Hemostatic dysfunction and resultant pathologic bleeding and/or thrombosis is a common complication of critical illness in children. There are several important considerations of this dysfunction that are imperative for accurate diagnosis and clinical management in the pediatric population. First, the hemostatic system changes and develops from birth to infancy, childhood, adolescence, and ultimately adulthood. Understanding this evolution is important in the interpretation of physiologic and pathologic hemostasis. Second, traditional tests of hemostasis have focused on platelet count and plasma-based measurements of clot formation. There are limitations in this traditional approach including lack of consideration for (1) platelet function and activity, (2) the imperfect sensitivity of plasma-based tests, particularly to abnormalities in fibrinolysis and conditions of enhanced clot formation, (3) the critical role of the endothelium in hemostasis and crosstalk between the endothelial, inflammatory, and coagulation systems, and (4) the stages of clot formation and lysis over time. Newer studies address some of these challenges by directly measuring thrombin generation in plasma and assessing the stages of hemostasis in whole blood. Successful diagnosis and management of hemostatic dysfunction in critically ill children requires both an age-dependent understanding of clot formation and regulation of hemostasis as well as of the strengths and weaknesses of the tests employed.

Hemostasis is the normal physiologic process by which blood is maintained in fluid state while also allowing for blood clot formation at the site of injury maintaining the integrity of the closed circulatory system after vascular damage (1). It is a highly complex series of interwoven processes that result in a rapid, localized, and highly regulated response. Key elements of hemostasis include (1) formation of the platelet plug (referred to as primary hemostasis), (2) soluble phase coagulation with propagation of clotting through the clotting cascade, (3) termination of clot formation, and (4) clot dissolution (fibrinolysis). Under physiologic conditions, a clot is formed at the site of injury to stop bleeding locally with clot lysis and tissue remodeling to follow (1). Abnormalities in any part of these processes can result in dysfunctional hemostasis manifest as abnormal bleeding or thrombosis.

The hemostatic process starts with injury to the vascular endothelium (2). The undisturbed endothelium has multiple guards against undesired coagulation by functioning to counteract platelet activation and aggregation, and to maintain blood fluidity (2). When the integrity of the endothelium is compromised, exposure of subendothelial elements trigger thrombus formation (3). The endothelium also responds to injury through vasoconstriction which results from impairment of the vasodilatory action of damaged endothelial cells as well as direct access of smooth muscle cells to locally generated vasoconstrictive agents (3).

Platelet activation and adhesion upon exposure to vascular injury occurs in a highly coordinated method involving tethering, rolling, activation, and firm adhesion (4, 5). The initial interactions between platelets and the extracellular matrix are highly dictated by local rheological conditions (4). Platelet activation is triggered by several highly adhesive macromolecules including collagen and von Willebrand factor (vWf) along with the weaker agonists adenosine diphosphate (ADP) and epinephrine (4–6). This activation is also triggered by thrombin (4, 5). As part of the platelet activation process, platelets undergo shape change and develop elongated pseudopods that aid in the subsequent adhesion process (7). Platelet adhesion is primarily mediated by attachment of the platelet surface receptor GPIb/IX/V complex to vWf in plasma and the subendothelial matrix (7). Further, platelet activation results in exposure and conformational change of the GPIIb/IIIa receptor on the platelet surface. The GPIIb/IIIa receptor then binds vWf and fibrinogen ultimately resulting in platelet-platelet cohesion (8, 9).

After platelet binding to subendothelial structures and subsequent signaling to the platelet cytoplasm, platelet granules fuse with the open canalicular system of the platelet membrane and empty their contents into the local environment. The paracrine and autocrine nature of these bioactive contents causes increased activation of nearby platelets in both number and degree (4). This results in secondary secretion and significant amplification of the platelet activation and adhesion processes. Alpha (α)granules contain a heterogeneous complement of proteins that effect multiple biologic systems beyond primary hemostasis and coagulation including inflammation, angiogenesis, wound healing, and others (4). Important to formation of the platelet plug is the α-granule release of vWf which further increases formation of the platelet scaffold via GPIIb/IIIa and GPIb/IX/V as well as release of fibrinogen which crosslinks with GPIIb/IIIa and provides an additional source of fibrinogen to what is present in the plasma (4, 10). Dense (δ)-granules secrete ADP and serotonin which stimulate and recruit additional platelets (11). Finally, platelet procoagulant activity is an important facet of platelet activation. This procoagulant activity includes both exposure of procoagulant phospholipids such as phosphatidylserine and the ensuing formation of enzyme complexes on the platelet surface essential to the clotting cascade (12).

