The ICU trauma resuscitation

Approaching the major trauma patient is a semi-standardized, team-based, protocolized process, well-described in courses like ATLS and smoothly honed in most busy trauma centers. Often, though, it isn’t an ICU process per se, because it tends to occur before patients reach our doors; unstable trauma patients visit the ED and then proceed directly to the operating room, only reaching the ICU after their initial stabilization and workup.

When do we get to play? Usually it’s in one of two situations:

  1. A patient’s injuries are non-operative. This includes patients “too sick for surgery,” those with borderline goals of care (i.e. still a full code but not very appealing surgical candidates), and patients with real injuries that are simply not amenable to emergent surgical repair.
  2. Patients who have already underwent damage control surgery and now have ongoing “medical” bleeding—aka oozing—due to coagulopathy.

So, a typical scenario might look like this:

A 28-year-old male with multiple gunshot wounds to the chest and abdomen is brought to the operating room by the trauma surgery team. Exploratory laparotomy and thoracotomies are performed with a portion of bowel resected, solid organs repaired or removed, and vessels ligated. Bleeding continues without a clear surgical source, and the patient’s metabolic profile is deteriorating. Packing is placed, drains are inserted, the abdomen is left open, and the patient is brought out to the ICU for medical stabilization before further definitive surgical repairs can be attempted. They arrive intubated, on multiple pressors, having received countless units of blood, and rapidly approaching cardiac arrest.

Similar non-trauma stories can also unfold after major surgical disasters, massive GI bleeds, and a few other situations. What do you do with these folks?

The basic “ABC” approach to life support can be followed, but the resuscitation can be boiled more simply down to three axioms:

  1. Replace the blood
  2. Stop the bleeding
  3. … and don’t miss anything while you’re busy with #1 and #2

Let’s discuss these in order.

Replace the blood

The treatment for hemorrhagic shock is blood transfusion.

The role for vasopressors is unclear and limited. A very low-dose of pressor (e.g. norepinephrine) in the background may help to gently squeeze the venous capacitance vessels and return pooled blood to the circulation. And as hypovolemic shock progresses, some degree of SIRS-like distributive physiology probably superimposes; after all, loss of vascular tone and cardiac failure is the final common pathway of most shock states.

But mostly, you see pressors used as a last-ditch effort to maintain the blood pressure when we get behind on volume. While this is not unreasonable, they should be weaned off ASAP. The primary shock pathology here is hypovolemic, and replacement of intravascular volume is the treatment. Although some of the details of this process remain controversial, the core principles are clear:

