for presumed upper GI sources, however; IR embolization is less effective here (due to the redundant blood supplies), and endoscopy will help localize the bleeding source for any needed embolization anyway.
  10. “Early” EGD usually means within 12 hours and is appropriate for active bleeds that are not catastrophic.
  11. Although massive bleeding can result from varices, with good medical treatment it is actually rare. In most cases judicious transfusion can be used to avoid overly increasing venous pressures.
  12. For truly rapid upper GI bleeds, intubate early (to prevent aspiration and facilitate EGD), ensure adequate IV access, and perform emergent EGD; endoscopy remains the first line treatment. Even when visualization is difficult it provides useful information by localizing the bleeding region. Normal (Hgb >7) transfusion targets are not relevant in active exsanguination. A PPI drip is reasonable but is not particularly high yield at this point.
  13. Balloon tamponade (Minnesota or Blakemore tubes) has its own risks, and few clinicians are expert at their placement. Overall it is rarely needed unless endoscopy is not immediately available, such as if a patient needs transfer to another institution.
  14. EGD can lead to rescue surgery if it visualizes bowel perforation, and to IR if a bleeding vessel is found that can’t be addressed endoscopically.
  15. Repeat endoscopy during the same hospitalization is rarely needed. For ulcers, it is common to re-scope in about 8 weeks to make sure it has healed and is not cancerous.


The Glasgow-Blatchford Bleeding Score (GBS) to stratify risk.

Episode 37: Airway management for COVID-19

Back again with Dr. Ross Hofmeyr (@rosshofmeyr), anesthesiologist in the Department of Anaesthesia and Perioperative Medicine at the University of Cape Town, to discuss an expert’s perspective on airway management in the COVID-19 patient.

Takeaway lessons

  1. Good practices for intubating COVID patients are, by and large, good practices for intubating anybody. Using a standardized protocol, appropriate PPE, applying best practices to optimize success, and pre-assigning roles has no downside.
  2. Support each other by using “call/response” checklists and buddy checking PPE.
  3. Ross’s protocol: one attempt at intubation, immediate placement of supraglottic airway if it fails, then proceed to another attempt. First line with video laryngoscopy using a Macintosh blade. No mask ventilation (to limit aerosolization) except as third line if SGA fails. Mask with two hands, two operators, and a PEEP valve.
  4. Patients need oxygenation, and to a much lesser extent ventilation, but not tubes per se. Whatever method achieves that in an emergency is okay.
  5. You need PEEP to preoxygenate the hypoxic COVID patient. High flow nasal cannula is okay, but a BVM with PEEP valve provides real PEEP and usually improves preoxygenation. HFNC with a mask on top is less clear as the large cannula can cause air leak.
  6. Learning to bag-mask ventilate on mannequins teaches bad habits. Learning in the OR with real humans and an anesthesia bag is a better place.
  7. Intubate everyone with head of bed elevated PLUS head in a sniffing position. Blankets are better than pillows. Start with more elevation than you need; it’s easier to remove than to insert.
  8. Move the bed. True 360 degree access to the bed makes a difference.
  9. Proper preparation makes most of the difference to success. Even experienced anesthesiologists have dramatically reduced first-pass success when removed from their usual OR setting, likely due to less preparation.
  10. By and large, different types of PPE should not affect intubation success if the team is highly-skilled.
  11. Ross’s team favors induction with fentanyl, etomidate, and succinylcholine (unless hyperkalemic, then rocuronium). The small advantage in speed with sux is worth it in these rapidly-deoxygenating patients.
  12. Use a verbal call/response checklist to make sure nothing has been missed, slow down the pace, and create a shared mental model among the team (particularly if not everyone is part of the usual group). This only takes a significant amount of time if you actually find deficiencies that need correcting (in which case you’ll be glad you took it), and it adds value almost every time.
  13. Many patients will be dehydrated and hypovolemic at the time of intubation, particularly if they’ve been on non-invasive for some time (often not eating/drinking) and most of all if they’ve been on non-humidified oxygen, such as regular cannula and/or masks.