The soluble phase of clotting primarily involves the clotting cascade (Figures 1, 2) (13). In general, the clotting cascade involves successive activation of a series of proenzymes or zymogens into active enzymes resulting in an incremental but amplified clotting response. The clotting cascade is classically described by the intrinsic, extrinsic, and common pathway, though further investigation has revealed these pathways to be much more complex and interrelated (Figure 1) (13).

The clotting cascade is initiated through the extrinsic pathway and starts with exposure of tissue factor (TF) at the site of injury. Blood is exposed to TF either directly by the subendothelial matrix or by cytokine-induced expression on endothelial cells or activated monocytes (14, 15). Platelets may also generate their own TF in proximity to the injured vessel (16). TF operates as a cofactor in the activation of factor VII and together these two factors form the extrinsic tenase multimeric complex which goes on to activate factors X and IX to factors Xa and IXa, respectively (17, 18).

The intrinsic or contact activation pathway is initiated through the interaction of negatively charged surfaces resulting in the activation of factor XII, high-molecular-weight kininogen, and plasma kallikrein among others (13). Activated factor XII in conjunction with high-molecular-weight kininogen activate factor XI which then activates factor IX (19). Activated factor IX forms a multimeric complex with activated factor VIII, which has been activated by factor X and thrombin produced in the extrinsic pathway. This multimeric complex, referred to as intrinsic tenase, subsequently activates sufficient factor X for clot formation (13, 19). This process is amplified because (1) factor VIII is activated by both activated factor X and thrombin and (2) activated factor IX is further activated by the thrombin-induced activation of factor XI (19, 20). As a result, there is a progressive increase in factor VIII and factor IX activation as factor Xa and thrombin are formed. Through these mechanisms, sustained and amplified generation of thrombin is achieved through the intrinsic pathway. Of note, while more activated factor X is generated through the intrinsic pathway due to amplifying steps in the cascade, the extrinsic pathway is physiologically more important clinically (21).

From there, both the extrinsic and intrinsic pathway proceed to the common pathway. Factor V, which is released from platelet α-granules, is cleaved by thrombin to form activated factor V (22, 23). Activated factors X and V bind on the platelet phospholipid surface to form the prothrombinase complex which converts prothrombin (factor II) to thrombin (activated factor II, factor IIa) (24). Thrombin then converts fibrinogen to fibrin and activated factor XIII crosslinks overlapping fibrin stands which leads to stabilization of the clot (25).

Intrinsic to the built-in amplification systems of formation of the platelet plug and the coagulation cascade, hemostasis also requires highly regulated mechanisms of control of clot extension and termination of clot formation (Figure 2). Systemic control of this localized response is modulated through several mechanisms including (1) dilution of procoagulants in the bloodstream, (2) elimination of activated factors through the reticuloendothelial system, and (3) control through antithrombotic pathways anchored on vascular endothelial cells (1). Physiologic inhibitors of coagulation include tissue factor pathway inhibitor (TFPI) which inhibits TF-mediated and factor VIIa-mediated factor X activation, and C1 esterase inhibitor which inhibits activated factor XII, plasma kallikrein, and several complement proteases (26, 27).

Cessation of clot formation is critical in maintaining a targeted response and mediating the extent of the clot. Clot termination involves two key components, antithrombin (antithrombin-III, AT) and the protein C pathway. AT is a serine protease inhibitor that neutralizes many enzymes in the clotting cascade by irreversibly binding to them (28). Its function is enhanced by endogenous heparins and heparan sulfate, which induce a conformational change and increase its affinity for thrombin approximately 300-fold (28–30). Vascular endothelial cells are coated with activated AT and therefore equipped to rapidly inactivate any excess generated thrombin (29, 30). The protein C pathway is initiated through the binding of thrombin to the endothelial membrane-expressed thrombomodulin which induces the ability of the thrombomodulin-thrombin complex to activate protein C (31, 32). In association with protein S, activated protein C then proteolytically inactivates activated factors V and VIII (33, 34).