  1. Use primarily blood products. The role of crystalloid or colloid fluids for ongoing bleeding should be limited to a temporizing measure while you wait for blood. These fluids carry neither oxygen, nor platelets, nor clotting factors, and therefore dilute them into ineffectualness; they also raise the blood pressure, increasing hydrostatic forces that blow off blood clots and impair attempts at hemostasis. Bolusing a balanced crystalloid (e.g. Plasmalyte) while blood is unavailable is better than proceeding to cardiac arrest—whether it is better or worse than using pressors is unclear—but if blood is present, use blood.
  2. Transfuse them in a balanced ratio. Historically, bleeding patients would receive mainly packed red blood cells. However, since this does nothing to help them clot, they would tend to keep bleeding. Patients bleed whole blood, and most contemporary data agrees that a fluid resembling whole blood is the best replacement. Since actual whole blood is rarely available in the civilian world, the best we can do is to give packed red blood cells (PRBCs), fresh frozen plasma (FFP), and platelets in approximately equal numbers: in other words, give one unit of PRBC, then one unit of FFP, then repeat. Since platelets are usually pooled into apheresis packs equivalent to 5–6 units worth, a pack of platelets should be given about every 6 units of PRBC/FFP. While an obsessively perfect 1:1:1 ratio is probably not necessary, deviating too far from it should be avoided.
  3. Warm them up and give them fast. If bleeding is brisk, you will only keep pace with equally brisk transfusion. Venous access must be large-bore; a minimum is two large IVs (18 gauge or much larger), but central access should be obtained if possible, since it is far more reliable and allows more options for medications. The most common large-bore central catheters are introducer sheaths (“Cordis” catheters) and hemodialysis catheters; through these, you should squeeze blood using a rapid infuser device, the most common being the Level 1 and the Belmont. These not only use powerful infusion pressures to produce incredibly fast flow rates, they have air detectors to ensure that large bubbles do not come along for the ride, and most importantly, fluid warmers to heat up your chilled blood products to body temperature.
  4. Use a massive transfusion protocol. Every hospital should have an “MTP.” In contrast to the usual sluggish process of crossmatching blood, verifying it ad infinitum, and transfusing it slowly, an MTP sets into motion a rapid-sequence protocol: the blood bank will prepare a “batch” of un-crossmatched blood (PRBCs, FFP, platelets), send them to you in a cooler, and then immediately begin to prepare another batch. This will continue until you tell them to stop.
  5. Transfuse enough, but no more. The goal is a well-perfused patient. Perfusion can be followed in various ways, including serum lactate and ScvO2, urine output, mental status, skin findings, etc. Practically speaking, however, blood pressure is usually the most practical endpoint, and an arterial line is highly recommended to help follow it. Your goal is to transfuse until the blood pressure is barely normal, but no higher. Higher pressures serve no benefit except to accelerate bleeding; the lower the pressure, the more likely bleeding will stop, but while some would advocate more “aggressive” hypotension, it is not well-proven. Therefore, the familiar MAP goal of 65 is probably reasonable. When the MAP dips, give balanced blood; when you reach 65, stop. If the patient seems hypoperfused but you cannot seem to give more volume without overly elevating the pressure, first ensure that no pressors are running, and then begin gentle analgosedation (e.g. small boluses of fentanyl). While no sedation is typically needed in deep shock, once well-resuscitated the patient may begin to rouse; a bit of opioid will keep them comfortable while also defraying their intrinsic adrenergic tone, relaxing their vasculature and allowing you to replace more of their circulating volume.
  6. Consider autotransfusion. Blood lost into chest tubes can be reclaimed and returned to the patient using commercial auto-transfusion devices (such as the Atrium). This can help supplement banked blood and is probably fresher and more physiologic. However, it is not quite the same as “whole blood,” because even a short stay in the canister renders it low on platelets and clotting factors (as well as high in potassium). It should therefore be calculated as equivalent to PRBCs in your 1:1:1 ratio.

Stop the bleeding

You can give blood until you exhaust your regional supplies, but if the bleeding never stops, it will be to no avail.

We have already discussed two of the most important principles of limiting coagulopathy: maintaining a low-normal BP, and avoiding crystalloid in favor of balanced blood products. However, is also critical to:

  1. Maintain normothermia. Hypothermia is common in trauma patients, since they are stripped naked and then filled with cold fluids. Battle this by giving all blood through a rapid infuser, and if possible, warm other infusions with a simple IV fluid warmer. Keep the room warm, cover the patient with more blankets than you think are necessary, use a forced-air heater (i.e. a Bair Hugger), and check the patient’s temperature frequently.
  2. Replete calcium. Serum calcium is chelated by the citrate anticoagulant in stored blood. Large-volume transfusions will therefore deplete calcium levels. This is a bad thing, since calcium is both an essential cofactor for clotting, as well as necessary for muscle contraction—including vascular tone and myocardial squeeze. A gram of calcium chloride will probably be needed every half-dozen units or so; follow this using ionized calcium levels, often measured on blood gasses.
  3. Prevent acidosis. Acidosis directly impairs platelet function and clot formation. Unfortunately, it is hard to directly treat; the best way is to simply resuscitate effectively and allow lactate to clear itself. Compensatory hyperventilation to the extent possible without breath-stacking or impairing hemodynamics is appropriate; do this by increasing the respiratory rate, not the tidal volume, which should be set to lung-protective ARDSnet volumes. (Some would controversially suggest using large volumes and a slow rate, to optimize preload to the heart, but this is not an evidence-based strategy.) Continuous dialysis is technically possible but usually not feasible or tolerated. Sodium bicarbonate probably has relatively little role, although a modest infusion of pure bicarbonate (i.e. not diluted in D5w or other fluids, and not rapidly pushed) may have some salutary physiologic effect. Or maybe not.
  4. Address specific deficiencies in clotting. Even with balanced-ratio transfusion, some components in the clotting cascade may need additional repletion. Maintaining a fibrinogen level >100 or >150 in a bleeding patient is appropriate; while some fibrinogen is present in FFP, far more can be given using cryoprecipitate, which is usually not included in a standard MTP except by request (you will need to follow levels). Since one unit of cryo is tiny, it is generally pooled; 10–20 units of cryo is the most common dose. Other clotting factors, such as prothrombin complex concentrates (PCCs) and factor VIII (FEIBA), or agents like DDAVP or idarucizumab (Praxbind) are usually only used for patients with known anticoagulation or factor deficiencies. One exception is tranexamic acid, which helps obtund fibrinolysis and should be given early (often in the ED or OR). The other is recombinant activated Factor VII, which is occasionally used as a last-ditch effort to jump-start clotting. It probably does this, but with a serious risk of causing pathological de novo thrombi; only in very rare cases, and probably using an institutional rather than whimsical approach, should it be considered.

