13.1 Anemia, Transfusion, Thrombocytopenia, Coagulopathy, and DIC (01.T.1–4)

Hemoglobin, platelets, and clotting factors are how an injured brain either receives oxygen or receives another hematoma. A 7 g/dL red-cell trigger that is excellent after pneumonia is not automatically the right trigger after aneurysmal subarachnoid hemorrhage (SAH). This independent OpenExamPrep section covers anemia and transfusion, thrombocytopenia, coagulopathy, and disseminated intravascular coagulation (DIC) as listed under hematologic disorders in the ABPN Content Specifications. It is not an ABIM or ABPN product. Drug-specific reversal of warfarin, dabigatran, and factor Xa inhibitors is taught with reperfusion pharmacology; the pointers here are only enough to keep those protocols from being mixed with DIC replacement or with antiplatelet-associated intracerebral hemorrhage (ICH).

Why transfusion items are written as brain versus body

The Transfusion Requirements in Critical Care (TRICC) trial (Hébert and colleagues, NEJM 1999) randomized euvolemic, non-bleeding medical and surgical ICU adults to a restrictive strategy (transfuse when hemoglobin fell below 7 g/dL, maintain 7–9 g/dL) or a liberal strategy (transfuse below 10 g/dL, maintain 10–12 g/dL). Thirty-day mortality was 18.7% versus 23.3%. The restrictive arm was at least as safe and was better in less acutely ill patients. Later ICU trials (TRISS and others) and the 2023 Association for the Advancement of Blood & Biotherapies (AABB) red-cell guideline kept a 7 g/dL threshold for most hemodynamically stable hospitalized adults, including general critical illness.

TRICC enrolled almost no patients whose oxygen delivery problem was a swollen, vasospastic, or ischemic brain. Cerebral oxygen delivery is hemoglobin × saturation × flow. After traumatic brain injury (TBI), SAH, or ICH, flow is already constrained by intracranial pressure (ICP), impaired autoregulation, or delayed cerebral ischemia (DCI). Observational series linked anemia to worse neurologic outcome, but those series could not say whether transfusion helps or merely marks severity.

Two 2024 randomized trials now sit on the examination table. TRAIN (Taccone and colleagues, JAMA 2024) randomized 850 adults with TBI, aneurysmal SAH, or ICH, hemoglobin below 9 g/dL in the first 10 days, and expected ICU stay of at least 72 hours, to a liberal trigger below 9 g/dL or a restrictive trigger below 7 g/dL. At 180 days, an unfavorable Glasgow Outcome Scale–Extended (GOSE) score of 1–5 occurred in 62.6% of the liberal group versus 72.6% of the restrictive group (adjusted relative risk 0.86; P = 0.002). Cerebral ischemic events were 8.8% versus 13.5%. HEMOTION (Turgeon and colleagues, NEJM 2024) randomized 742 adults with moderate or severe TBI and anemia to transfusion at hemoglobin ≤10 g/dL versus ≤7 g/dL. Unfavorable 6-month GOSE using a prognosis-based sliding dichotomy occurred in 68.4% versus 73.5% (adjusted absolute difference 5.4 percentage points; 95% confidence interval −2.9 to 13.7). The primary endpoint was not met. Acute respiratory distress syndrome was more common with the liberal strategy (3.3% versus 0.8%). Venous thromboembolism rates were 8.4% in each arm.

The usable examination sentence is therefore two-part. After most ICU illness — sepsis, pneumonia, gastrointestinal bleeding once hemostasis is achieved (Villanueva, NEJM 2013, used 7 versus 9 g/dL in upper-tract bleeding) — a restrictive 7 g/dL trigger remains the default. After TBI, SAH, and ICH, do not recite TRICC as if it settled the brain. TRAIN supports considering a higher (9 g/dL) threshold in acute brain injury. HEMOTION shows that pushing all the way to 10 g/dL in TBI is not a proven neurologic win and may cost lung injury. Individualize when brain-tissue oxygen is low, DCI is active, or active coronary ischemia coexists. Transfuse one unit at a time and recheck hemoglobin.

