Validation of thromboelastometry in horses


Prof. Saverio Paltrinieri, Dipartimento di Patologia Animale, Igiene e Sanità Pubblica Veterinaria, Via Celoria 10, 20133 Milano, Italy


Background: Thromboelastometry is used for identifying or monitoring coagulation abnormalities. It has been validated in several species but not in horses and the characteristics of the equine thromboelastogram have not yet been detailed.

Objectives: The purpose of this study was to validate a thromboelastometer to be used with equine blood and to define the normal equine thromboelastogram.

Methods: A Rotem-gamma thromboelastometer (Pentapharm GmbH, Munich, Germany) was used on 38 citrated blood samples to investigate native coagulation, the intrinsic and extrinsic pathways, the function of fibrinogen (largely dependent on its concentration), and the presence of fibrinolysis. Using classic validation approaches, we evaluated the imprecision of the method and the influence of hemolysis and storage time and temperature. The normal thromboelastogram was defined in both saddle and racing horses (the latter sampled before and after the race).

Results: For imprecision tests, the analytical variations were <10%. The equine thromboelastogram had a pattern similar to those of other species, but the intrinsic and extrinsic pathways were less and more efficient, respectively. Reference intervals in racing horses, especially after exercise, were different from those of saddle horses, most likely due to a higher RBC mass. Coagulability decreased in hemolyzed samples and significant changes were found between nonrefrigerated and refrigerated blood samples stored for 20 hours.

Conclusions: The Rotem-gamma thromboelastometer is a precise instrument for use with equine blood samples. The equine thromboelastogram is similar to that of other species, but reference intervals vary with aptitude and exercise. Hemolysis and refrigeration alter thromboelastometric results.


In equine medicine, several situations such as colic syndromes, primary coagulopathies, disseminated intravascular coagulation, poisoning (anticoagulants, insect, or snake bites), sepsis, and surgical complications are associated with coagulation abnormalities.1–9 Nevertheless, only a few studies on coagulation parameters in horses are available. The reports cited above, as well as other methodological studies on laboratory techniques, mainly investigated coagulation times, such as prothrombin time (PT) and activated partial thromboplastin time (aPTT), coagulation factors, and indicators of fibrinolysis such as D-dimers, fibrinogen degradation products, and fibrinogen.10–12 The latter is a major acute phase protein in horses and tends to reflect inflammatory states rather than altered coagulation.13 The availability of a method able to provide, in a short time and accurate manner, an overview of the different steps of the coagulation cascade in horses would be very useful.

Thromboelastometry assesses the efficiency of the whole coagulation process as well as the individual steps of the coagulation cascade.14 Moreover, contrary to traditional parameters of hemostasis, thromboelastometry also takes into account the function of both cellular and soluble components of coagulation and how they interact with each other in generating clots.14 Because of this, thromboelastometry has gained importance in human medicine and surgery.15,16 To date, information on the possible usefulness of thromboelastometry in veterinary medicine is scarce and mainly refers to animal models of human diseases, except for rare studies on canine thromboelastometry.17–20 In 1 report, the thromboelastogram of a horse with Glanzmann's thrombasthenia was compared with that of a healthy horse; the authors concluded that thromboelastography might be very useful in the equine species.21 We are unaware of other reports about the application of this technique in horses.

Before recommending extensive use of thromboelastometry in horses, it is important to assess the analytical performance of the technique in this species, especially because of peculiarities in equine coagulation values compared with other species (eg, shorter PT, longer aPTT, higher concentration of fibrinogen, higher activity of factors I, II, V, and X).8,22 The aim of this study was to define the analytical performance (imprecision, influence of storage time and temperature, influence of hemolysis) of a thromboelastometer in horses, to establish the characteristics of the normal equine thromboelastogram, and to develop reference intervals for horses at rest and postexercise.

Material and Methods

Animals and study design

This study was conducted on blood samples collected from 38 privately owned horses, with informed consensus of the clients, during a 6-month-period (September 2006–March 2007). Fifteen were Standardbred racing horses (12 mares, 2 stallions, and 1 gelding), 2–5 years of age (mean±SD, 2.8±0.94 years; median 3 years), and from a single stud. Twenty-three were Italian saddle horses (12 mares, 1 stallion, and 10 geldings), 2–23 years of age (13.3±6.5 years; median 15 years), and from a second stud. Horses were randomly selected; the only exclusion criterion was the presence of clinical signs or, for racing horses, poor performance during competition. To assess health status, all horses received a complete physical examination just before blood sampling. For racing horses, which were also sampled to evaluate the effect of exercise, the physical examination was repeated after the competition. Particular attention was paid to any possible clinical signs associated with alteration or activation of coagulation, including exercise-induced pulmonary hemorrhages (EIPH) in the racing horses. To exclude this disorder, the respiratory tracts of these horses were inspected by endoscopy immediately after the second sampling.

