Hematologic Issues (Critical Care Manual)


Transfusion Criteria

A study of 838 critically ill, euvolemic ICU patients showed that there were no differences in 30 or 60-day mortality when treated by restrictive (hemoglobin 7.0 – 9.0 g/dL) or liberal (hemoglobin 10.0 – 12.0) transfusion criteria. The only statistically significant differences were 30-day mortality in patients with an APACHE score < 20 (8.7% in restrictives, 16.1% in liberal, p=0.03) and those < 55 years of age (5.7% in restrictives and 13.0% in liberal, p=0.02), but not among patients with clinically significant cardiac disease (20.5 percent and 22.9 percent, respectively; P=0.69). There were no significant differences in 30-day mortality between treatment groups in the subgroup of patients with a primary or secondary diagnosis of cardiac disease, in the subgroup of patients with severe infections and septic shock, or in the subgroup of patients with trauma. This cardiac data runs counter to other data, such as Lancet 348: 1055, 1996 which was a retrospective study of a cohort who refused blood products prior to cardiac surgery and also [Am J Respir Crit Care Med 155: 1618, 1997] which showed a statistically insignificant trend in cardiac patients [NEJM 340: 409, 1999]

For a more extensive discussion on transfusion criteria, see Blood Component Therapy

Blood Products in Trauma [from Andrews]

Platelet transfusions should be initiated (4-8 pack) if counts are < 50,000 or the patient has received more than 1.5 times their volume of infused blood. Cryoprecipitate (has higher fibrinogen than FFP) should be given for fibrinogen < 100 mg/dL. Brain injury leads to tissue thromboplastin release, which can cause DIC (elevated PTT and INR, fibrinogen < 50 mg/dL, platelets < 50,000, increased fibrin split product). Also note that hypothermia (in particular T < 20C) adversely affects all aspects of coagulation.


Often advocated in the SAH patients to improve rheology – 30-35% is usually the quoted target. However, blood flow studies in healthy humans show that maximal oxygen delivery occurs at a hematocrit of 40-42% [J Cereb Blood Flow Metab 7: 259, 1987; Neurosurgery 9: 514, 1981]. This may not be valid in the injured brain, where autoregulation may be failing. Animal studies suggest that in an infarcting brain, a hematocrit of 30% minimizes focal ischemia [Neurosurgery 80: 469, 1994; J Neurol Sci 150: 115, 1997]. 30-35% may be a reasonable compromise [Andrews]

Venous Thromboembolism


Clinical risk factors include history of thromboembolism, major surgery, trauma, malignancy, age > 40, obesity, varicose veins, prolonged immobilization, dehydration, heart failure, nephrotic syndrome, stroke, myeloproliferative syndrome, Behcet’s disease, pregnancy/pueperium. Pharmacologic risk factors include OCPs and HRT. Also, 5% of the Caucasian population has a factor V Leiden mutation [Br J Anesth 93: 275, 2004]

TED stockings have been proven effective in both abdominal and neurosurgical operations [Arch Int Med 149: 679, 1989; 154: 67, 1994] but are not recommended as the sole prophylactic method in any situation. SCDs are generally considered to be more effective than TED, and are often used in neurosurgical patients [JAMA 268: 1727, 1992] and other patients where bleeding would be intolerable.

Low doses of heparin activate antithrombin III, inhibiting the conversion of prothrombin to thrombin without affecting the rest of the coagulation process. The usual dose is 5000U q8h. Two RCTs in patients with proximal DVT reported that IV UFH administered for 5 to 7 days is as effective as UFH administered for more prolonged periods, provided that it is followed by adequate long-term anticoagulant therapy [Lancet 2:1293, 1996; NEJM 322:1260, 1990]. In surgical cases, an initial dose should be given 2 hours pre-operatively and continued until the patient is ambulatory or for 7-10 days, whichever comes first [Marino]. Low dose UFH does not provide adequate anticoagulation for orthopaedics, trauma, or spine injury cases.