SASA (South African Society of Anaesthesiologists) COVID-19 protocol and recommendations

Episode 36: Preventing and managing complications

Back in the arena with one of our favorites, Matt Siuba (@msiuba), Cleveland Clinic intensivist and Mr. Zentensivism, to discuss complications in critical care and how to prevent and manage them. Today we focus on atrial fibrillation with RVR and bleeding after thoracentesis and related other procedures.

Takeaway lessons

  1. Rapid atrial fibrillation in the ICU should be considered a “symptom,” not a disease per se. Look for stressors or triggers for tachycardia, such as infection, agitation, etc. Resume home agents if they exist — or don’t hold them to begin with — especially beta blockers, as rebound can occur with discontinuation. Don’t get too hung up about converting the rhythm. Give magnesium early and often, acknowledging that rapid administration tends to provoke rapid loss to the urine and you may be better served to stretch it out.
  2. A-fib with a rate below the 130s-140s is unlikely to be the cause (rather than an effect) of shock, outside of structurally abnormal hearts that need filling time or atrial kick (such as diastolic failure).
  3. Remember that you have time to address rapid A-fib in a stable, minimally symptomatic patient, regardless of the rate. You can only make them less stable. Go slow and be thoughtful.
  4. Good reasons to perform therapeutic thoracentesis include work of breathing. Less common reasons include hypoxemia. If you suspect you may need to re-tap, consider leaving in a drain.
  5. Under ultrasound, put color doppler on the thoracic wall to confirm there are no unexpected vessels at your puncture site; do this in two planes and use a superficial probe.
  6. You do not need to use real-time ultrasound guidance for the thoracentesis puncture unless the pocket is quite small; you can always ultrasound the wire after it’s in place if the wire entry felt weird. It takes some practice to maintain a good relationship to the rib while also guiding yourself under ultrasound.
  7. Anchor your needle hand to the patient so unexpected movement will not shift your position, and use the smallest needle necessary. Consider performing smaller thoracenteses with a micropuncture kit rather than with a larger catheter like a pigtail; insert the micropuncture sheath and use it to drain the fluid. Small needle, small catheter, safe.
  8. A “dry tap” with your thora needle should prompt a different technique, not repetitions of the same one. Change something or check your position to ensure you’re not below the diaphragm. After one or two attempts, consider handing over to someone more experienced.
  9. Finding blood in your pleural tap should make you pause, but not panic. Traditionally you can send it for a hematocrit, but this is rarely very useful. Generally you can complete the tap and see if it clears. Afterwards, reinvestigate the space under ultrasound to ensure no blood is reaccumulating, and monitor the patient closely; occasionally they may need a CTA and embolization. Consider leaving a drain to monitor output, although be sure to flush it regularly to prevent clotting. Investigate for other reasons there may be hemothorax, such as trauma, previous taps, or malignant exudates.
  10. If you suture a line or other device and it won’t stop bleeding, you may have caught a superficial vessel (e.g. the EJ when performing an IJ). Take those sutures out or it’ll never stop.
  11. Complications happen. They should generally prompt introspection to consider whether practice should be changed: could I have been better prepared to do that? Was I rushed? Was my mindset wrong? Should I be using a different technique? And so on. However, sometimes practice is optimal, and complications are simply the inevitable result of intrinsic risk; in such situations, changing practice can only mean worsening it. Errors of omission (failing to perform needed interventions) should not be judged as worse than errors of commission (complications of the intervention).
  12. When everything is done right, and something bad happens, everything was still done right.
  13. Learn from near-misses more than from complications; they are more common and it’s safer for everyone. But to do this, you must acknowledge sticky situations, not ignore them or gloss over them; the tricky or “challenging” case should not make you applaud that you had the moves to recover from it, but to ask how you can prevent it in the future.
  14. Support each other when complications occur, as some amount of self-blame is common, and can easily become excessive and harmful — even when an error truly was made.


A-fib in the ICU: Drikite L, Bedford JP, O’Bryan L, et al. Treatment strategies for new onset atrial fibrillation in patients treated on an intensive care unit: a systematic scoping review. Crit Care. 2021 Jul 21;25(1):257. doi: 10.1186/s13054-021-03684-5. PMID: 34289899; PMCID: PMC8296751.