Blood samples were obtained from the jugular vein. To avoid unnecessary manipulation, which can result in activation of coagulation, the venipuncture site was not shaved. A 20-gauge needle was inserted into the vein after delicate disinfection of the hair. Blood (4.5 mL) was collected into nonsiliconized Vacutainer tubes (Venoject-Serum, code VT-050SU, Terumo Italia Srl, Rome, Italy) to obtain serum. To perform hematologic and thromboelastometric tests, an additional 4.5–13.5 mL of blood was then collected into glass Vacutainer tubes specific for coagulation tests, each containing 0.5-mL buffered sodium citrate (0.129 mol/L) with a maximum capacity of 4.5 mL (Venoject-Coagulation, code VT-050SBCS, Terumo).

Blood was transported to the laboratory within 2 hours. Except for samples used to study the influence of storage temperature, samples were transported at environmental temperature (approximately 20°C). To assess the effect of storage temperature, 1 tube was stored at 4°C just after sampling and kept refrigerated during transportation to the laboratory where it was maintained at 4°C except during the analysis.

A CBC was performed on each blood sample using an impedance counter already validated for equine blood (Hemat 8, SEAC, Calenzano, Florence, Italy). The CBC included RBC, WBC, and platelet counts; hemoglobin concentration; HCT; MCV; MCH; MCHC; red cell distribution width; mean platelet volume; plateletcrit; and platelet distribution width. On blood smears stained with May Grünwald–Giemsa, the differential leukocyte count was microscopically determined. Furthermore, blood smears were reviewed for platelet clumping to ensure accuracy of the platelet number determined by the analyzer. A platelet estimate was considered adequate when a minimum of 8 platelets per × 100 oil immersion field was observed.

Blood collected in tubes without anticoagulant was centrifuged, and serum was obtained, to perform a basic panel of biochemical tests using an automated analyzer (Cobas Mira, ABX Diagnostic, Montpellier, France) and reagents provided by Real Time Diagnostic (Florence, Italy). The panel included alanine aminotransferase (ALT), alkaline phosphatase (ALP), and γ-glutamyl transferase (GGT) activities, and total protein, albumin, calcium, inorganic phosphorus, urea, and creatinine concentrations. Selected samples were used in all or some of the validation procedures described below, depending on available blood volume.

Thromboelastometric analyses

Thromboelastometric analyses were performed on citrated whole blood using the Rotem-gamma thromboelastometer and reagents provided by the manufacturer (Pentapharm GmbH, Munich, Germany). Thromboelastometric tests were performed at 37°C according to the manufacturer's instructions and using the automated pipette included with the instrument, which for each test dispenses 300 μL of blood and 20 μL of activator or inhibitor in a cup. To assess “native” coagulation, which depends on both the intrinsic and extrinsic pathways and includes possible activation of the fibrinolytic pathway, blood was recalcified with the Startem reagent, which contains calcium chloride and allows normal coagulation to progress. To evaluate the intrinsic pathway, after recalcification as described above, coagulation is activated with the Intem reagent, which contains ellagic acid as a contact activator. To evaluate the extrinsic pathway, after recalcification as described above, coagulation is activated using the Extem reagent, which contains tissue factor. The function of fibrinogen within the extrinsic pathway is assessed using both the Extem and the Fibtem reagent; the latter contains cytochalasin D, which blocks the cytoskeleton and thus contraction of platelets. The resulting thromboelastogram is due to the function of fibrinogen alone, and correlates with fibrinogen concentration.14,16 Fibrinolysis is assessed using both the Extem and the Aptem reagents; the latter contains aprotinin, which blocks plasmin activity. The difference between thromboelastograms obtained by activation of the extrinsic pathway, with or without the addition of aprotinin, indicates the presence of fibrinolysis.