For patients who will need long-term anticoagulation, the currently recommended approach is to start heparin and vitamin K antagonists together at the time of diagnosis, and to discontinue heparin when the INR is stable and > 2.0; this usually occurs after 5 to 7 days of heparin therapy [Chest 126: 401, 2004]

LMWH can be dosed less frequently, has a lower incidence of bleeding, and a lower incidence of heparin-induced thrombocytopenia [Blood 79: 1, 1992; NEJM 332: 1330, 1995]. As of 1997 these advantages were theoretical but not borne out clinically. LMWH is superior to UFH for ortho (hip, knee), major trauma, and spinal cord patients [Chest 126S: 338S, 2004]. In patients who are receiving spinal anesthesia, the first dose of LMWH should wait until 12-24 hours after surgery in order to avoid hematoma [Chest 124S: 379S, 2003]. Lastly, LMWH should be renally-dosed, but dalteparin does not need to be.

Low-dose warfarin (10 mg PO initially, titrated to an INR 2-3) can be used as a replacement for low dose heparin but is more cumbersome and thus generally avoided. If used, start with 10 mg the night before (will not create a coagulopathy by the time of surgery), then 2.5 mg qHS for the first few nights, then draw labs.

Fondaparinux (Arixtra) is no more effective than LMWH [Chest 126S: 338S, 2004]. Its primary advantage is that it does not cause HIT. It’s major disadvantages are that it is contraindicated in renal failure (CrCl < 30) and in patients < 50 kg (increased risk of bleeding). It is the ideal drug in HIT patients, however.

Vena cava filters are indicated in documented iliofemoral vein thrombosis with a) contraindication to anticoagulation b) documented PE during full anticoagulation c) free-floating thrombus or d) high-risk condition for fatal PE or in the absence of iliofemoral thrombosis but when a) long term prophylaxis is necessary or b) there is a high risk for both thromboembolism and hemorrhage. The incidence of PE with a Greenfield filter is 5% [Arch Int Med 152: 1985, 1992; Radiology 216: 54, 2000]

Ann Intern Med 144: 390, 2006

Prospective cohort study to evaluate PE in 211 patients with COPD and exacerbation of unknown origin – modified the original Geneva protocol for this particular patient population (replaced surgery with malignancy).

Arch Intern Med 161: 92, 2001

Study of > 2000% patients in an emergency ward for risk of pulmonary embolism (ie not post-surgical patients).

Chest 119: 132S, 2001

Clinical risk factors in this meta-analysis included increasing age; prolonged immobility, stroke, or paralysis; previous VTE; cancer and its treatment; major surgery (particularly operations involving the abdomen, pelvis, and lower extremities); trauma (especially fractures of the pelvis, hip, or leg); obesity; varicose veins; cardiac dysfunction; indwelling central venous catheters; inflammatory bowel disease; nephrotic syndrome; and pregnancy or estrogen use.


Clinical examination in these patients is neither sensitive nor specific enough to be of diagnostic value. In fact, hypoxemia may be absent in 30% of patients with a PE [Arch Int Med 146: 1699, 1986]. Additionally, a normal A-a gradient does not exclude the possibility of a PE [Chest 107: 139, 1995]. Still, red flags to look for include calf pain, edema, venous distension, pain on dorsiflexion, dyspnea, pleuritic chest pain, cough, hemoptysis. The highest negative predictive values are normal D-dimer (92%), normal dead-space ventilation (92%), and absence of tachycardia (85%). Dyspnea and tachypnea are only 75%, and hypoxemia is only 70%. Unfortunately, plasma D-dimers have very little positive predictive value in the ICU – up to 80% of ICU patients have elevated D-dimer values in the absence of VTE [Crit Care Med 28: 414, 2000]. Their value as a negative predictor is much better, but of marginal utility because so few ICU patients have normal D-dimers. Dead space is useful in ER patients but has not been studied in ICU patients, most of whom have some dead space – it may be that changes in dead space are more useful, however this has not been studied.