Episode 35: When to operate in trauma with Dennis Kim

Looking at trauma from the perspective of a surgeon, with a focus on the perennial dilemma of when a patient needs surgery. Our guest is trauma surgeon Dr. Dennis Kim (@traumaicurounds), associate professor of Clinical Surgery at UCLA and medical director of the Harbor-UCLA Medical Center SICU, as well as host of the Trauma ICU Rounds podcast.

Takeaway lessons

  1. Trauma patients who are hypotensive or otherwise unstable should be assumed to be bleeding, bleeding, bleeding until proven otherwise, and should have a very low threshold to proceed directly to the operating room for exploration.
  2. Airway is not the first priority in most trauma patients and can often wait until a patient is resuscitated—in many hemorrhaging patients, it can wait until the OR. Likewise, many penetrating injury patients with palpable pulses can wait for further resuscitation (whether blood or anything else) until surgery. The treatment for bleeding is hemostasis.
  3. The exception is patients with concomitant brain injury, in whom permissive hypotension should not be allowed. However, don’t delay the unstable patient from the OR by getting a CT of the head.
  4. Don’t forget examination of the back, and hair-bearing areas like the axillae and groin, which can easily hide penetrating wounds.
  5. Consider using the shock index, the heart rate/SBP, to detect underlying shock. Over .8 or 1 is highly suspicious.
  6. Operative prep for exploratory laparotomy is usually from the chin to the knees. Although a midline laparotomy incision is the typical starting point, injuries can track more widely than you expect, and there should be the ability to open the chest or access the groin (for femoral exploration, conduits, etc) without repositioning or re-prepping.
  7. Damage control surgery involves evacuating hematoma, packing to provide initial hemostasis, then securing bleeding (by coagulation, suturing, packing, etc), resecting or reperfusing ischemic tissue, and securing injured bowel. In patients with continued metabolic or coagulopathic instability, surgery typically stops there with the abdomen left open and a wound vac (e.g. Abthera) placed. More stable patients may tolerate more extensive initial repairs.
  8. The most expeditious repair for open bowel injury is simply stapling the bowel shut in discontinuity. However, there are some arguments for repairing it early (by anastamosis/stoma creation), as discontinuous bowel becomes edematous, becomes an obstacle to later closure, and may be difficult to eventually reconnect.
  9. Orthopedic injuries should be manually reduced and perfusion ensured, splinted (e.g. pneumatically), then definitively addressed by Orthopedics at their convenience. Traction splinting is usually not done in the ED. In patients planned to receive a contrast CT, perfusion to a threatened limb can often be easily evaluated by simply adding an arterial study (extremity run-off) to your pan-scan.
  10. Ultrasound (eFAST) and plain x-rays (chest and pelvis) are useful tools for rapid evaluation in the ED. Although not as definitive as CT, they are safer and quicker, and can rapidly rule-in many problems needing immediate intervention.
  11. Instead of giving TXA as the CRASH dose of 1 g up front plus a 1 g drip, give 2 g upfront. The drip tends to get forgotten.
  12. Bowel edema noted on CT should raise suspicion for occult bowel perforation, which is very difficult to primarily visualize on CT.
  13. Serial abdominal exams are a valid way to follow a questionable abdomen. These should ideally be repeatedly done by the same person, looking for worsening tenderness, pain, or rigidity, and combined with lab trends (e.g. trending a CBC every 4–8 hours to follow the leukocytosis).
  14. The lungs mirror the abdomen! Worsening respiratory status should raise suspicion for a worsening abdominal process, such as evolving infection. Evaluate such things with a contrast CT; non-contrast scans are very difficult to interpret in a complex abdomen.
  15. When assessing for possible infection in a post-surgical abdomen, antibiotics, patience, and possible IR-guided drainage (after enough time has passed for abscesses to organize) are usually the mainstays. Surgical exploration is helpful only when a specifically surgical process exists, such as a leaking dowel, and becomes increasingly risky as time passes and adhesions develop. Diagnostic laparoscopy or enteral contrast can occasionally be helpful in picking up a surgical leak.