Also important to the limitation of clot formation is the regulation of vascular and platelet reactivity which is primarily modulated through prostacyclin, thromboxane, and nitric oxide. Specifically, undisturbed endothelial cells adjacent to endothelial injury release arachidonic acid which is subsequently converted to thromboxane A2 by cyclooxygenase-1 in platelets and prostacyclin via cyclooxygenase-1 on the endothelium (35). Prostacyclin blocks platelet aggregation and counteracts thromboxane A2-mediated vasoconstriction (36). In addition to its vasodilatory effects, nitric oxide inhibits platelet adhesion and aggregation (37). In fact, platelets may enhance their synthesis of nitric oxide in the setting of platelet adhesion to collagen providing negative feedback to limit excessive platelet adhesion and vasoconstriction at the site of injury (38).

Clot removal following hemostasis is achieved through fibrinolysis. Central to this process, plasmin is formed from the cleavage of plasminogen by bound fibrin and tissue-type plasminogen activator (38, 39). Urokinase is a secondary plasminogen activator, primarily acting in the extravascular compartment (39). Once formed, plasmin cleaves fibrinogen, fibrin, activated factor XIII, as well as a variety of plasma proteins and other clotting factors (39–41). Fibrinolytic activity can be generated either on the surface of the fibrin-containing thrombus or on cells that express profibrinolytic receptors (38). Plasmin activity is regulated by vascular endothelial cells that secrete both serine protease plasminogen activators and plasminogen activator inhibitors (42). Fibrinolysis works in conjunction with wound healing and tissue remodeling to restore vessel patency following injury.

The above discussion focuses primarily on plasma proteins with only limited mention of the role cellular components play in hemostasis. Indeed, a major limitation of plasma-based coagulation tests is that they do not take into consideration the myriad roles platelets, endothelial and leukocytes play in this process (43). Platelets are recruited to areas of vascular injury and adhere to endothelial cells and subendothelial structures (e.g., collagen) via specific receptors on these structures, a process augmented through binding to von Willebrand factor (vWf), thereby forming a platelet plug (Figure 3A). As noted, endothelial cells are critical in modulating the balance between activation of coagulation and fibrinolysis mediated through the binding of thrombin to thrombomodulin (Figure 3B). Additionally, platelet-neutrophil interactions are important in the formation of Neutrophil Extracellular Traps (NETs) and the process of NETosis which can induce a process referred to as immunothrombosis (Figure 4) (44, 45).

Developmental hemostasis describes the physiologic changes that occur with an increase in age. These changes are particularly relevant in the pediatric population from birth through infancy when plasma levels of several important procoagulant and anticoagulant factors rapidly increase or decrease to normal adult levels. Because of the dynamic development of the hemostatic system, an understanding of age-specific physiology and quantification of normal reference ranges for coagulation parameters is of particular importance in pediatric medicine.

Early study of developmental hemostasis included the establishment of reference ranges for basic global measures of coagulation in infants using cord blood (46, 47). Andrew et al. were the first to define the field of developmental hemostasis in a series of articles describing reference values for coagulation parameters in (1) healthy term infants from birth to 6 months of age, (2) healthy preterm infants from birth to 6 months of age, and (3) healthy children and adolescents between 1 and 16 years of age (Table 1) (48–50). This work defined normal values of global measures of coagulation, individual factors that facilitate clot formation, and inhibitors of coagulation (Table 2).

Overall, Andrew et al. found that coagulation tests varied with age and different parameters showed different patterns of maturation (Tables 1, 2) (48–50). At birth, plasma levels of many important coagulation factors were found to be around half of that found in adults. Full maturation to adult levels of these factors varied between a few months to 16 years of age (48–50). While there were continued increases in some factors through later childhood, most of these factors reached near normal adult levels by 6 months of age. Specifically, in both term and preterm infants the mean values for the vitamin K-dependent factors, contact factors, and selected inhibitors including antithrombin (AT), protein C and protein S were well-below adult values. In contrast, fibrinogen, factors V, VIII, and XIII, vWf and the inhibitors α₁-antitrypsin, α₂-antiplasmin, α₂-macroglobulin, and C₁ esterase inhibitor were noted to be at normal or somewhat supra-normal adult levels at birth. Globally, both prothrombin time (PT) and activated partial thromboplastin time (aPTT) were prolonged in neonates. While there was wide variation in normal PT in infancy, mean PT normalized to adult levels by 1 month of age whereas aPTT normalized by 6 months of age (48, 49). Differences between preterm and term infants tend to be present but relatively small (48, 49). By 6 months of age both preterm and full term infants exhibit equivalent levels of all but four components of the coagulation system, namely factor VIII (mean 0.99 in preterm compared to 0.73 in term), plasminogen (mean 2.75 in preterm compared to 3.01 in term), antithrombin (mean 0.9 in preterm compared to 1.04 in term), and heparin cofactor II (mean 0.89 in preterm compared to 1.2 in term) (49).