Don’t miss anything else

Replacing blood and stopping the bleeding are the cornerstones of trauma resuscitation. But while you’re running around doing that, you must be careful not to become so tunnel-visioned that you miss other, critical problems. Remember to:

  1. Send regular labs. The process described above is mostly not lab-driven; labs are too slow, and transfusion should be based upon vitals and clinical findings. But you should still send labs at least every few hours, including blood counts, chemistries, blood gasses, and coagulation studies (including fibrinogen). They serve as your score card, letting you know whether you’re making progress or losing ground; but they also catch idiosyncratic problems you’d otherwise miss. A common example is hyperkalemia, caused by the high potassium level in stored blood; it can be managed by the usual shifting strategies, such as insulin/dextrose and albuterol.
  2. Attend to the airway and breathing. Some degree of hypoxia is common, even if the lungs were initially uninjured; the inflammatory state here is profound, and TRALI is a real possibility. Vent management should proceed as usual, although your ability to use PEEP will be limited by hemodynamics. In general, respiratory issues should be addressed but kept on the back burner; they will rarely be the cause of death.
  3. Keep an eye on the heart. Cardiac stunning can occur due to direct blunt trauma or stress cardiomyopathy—never mind actual STEMI, which occasionally occurs as well. If cardiogenic shock develops, inotropic support may be necessary. Echocardiography is invaluable.
  4. Do not miss a surgical or correctable condition. Just because the patient already had surgery, or was not previously a candidate for it, does not mean the patient cannot develop a new indication for a directed procedure. While they may have arrived with primarily “medical” bleeding, a sudden increase in drain or tube output may indicate a new focal hemorrhage requiring surgical repair. IR-directed therapy may also become appropriate and necessary. Pneumothorax can always occur, as can cardiac tamponade; both are easily correctable and should not fall off your shock differential even while you treat for hemorrhage. Bedside ultrasound and the occasional portable xray can supplement your eyes and brain, and the surgical team—far more familiar with the patient’s anatomy than you—should be apprised of the patient’s status whenever in doubt.


Trauma resuscitation in the ICU is an engaging and sometimes exhausting activity. Its principles are simple, and can be applied in many non-traumatic patients in the surgical or cardiac surgery ICUs, as well as the occasional medical patient with massive GI or similar bleeding. Just remember:

Replace the blood
  1. Use primarily blood products
  2. Transfuse them in a balanced ratio
  3. Warm them up and give them fast
  4. Use a massive transfusion protocol
  5. Transfuse enough, but no more
  6. Consider autotransfusion
Stop the bleeding
  1. Maintain normothermia
  2. Replete calcium
  3. Prevent acidosis
  4. Address specific deficiencies in clotting
And keep your eyes peeled for anything else
  1. Send regular labs
  2. Attend to the airway and breathing
  3. Keep an eye on the heart
  4. Do not miss a surgical or correctable condition