Episode 33: Ischemic stroke with Thomas Lawson

Evaluation of ischemic stroke, decisions for tPA and thrombectomy, supportive critical care, and monitoring for cerebral edema—with returning guest Thomas Lawson (@TomLawsonNP), nurse practitioner in the neurocritical care unit at OSU Wexner Medical Center and James Cancer Hospital. Thomas is now also a PhD student at the OSU College of Nursing where he studies the epidemiology of delirium in critically ill stroke patients, and recently cofounded the Board Review Associates AG-ACNP board review course.

Takeaway lessons

  1. The first priorities after suspected ischemic stroke is head CT to rule out hemorrhage, screening for tPA rule-outs, and establishing a “last known normal” time. The window for systemic tPA is 3-4.5 hours from known onset of symptoms, although the sooner the better.
  2. Patients who are not candidates for tPA should receive some form of perfusion imaging (such as a CT perfusion scan) to characterize the size of the infarct and at-risk penumbra, as well as angiography (e.g. CTA) to assess for the presence of large-vessel occlusions (e.g. ICA, M1), which may mean there is an opportunity for endovascular interventions such as thrombectomy. Smaller vessel occlusions are a less evidence-supported zone and are more dependent on the interventionalist’s judgment.
  3. If there is an opportunity for thrombectomy in a patient who is not a tPA candidate (and these days, potentially even one who is), yet your center cannot perform it, consider stat transfer to somewhere that can.
  4. The preferred initial (and follow-up) neurologic exam is the NIH stroke scale. Check a blood glucose and blood pressure as well.
  5. Post-tPA and/or thrombectomy patients should go to a neuro-capable ICU for hourly neurologic checks and blood pressure control.
  6. BP should be maintained under 180/105 after tPA, or usually somewhat lower (e.g. <160) after thrombectomy alone. Use labetalol and/or hydralazine but have a low threshold for using a drip, usually nicardipine. The newer, quicker-titrating drip clevidipine can be useful too. Later, blood pressure targets can be down-titrated and eventually brought towards “normal” — but watch out for pressure-dependent neurologic function from imperfectly-reperfused stenoses.
  7. Sodium should ideally be normal-ish early. If edema occurs later in the setting of large strokes, you may need to drive sodium up to provide an osmolar gradient, either by a hypertonic infusion or intermittent boluses.
  8. Chemical DVT prophylaxis should be held 24 hours after tPA (along with aspirin and pretty much anything else that could make you bleed), after which you’ll repeat a CT to screen for bleeding. If still none, you can and should start heparin or LMWH. When to start or resume full anticoagulation is a more nuanced question, as large strokes always portend a risk of hemorrhagic conversion.
  9. A non-urgent MRI is usually nice to evaluate the degree of infarction, although it infrequently changes care.
  10. Medium- and long-term recovery is variable and depends heavily on the patient and various other factors including luck. Early, high-quality rehab is key.

Episode 31: Practical mobility, awakening, and delirium prevention with Kali Dayton

The art of taking a critically ill, heavily sedated, floridly delirious patient on aggressive vent support and pulling them out of the loop of sedation, immobility, and delirium. With Kali Dayton, ACNP-BC (@HomeIcu), critical care nurse practitioner and host of the Walking Home from the ICU podcast, where she looks closely at these issues, including interviews with survivors describing their ICU experiences.

A spiritual successor to our talk with Dale Needham, this time focusing more on details and practical approaches.