Vascular ultrasound offers two modalities – in venous compression ultrasound, pressure from the transducer fails to compress the veins in cases of DVT. In Doppler mode, blood velocities are measured. “Duplex ultrasound” is the combination of both compression and Doppler ultrasound and has a sensitivity of 95 – 100% and a specificity of 97 – 100% for DVTs of the thigh [Emerg Med Clin NA 22: 775, 2004]. Remember, however, that up to 30% of patients with a PE will show no evidence of thrombus in the legs [Ann Int Med 98: 891, 1983]. Also note that for calf DVTs, ultrasound has a sensitivity of 33 – 70% [Clin Chest Med 24: 1, 2003]

D-dimer is highly sensitive but not specific. A negative ELISA has a 95% negative predictive value in low risk patients but latex agglutination/high-risk patients are not as accurate [JACC 40: 1475, 2002]

Patients with normal lungs should be V/Q scanned if suspicion is still high, and those with lung disease should get a spiral CT [Marino]. The problem with V/Q scanning is that very few ICU patients have normal lungs – in those that do, a normal V/Q scan excludes PE, and a high probability scan has a 90% PPV. Low-probability scans in conjunction with negative duplex are sufficient to stop the workup and follow patients clinically.

Spiral CT scans have a ~ 93% sensitivity and ~ 97% specificity for larger, proximal emboli and only ~ 70% for smaller, more peripheral, and clinically questionable emboli [AIM 160: 293, 2000]. Fortunately, there is no data that these peripheral emboli are dangerous or that stopping therapy based on a negative CT scan is in any way unsafe [JAMA 293: 2012, 2005]. Spiral CT scans are more sensitive for proximal emboli [Annals 132: 277, 2000] as compared to V/Q scanning- their only disadvantage is that they require that a patient can hold their breath for 30 seconds or more.


Heparin is the mainstay of therapy for DVT, and a goal PTT of 1.5 – 2.5 should be reached as soon as possible [Arch Int Med 152: 1589, 1992]. Weight-based dosing of heparin has been proven superior to standard dosing [Ann Int Med 119: 874, 1993]. UFH should be checked every 6 hours by monitoring PTTs. Warfarin can be started on the first day of therapy, and when INR > 2.0 the heparin can be stopped. Evidence suggests that 5 days of heparin is usually sufficient [NEJM 322: 1260, 1990]

Possible Superiority of SQ UFH for DVT Treatment

For the initial treatment of DVT, UFH can be administered SQ twice daily. The relative value of the IV and SQ administration of UFH has been evaluated in eight clinical studies and reviewed in a meta-analysis [Ann Intern Med 116: 279, 1992]. SQ UFH administered twice daily appeared to be more effective and at least as safe as IV UFH. When patients are receiving SQ heparin, the aPTT should be drawn at 6 h after the morning administration and the dose of UFH adjusted to achieve a 1.5 to 2.5 prolongation.

Equivalence of UFH and LMWH for DVT Treatment

The most recent meta-analysis [Arch Intern Med 160:181, 2000] included 13 randomized studies comparing IV heparin and LMWH for the initial treatment of acute DVT. There was no statistically significant difference in risk between LMWH and UFH for recurrent VTE, PE, and major bleeding. A statistically significant difference for risk of total mortality was observed in favor of LMWH (RR, 0.76; 95% CI, 0.59 to 0.90). The survival benefit was essentially accounted for by patients with malignancy. No apparent differences were observed in efficacy and safety among the different LMWHs. Because dosing of LMWH is easier (1 mg/kg SQ q12h) and it does not have to be checked via PTTs, LMWH is replacing UFH in the initial treatment of DVT [Marino]

Note that serum transaminase levels may increase artificially in up to 80% of heparinized patients [NEJM 325: 1585, 1991] and that 5 – 10% of patients on heparin may experience hyperkalemia due to heparin-induced aldosterone suppression, leading some to recommend checking potassium q3d in patients on heparin [Am J Med 98: 575, 1995]

LMWH as effective as UFH for treatment of PE

Meta-analyses of patients with DVT (with likely asymptomatic PE in a substantial proportion of these patients) have shown that LMWH treatment administered SC in doses adjusted to body weight only is at least as effective and safe for initial treatment as IV, dose-titrated UFH [Arch Intern Med 160: 181, 2000] Five studies [NEJM 337:657, 1997; Arch Intern Med 160: 229, 2000; NEJM 337: 663, 1997; Ann Intern Med 123: 1, 1995; Thromb Haemost 74: 1432, 1995] in patients presenting with symptomatic nonmassive PE or VTE confirmed these findings of a comparable safety and efficacy of LMWH administered SC. In patients with acute nonmassive PE, LMWH is recommended over UFH [Chest 126: 401, 2004]