Subsequent studies went on to confirm the highly selective pattern of maturation of the coagulation system described by Andrew et al. using different analyzer and reagent combinations (Table 3) (51–55). Compared to Andrew et al. these studies report the same trends by age and value relative to adult values but establish different reference intervals due to use of different equipment and reagents instead of the manual methods employed by Andrew (51–55). In addition to updated analysis techniques, several studies have expanded the panel hemostatic markers (Table 3) (51–53). Flanders et al. described differences by sex in protein S levels with females having higher than adult values at all ages and males having higher than adult values only in the 16–17 year old age group (101–103% in females 7–17 years old vs. 98% in female adults; 116% in males 16–17 years old vs. 108% in male adults) (51).

Monagle et al. described elevated D-dimer, reduced TFPI, and reduced endogenous thrombin potential in healthy neonates and children compared to established adult values (52). Appel et al. described variation in development of vWf and factor VIII across blood groups, specifically (1) a higher median vWf antigen in non-O compared to O groups at 1–5 years old (97 vs. 71%) and >19 years old (118 vs. 99%), (2) a higher median ristocetin cofactor activity in non-O compared to O groups at 1–5 years old (82 vs. 66%) and >19 years old (106 vs. 82%), and (3) a higher median factor VIII in non-O compared to O groups at 1–5 years old (121 vs. 100%) and >19 years old (131 vs. 116%) (53). Attard et al. combined the results of several cohorts of infants at different ages and found maturation patterns that mirrored prior studies along with a high level of interindividual variability in protein levels (54). Finally, Toulon et al. looked across seven centers and found values that were not significantly different between centers when using the same analysis techniques and reagents (55). In contrast to other reports in which primarily activity of clotting proteins is measured, Attard tracks changes in antigenic levels of hemostatic factors over age (54). It is assumed that the relationship between antigen and activity is (relatively) fixed and that a decrease in antigen implies a commensurate decrease in activity with clinical decisions being made accordingly. However, one cannot rule out the possibility that changes in antigenic level or protein activity noted with age reflect an element of post-translational protein modification.

Taken together, these studies show that most coagulation parameters are highly dependent on age and undergo the greatest changes during the 1st year of life. These changes require the use of age-specific reference intervals in the clinical assessment of hemostasis and the diagnosis of pediatric bleeding and thrombotic disorders. Despite these age-dependent differences, the changing coagulation system throughout infancy, childhood, and adolescence should be viewed as physiologic, i.e., “balanced,” with decreased factor levels and prolonged PT and aPTT in infants compared to normal adolescents and adults not necessarily reflecting an increased risk for hemorrhage. In fact, the pediatric hemostatic system may favor a reduced risk of thrombosis without an increased risk of bleeding (52, 56). While generation of thrombin in infants has been shown to be less than that in adults, this is counterbalanced by the reduction in AT and the fact that that the major physiologic inhibitor of thrombin in infants, α₂-macroglobulin, is less efficient than is AT. However, other studies have shown that thrombin generation (endogenous thrombin potential; ETP) is “normal” in infants if measured in the presence of added thrombomodulin and that thrombin generation in very pre-term newborns is not different than that in near term infants (57, 58). That infants exhibit a “balanced” hemostatic system is supported by clinical findings such as decreased thromboembolic disease following surgery or prolonged immobilization in children compared to adults with similar risk factors (52, 59, 60). That said, the molecular basis underlying age-related numerical changes in coagulation measures as well as the clinical manifestation of these changes is not fully understood. Animal studies have demonstrated structural differences in coagulation proteins in newborn compared to adult models but this work has yet to be translated to humans (61). Further translational study is necessary to elucidate the important connections between coagulation parameters and clinical thrombotic and bleeding risk.

Evaluation of abnormalities in pediatric hemostasis requires (1) knowledge of different hemostatic testing options, (2) an understanding of the strengths and weaknesses of each test, and (3) an awareness of utility in the pediatric population, particularly in the context of the developmental hemostasis consideration discussed above. The following are key tests of pediatric hemostasis with specific focus on platelet evaluation, clotting assays, and global measures. Logistics of testing as well as strengths and weaknesses are summarized in Table 4.