Takeaway lessons

  1. Good care to optimize long-term function is also good care to optimize short-term survival and morbidity.
  2. Benzodiazepines are normally a poor choice for sedation given their deliriogenic properties. However, using benzos in patients with alcohol dependence is more appropriate. It can also be rationally used in more subacute patients in whom benzos aren’t being used as sedation but only as anxiolytics—i.e. low doses of an agent like clonazepam to preserve level of arousal but treat anxiety, much like it’s used in the outpatient setting.
  3. The combative behavior of delirious patients isn’t inexplicable; it’s a rational response to their perceived situation, which often involves vivid hallucinations of sexual abuse, torture, fractured realities, threats to loved ones, and similar horrors.
  4. Favor dexmedetomidine in patients who do need a sedative drip, but aim only for calm, not a depressed level of consciousness.
  5. Delirious (non-combative) patients can often still be mobilized to the extent tolerated, and it tends to actually improve their mental status. Limited activity is better than none. Concerns for self-extubation are usually easily managed by gentle restraint or redirection, as these patients are usually physically weak. A dexmedetomidine drip is not necessarily a contraindication to mobility.
  6. Ventilator settings are rarely a contraindication to mobility. Increased FiO2 may be necessary and is acceptable. Modes can be adjusted as needed. While exertion may increase respiratory needs, this change is rarely precipitous or “dangerous”; adjustments can simply be made as needed.
  7. Fatigue induced by exercise is a good thing and may facilitate further reduction in sedation. Allow patients to nap, but not too long (to preserve a normal sleep-wake cycle). In an ideal world, aim for three mobility sessions daily: two on the day shift and one before bed.
  8. Proning does not necessarily mandate deep sedation and/or paralysis. It can be a “therapy, not a lifestyle,” with patients proned for a period of time (but awake and interactive) and then turned back up to perform mobility and other activities.
  9. Awake, non-delirious patients can require more “work” to mobilize etc, but in many other ways require less work. They understand their situation, can assist with their own care, protect their own tube, etc. They are part of their care, not working against it.
  10. With good care, tracheostomies are rarely needed for the most common reason of oversedation and weakness. Mobility and light sedation can be practiced without them. However, they may still occasionally be needed for truly refractory lung disease or anatomic issues like airway abnormalities.
  11. Sedated patients appear to be resting and comfortable, but they are not, and follow-up interviews reveal they are actually internally suffering from their delusions. On balance, most would much rather be awake but experiencing their true reality (even if bored or uncomfortable) rather than sedated and experiencing the horrific false realities of delirium.

Episode 30: Diabetic ketoacidosis

Diagnosing and treating DKA, including fluid management, lab studies, insulin management, managing acid-base abnormalities, transitioning off your drips, and all the rest.

Takeaway lessons

  1. Calculate your anion gap and perhaps your strong ion difference (or bicarb gap). In most cases, consider checking a b-hydroxybutyrate and a lactate to confirm the diagnosis, but hyperglycemia + anion gap generally equals DKA.
  2. Ask what triggered DKA. The most common causes are medication non-compliance (or an inadequate regimen), and a stressor like infection.
  3. Bolus fluid until euvolemic, just like any patient. These people are often severely hypovolemic, particularly from polyuria, but they vary; you’ll need to assess them and decide their needs. Ultrasound and clinical examination are helpful.
  4. Start an insulin drip, with or without a bolus. A common regimen is 1 unit per 10 kg of bodyweight as both a bolus and a starting drip rate. Check hourly fingersticks and adjust as you go to reduce the glucose at a modest rate.
  5. Check q4h basic chemistries to follow electrolytes and the anion gap. If potassium gets down into the normal range, give more. If it gets low, stop the drip; it’s going to get lower. You can check a blood gas up front (a VBG is fine) but it usually doesn’t need to be trended.
  6. Once the glucose drops below 200-250, start some IV fluid containing dextrose. This prevents “overshoot” and allows you to continue the insulin drip at a low rate until the ketosis has cleared.
  7. Once the gap has been closed for two consecutive checks, you can transition to a subcutaneous regimen. Give long-acting insulin, wait two hours, then turn off the drip. Calculate the dose by either restarting their home regimen (if it was previously effective) or by estimating their 24-hour insulin requirement, splitting it into 50% basal and 50% short acting, then cutting that basal dose to about 50-80% to create a safety margin. Give short-acting as either fixed prandial doses or sliding scale, either qACHS (with meals and at night), or q4-q6h.
  8. Once the drip comes off, they should eat some kind of meal.
  9. Check one more chemistry, then they can usually leave the unit.
  10. Alcoholic, starvation, or medication-related ketacidosis presents like DKA, but without severe hyperglycemia. If severe, treat them similarly, but since it won’t take long to drop their glucose, start supplemental dextrose early. Mild to moderate cases don’t need insulin at all, only nutrition.
  11. In general, disable insulin pumps upon admission; they can potentially be restarted once DKA has resolved. Endocrinology is helpful for this.