Use of Aspirin as Alternative to Heparin/Enoxaparin

A small (275 patients) randomized controlled trial suggested that aspirin may be as efficacious as enoxaparin in patients undergoing total knee arthroplasty and who use the VenaFlow compression device [Westrich GH et al. J Arthroplasty 21 (6 Suppl 2): 139, 2006]. The purported equivalence of aspirin is refuted by data from air travel studies, which suggest that, as compared to heparin, aspirin is an inferior anticoagulant [Cesarone MR et al. Angiology 53: 1, 2002]. Aspirin clearly offers some efficacy, with the PEP trial (17,444 patients) showing a 36% reduction in VTE risk after orthopedic surgery (absolute risk of VTE/PE reduced from 2.5% to 1.6%) [PEP Trial (no authors). Lancet 355: 1295, 2000]

Thrombolysis for treatment of PE – usually not recommended

In a systematic review [Can Med Assoc J 146: 1317, 1992] of nine trials in patients with acute PE, thrombolytic therapy led to a more rapid resolution of the radiographic and hemodynamic abnormalities associated with acute PE than did anticoagulant therapy alone, although these benefits were short-lived. No difference was detected in clinically relevant outcomes such as the death rate or the resolution of symptoms between patients receiving thrombolytic therapy and those receiving anticoagulant therapy alone [Can Med Assoc J 146: 1317, 1992]. Patients with VTE who receive thrombolytic therapy have a 12% risk of major bleeding [Chest 121: 877, 2002] and a 1 to 2% risk of intracranial bleeding. Because of the favorable results with anticoagulants, thrombolytic therapy should usually be reserved for treatment of patients with acute massive embolism, who are in hemodynamically unstable condition and do not seem prone to bleeding [Chest 126: 401, 2004]. If used, thrombolytic therapy should always be followed by heparin or coumadin.

Neurosurgical Considerations

Ample literature suggests that the addition of heparin to the neurosurgical patient increases the risk of bleeding, oftentimes major and requiring surgery [Surg Neurol 50: 521, 1998; Surg Neurol 59: 363, 2003; Eur Spine J 13: 1, 2004]. A substantial review article from 1994 proposed that there is no data to suggest that the effects of SCD and TED are additive, that moderate risk patients should receive SCD/TED (or low-dose heparin if no SAH or ICH), and that high-risk patients should receive both [Neurosurgery 34: 280, 1994].

Interestingly, there is now a fledgling body of data suggesting that enoxaparin may be indicated in craniotomy patients. A multicenter, randomized, double-blind trial of > 300 neurosurgical patients (~ 80% of whom underwent craniotomy) given SQ enoxaparin (40 mg once daily for 7 days, starting < 24 hours post-op) vs. placebo showed that 17% of treatment and 32% of placebo had DVT (p=0.004). Two patients in the placebo group died of autopsy-confirmed PE. Major bleeding occurred in four patients in both groups [NEJM 339: 80, 1998]. An earlier multicenter, randomized, double-blind trial of neurosurgical patients given nadroparin, initiated postoperatively, compared to graduated compression stockings showed that 18.7% of nadroparin patients and 26.3% of control patients developed VTE (p = 0.047). The rates for proximal deep-vein thrombosis/pulmonary embolism were 6.9% and 11.5% (p = 0.065). The corresponding percentages for proximal deep-vein thrombosis/pulmonary embolism were 5.8% and 10.2% (p = 0.36). Major bleeding complications occurred in 2.5% of treated patients and 0.8% of control patients (p = 0.87). A higher mortality was observed in the treated group, however, none of these deaths was judged by a blinded adjudication committee to be related to the study drug [Thromb Haemost 75: 233, 1996]

Thromb Haemost 75: 233, 1996

Trial Multicenter, randomized, double-blind trial
Treatment Nadroparin, initiated postoperatively, vs. graduated compression stockings (exact time unknown)


  • VTE: 26.3% (control) vs. 18.7% (nadroparin, p = 0.047)
  • Proximal DVT/PE: 11.5% and 6.9% (p = 0.065)
  • Major bleeding complications: 0.8% (control) vs. 2.5% (nadroparin, p = 0.87)
  • A higher mortality was observed in the treated group, however, none of these deaths was judged by a blinded adjudication committee to be related to the study drug.