Thrombocytopenia: EVD, craniotomy, and the PATCH exception

Prophylactic platelet transfusion for hypoproliferative thrombocytopenia (chemotherapy, aplastic marrow) is typically given when the count falls below 10 × 10^9/L in a non-bleeding patient, with a higher floor if fever, mucositis, or a falling trend is present. That AABB-style number is not the neurosurgical number.

For craniotomy and, in most shops, external ventricular drain (EVD) placement, the customary target is a platelet count near 100 × 10^9/L. That threshold is expert practice, not a randomized cutoff. Retrospective series associate counts below 100 × 10^9/L with more tract hemorrhage, but operator passes, technique, and coexisting coagulopathy confound the association. If herniation is minutes away, place the drain while platelets infuse rather than waiting for a recount. For lumbar puncture in a non-neurosurgical context, many services use 50 × 10^9/L; do not import that number into EVD consent.

Antiplatelet-associated ICH is a different problem: the count is often normal and the defect is function. The PATCH trial (Baharoglu and colleagues, Lancet 2016) randomized adults with spontaneous supratentorial ICH on antiplatelet therapy, within 6 hours of onset, Glasgow Coma Scale 8–15, and no planned surgery, to platelet transfusion versus standard care. Transfusion increased the odds of death or dependence at 3 months (adjusted common odds ratio 2.05) and increased serious adverse events. The examination implication is blunt: do not routinely transfuse platelets for nonoperative ICH on aspirin, clopidogrel, or dipyridamole. Desmopressin (0.3–0.4 micrograms/kg IV) is used in some hemorrhage bundles as a platelet-function adjunct; it is not a PATCH-tested substitute and it is not a reason to ignore PATCH.

If the same patient needs emergent craniotomy or EVD, PATCH does not apply. Many neurosurgeons still request platelets for recent P2Y12 inhibition (clopidogrel, ticagrelor, prasugrel) because the surgical field cannot wait for a P2Y12 reaction unit assay. That is a procedure decision, not a medical ICH decision. Document the distinction on the examination: count versus function, operating versus not operating.

Other thrombocytopenia patterns in the neuro ICU include heparin-induced thrombocytopenia (HIT, next section), thrombotic thrombocytopenic purpura (TTP, section 13.3), drug-induced immune thrombocytopenia (beta-lactams, vancomycin, glycoprotein IIb/IIIa inhibitors), post-cardiopulmonary bypass consumption, and hypersplenism in cirrhosis. A falling count plus thrombosis is HIT or TTP until proven otherwise. A falling count plus oozing plus a rising D-dimer is DIC until proven otherwise.

Coagulopathy: DIC versus dilution versus liver

DIC is thrombin storm with secondary fibrinolysis. An underlying driver is required: sepsis, trauma, obstetric catastrophe, acute promyelocytic leukemia, or sometimes brain tissue thromboplastin after severe TBI or craniotomy. The laboratory pattern is falling platelets, falling fibrinogen, rising D-dimer or fibrin degradation products, and a prolonging prothrombin time (PT) / international normalized ratio (INR). Factor VIII is consumed. Schistocytes may be few. The International Society on Thrombosis and Haemostasis (ISTH) overt DIC score (platelet count, fibrin marker, PT prolongation, fibrinogen) of 5 or more, in a patient with a known driver, supports overt DIC. Treat the driver first. Replace what is missing if the patient is bleeding or about to undergo a neuraxial procedure: cryoprecipitate or fibrinogen concentrate when fibrinogen is below about 100–150 mg/dL (many trauma protocols use 150–200 mg/dL), platelets for bleeding with counts below about 50 × 10^9/L, and plasma for bleeding with a substantially prolonged PT. Heparin for chronic DIC is a hematology consult, not a brain-bleed reflex.