There are a number of additional considerations in the assessment of hemostasis in the pediatric population. First, there are numerous technical challenges. This includes practical difficulty drawing sufficient blood volume from small infants and children given their small vessels for certain tests (e.g., LTA) and the overall smaller total blood volume infants and children which limits the quantity of blood that can safely be obtained to perform comprehensive diagnostic testing. Consequently, tests that use small volumes may be preferred and careful consideration is necessary to parsimoniously select appropriate tests given the clinical situation. While central lines help with the technical difficulties of obtaining adequate testing volumes, samples can be contaminated by anticoagulation or other medications running through the line itself. In fact, samples collected through central lines may require 10 mL of blood be withdrawn prior to sample collection for the result to be considered valid. Techniques to mitigate this waste include using the initial sample collection for other labs or reinfusing blood that would otherwise be discarded.

There are also a number of infant and pediatric conditions and diseases that require special testing considerations. Pediatric trauma patients, particularly infants, demonstrate significant endothelial injury, present with acute traumatic coagulopathy and/or coagulopathy related to iatrogenic causes, and, while they may be at less risk of bleeding than their adult counterparts, they are also at risk of thrombosis during recovery from their injury (120–123). While there have been several pediatric-focused studies around damage control resuscitation and thromboelastographic-guided transfusion, there are still gaps in how these tests account for the developmental changes in hemostasis noted in infants and children (99, 116). Pediatric patients with liver dysfunction may have significant hemostatic abnormalities from impaired synthetic function including coagulation factor defects, thrombocytopenia and platelet dysfunction, alterations in the fibrinolytic system, and alterations in endogenous inhibitors of coagulation. While clotting times may be significantly elevated in these patients, these abnormal results often do not correlate with bleeding risk (124, 125). Indeed, assessment of hemostatic balance in patients with liver disease generally demonstrates a neutral or mildly pro-thrombotic hemostatic system (126, 127). The concept of “hemostatic balance” has also been conceptually addressed as alternatively being a condition of a loss in “hemostatic reserve” by some hematologists, a concept that carries some merit (128). In the oncologic population, patients may be at risk of hyperviscosity, thrombosis, and cytopenias that may increase bleeding risk depending on the type of cancer and therapies. The hemostatic system is frequently monitored in these patients, but transfusion and other therapeutic decisions necessitate an understanding of the limitations and normal variations of these standard hematologic measurements (129).

Patients on extracorporeal life support present with unique hemostatic risks as both the extracorporeal circuit and the underlying disease process induce important dysregulation of the hemostatic system that may favor either hemorrhage or thrombosis (119, 120). Hematologic monitoring of these patients may be one of the most active parts of their management and typically includes frequent evaluation for bleeding, consumptive coagulopathy, hemolysis, and adequacy of anticoagulation. In addition to aPTT to assess anticoagulant effect of unfractionated heparin (UFH), anti-Xa assay (assessing effect of UFH or low molecular weight heparin [LMWH] or some of the new direct acting oral anticoagulants [DOACs]) or the Activated Clotting Time (ACT) are utilized to assess bleeding/thrombosis risks (130, 131) Increasingly, the anti-Xa assay is becoming the preferred test to monitor UFH as well as anti-Xa inhibitors due to its insensitivity to low levels of AT. However, therapeutic range for anti-Xa has been extrapolated from those for UFH by matching aPTT result to anti-Xa level in plasma samples to which heparin was added ex vivo. This practice does not account for the primarily anti-IIa action of UFH potentially introducing a degree of uncertainty regarding the extrapolated therapeutic threshold (132, 133). ACT is the preferred monitoring test for UFH given during cardiopulmonary bypass surgery with “therapeutic” (i.e., desired) values being surgeon and device specific (134, 135).

Finally, trauma patients often present in the field or to community hospitals and must be transported to the regional trauma center for care. The use of point of care (POC) tests of hemostasis have only recently been investigated in this patient population (136, 137). While some studies suggest the feasibility of these devices, there are as of yet insufficient data to adequately assess their use.

Understanding of hemostatic dysfunction remains a unique challenge in critically ill infants and children. It requires knowledge of physiologic hemostatic pathways, comprehension of the development of the hemostatic system by age, an understanding of logistics, advantages, and limitations in hematologic testing, and an awareness of the specifics of application to pediatric populations. Taken together, this knowledge will enable the practitioner to more accurately assess hemostatic dysfunction and determine bleeding and thrombosis risk resulting in more successful diagnosis and management of this vulnerable population.

AN drafted all tables included. RP created all figures included in manuscript. Both authors participated in reviewing appropriate published literature, writing all drafts, and the final submitted draft manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.