Neurosurgery 43: 1074, 1998

Where Randomized study of 68 craniotomy patients
Study SCD, enoxaparin (30 mg SQ), or both, initiated before induction (in the holding room)
Outcome No statistically significant difference in the incidence of post-operative DVT. 5 of 46 patients on LMWH suffered a clinically significant intracranial hemorrhage, thus the study was terminated early.

NEJM 339: 80, 1998

Where Multicenter, randomized, double-blind trial of > 300 neurosurgical (tumor) patients (~ 80% of whom underwent craniotomy)
Study SQ enoxaparin (40 mg once daily for 7 days, starting < 24 hours post-op and lasting 8 +/- 1 days) vs. placebo
Outcome – 32% of placebo vs. 17% of treatment had DVT (p= 0.004)
Two patients in the placebo group died of autopsy-confirmed PE. Major bleeding occurred in four patients in both groups.

Chest 122: 1933, 2002

Trial Randomized, prospective, double blind trial of 150 craniotomy patients (tumor)
Treatment 40 mg/d enoxaparin vs. heparin (5000 U bid) starting on the morning of the first post-operative day combined with TED/SCDs in all patients.


  • No patient developed symptomatic DVT or PE prior to discharge.
  • 9.3% developed asymptomatic VTE with no difference between the two groups
  • 3 patients in the group had post-operative bleeding, one of which was intraparenchymal

Surgical Neurology 59: 363, 2003

Trial Randomized, prospective trial of 100 craniotomy patients
Treatment SC heparin 5000 bid vs. dalteparin 2500 qday begun at induction of anesthesia and continued for 7 days or until ambulating


  • Two dalteparin patients developed DVT (one syptomatic) vs. no heparin patients.
  • Two hemorrhages occurred in dalteparin patients (non-operative) and one hemorrhage occurred in the heparinized patients (craniotomy required).
  • Two dalteparin patients had to stop medication secondary to thrombocytopenia.

Vena Cava Filters

A study of 400 patients with confirmed DVT (+/- PE) were randomized to heparin vs. lovenox (~ 200 in each group) and independently also randomized to filter vs. none (~ 200 patients per group). At day 12, IVC filters prevented PEs (1.1% vs. 4.8%, p = 0.03) but at 24 months this was statistically insignificant (6% vs. 12%, p = 0.16) and there was no difference in mortality. Permanent filters did increase the risk of recurrent DVT (21% vs. 11.6%, p = 0.02). There were no differences between enoxaparin and heparin [Decousus H et. al. NEJM 338: 409, 1998]. An 8 year follow up showed that permanent filters reduced PE (15% vs. 9%), caused DVT (36% vs. 28%, especially thrombosis of filter), had no effect on mortality or post-thrombotic syndrome [Ramacciotti E et. al. Circulation 112: 416, 2005]. A recent Cochrane Database Review concluded that there is currently insufficient evidence to make recommendations regarding permanent filters [Young T et. al. Cochrane Database Syst Rev (3): CD006212, 2007]

Temporary filters prevent early PE but have no effect on late DVT. Several registries suggest that very few are ever removed (35% removed [Am J Surg 190: 858, 2005], 31% removed [J Vasc Surg 40: 958, 2004], 13% removed [J Vasc Interv Radiol 16: 1189, 2005]). In a study of 32 patients in Japan, filters were placed in patients with residual proximal DVT and a transient risk factor and also pre-operative DVT patients without PE. 18 patients received permanent filters (group A) and 32 received temporary filters (group B). Mortality was 35% in the permanent and 16% in the temporary group (p = 0.14). PE occurred in 18% of the permanent group and 0% of the temporary group (p = 0.10) [Heart Vessels 21: 221, 2006]