Liver failure is synthetic failure plus portal sequestration. Factors II, V, VII, IX, X, and fibrinogen fall. Factor VIII is usually normal or high because extrahepatic endothelium still makes it and it behaves as an acute-phase protein. INR is a poor predictor of bleeding in cirrhosis; viscoelastic testing (TEG or ROTEM) often shows adequate clot strength despite a terrifying INR. Vitamin K 10 mg IV is reasonable when deficiency is possible (cholestasis, poor intake, antibiotics). Plasma “to correct the INR” before an EVD, without bleeding and without a viscoelastic or factor-based indication, floods the patient with volume and still may not change the INR.

Dilutional coagulopathy follows massive crystalloid or packed red-cell resuscitation without plasma, platelets, or fibrinogen. Fibrinogen is the first factor to hit a hemostatic floor because its plasma concentration is low to begin with. D-dimer is not sky-high the way it is in DIC unless trauma fibrinolysis coexists. The history is the diagnosis: liters in, factors out.

Massive transfusion and citrate hypocalcemia

Definitions of massive transfusion vary (10 red-cell units in 24 hours, or 3–4 units in about 1 hour, or replacement of a circulating blood volume). In practice the neurointensivist meets it after polytrauma, scalp and sinus arterial bleeding, or a ruptured aneurysm that has not yet been secured. The PROPPR trial (Holcomb and colleagues, JAMA 2015) compared plasma:platelet:red-cell ratios of 1:1:1 versus 1:1:2 in trauma. The 1:1:1 ratio achieved hemostasis more often and reduced death from exsanguination at 24 hours; 24-hour and 30-day all-cause mortality were not significantly different. Damage-control resuscitation still means limit crystalloid, keep the patient warm, correct acidosis, and give products in a balanced ratio rather than red cells alone.

Packed red cells, plasma, and platelets are anticoagulated with citrate. The liver metabolizes citrate to bicarbonate. When transfusion is fast, when the liver is ischemic or cirrhotic, or when the patient is hypothermic, citrate binds ionized calcium. Ionized hypocalcemia presents as hypotension that does not make sense for the product volume, a prolonged QT, coagulopathy (calcium is a clotting cofactor), and sometimes tetany. Check ionized calcium frequently during a massive transfusion protocol and replete with intravenous calcium chloride or calcium gluconate. Citrate also binds magnesium. Stored red cells leak potassium; irradiated or older units can cause hyperkalemia, especially through a central line into a small circulation.

Other transfusion complications the examination still uses: transfusion-associated circulatory overload (TACO) (hypertension, pulmonary edema, raised BNP, diuretic-responsive) versus transfusion-related acute lung injury (TRALI) (acute hypoxemic respiratory failure during or shortly after transfusion, often hypotensive, non-cardiogenic edema). Both can raise ICP by ruining oxygenation and, in TACO, by venous congestion. Stop the transfusion, support the airway, and distinguish the two so you do not diurese TRALI or ignore TACO.

Reversal pointers without repeating the reperfusion chapter

Hold the anticoagulant. For warfarin-associated ICH, give 4-factor prothrombin complex concentrate dosed to INR and weight plus vitamin K 10 mg IV; plasma is slower and volume-heavy. For dabigatran, idarucizumab 5 g IV. For oral factor Xa inhibitors, andexanet alfa is the specific reversal agent; ANNEXA-I (NEJM 2024) improved hematoma-expansion control versus usual care and increased thrombotic events, including ischemic stroke. Unfractionated heparin reverses with protamine. Those doses, labels, and ANNEXA-I percentages belong in the reversal chapter. The hematology trap is using those drugs to “treat DIC” or using PATCH-contraindicated platelet dumps as a substitute for specific reversal.

Exam traps

Do not apply TRICC’s 7 g/dL trigger as a moral law after TBI or SAH. Do not claim HEMOTION proved that 10 g/dL is required. Do not transfuse platelets for nonoperative antiplatelet ICH. Do not treat a cirrhotic INR with plasma the way you treat DIC. Do not forget ionized calcium when the massive transfusion refrigerator is open.