The following parameters of the thromboelastogram (Figure 1) were evaluated: coagulation time (CT), the time (in seconds) between activation of coagulation and formation of the first measurable clot, which evaluates the activity of plasma coagulation factors and corresponds to the time from the start of the test until an amplitude of 2 mm is reached; clot formation time (CFT), the time (in seconds) necessary to increase the elasticity of the clot from 2 to 20 mm, and which corresponds to the initial activation of platelets and fibrinogen; angle (α, expressed in degrees), which corresponds to the slope of the tangent on the elasticity curve and indicates a tendency towards hypo- or hypercoagulable conditions; and maximum clot firmness (MCF), the maximum amplitude (in mm) reached during the test, which corresponds to the maximum strength of the clot and depends on both platelet and fibrinogen activation in the presence of factor XIII (which stabilizes the clot). Two additional parameters were calculated on the basis of the first derivative generated by the instrument23: maximum velocity (Vmax), the maximum rate of clot formation (in mm/min), which is the time required to reach the maximum peak of thrombin generation; and the area under the curve (AUC), the area under the velocity curve (in mm2), which equals the endpoint minus the starting point of the elasticity curve.

Figure 1.

 Schematic representation of a thromboelastogram. CT, coagulation time; CFT, clot formation time; α, angle; MCF, maximum clot firmness.

Thromboelastometric runs were stopped 30 minutes after the end of the CFT, because preliminary tests demonstrated the absence of significant changes after this period of time. Theoretically, conclusive information about fibrinolysis can be achieved only after 60 minutes, when lysis is complete. From a practical point of view, however, when fibinolysis is activated, the MCF begins to decrease earlier. Thus, fibrinolysis is first measured at 30 minutes by the parameter “LI30” which compares the amplitude at 30 minutes with the MCF.

Analytical validation of the thromboelastometer

Analytical imprecision was evaluated for each thromboelastometric test by repeated analysis of 2–4 replicates of the same 10 samples. The influence of storage was evaluated for all thromboelastrometric tests except fibrinolysis on 5 samples that were aliquoted, handled, and stored as described above. Each aliquot was analyzed by thromboelastometry at 2, 4, and 20 hours post-sampling. The effect of hemolysis was investigated on 5 samples (2 for fibrinolysis). Hemolysis was mechanically induced by repeated (5–10 times) rapid aspiration of an aliquot of the sample using a syringe and a 24-gauge needle, followed by strong squeezing of blood into the tube. Cell counts and thromboelastometry were performed before and after lysis. The RBC count and HCT before and after lysis were compared to calculate the percentage of hemolysis. In samples from 14 horses (5 for fibrinolysis), thromboelastometric tests were performed simultaneously to define the characteristics of the normal equine thromboelastogram, thus avoiding possible differences due to different work sessions.

Reference intervals for equine thromboelastograms at rest and postexercise

Reference values were established using all “baseline” samples (mean values from imprecision tests, values recorded at 2 hours in nonrefrigerated samples, values recorded prelysis, and samples used to define the normal thromboelastograms) and using additional blood samples from resting, racing, and saddle horses. The population used to establish reference intervals was thus composed of 30 horses for native coagulation (saddle horses=22, racing horses=8), 36 horses for the intrinsic pathway (saddle horses=20, racing horses=16), 35 horses for the extrinsic pathway and fibrinogen function (saddle horses=21, racing horses=15), and 18 horses for fibrinolysis (saddle horses=11, racing horses=8).

Differences in reference values after exercise were assessed on 14 horses on which thromboelastometry was performed 1 hour after a standardized race of 1600 m. Specifically, the effect of exercise on both the intrinsic and extrinsic pathways was investigated in 14 horses, on fibrinogen in 13 horses, on native coagulation in 7 horses, and on fibrinolysis in 6 horses.

Statistical analysis

Statistical analyses were done using statistical software (Statistica, Stat Soft Inc., Tulsa, OK, USA). For imprecision tests, coefficients of variation (CV) were calculated using the estimate of the pooled variance of differences between sequential readings. For each storage condition, results at different times were compared using the Friedman ANOVA test. A Wilcoxon paired t-test was used to compare the results of refrigerated versus nonrefrigerated samples at each sampling time. Results obtained pre- and postlysis were compared using a Wilcoxon paired t-test. Spearman correlation was used to investigate correlation between the percentage of lysis and thromboelastometric results. The results of thromboelastometric tests were compared using a Friedman ANOVA followed by the Bonferroni test. Results of the intrinsic and extrinsic pathways were also expressed as a percentage of native coagulation. Similarly, the function of fibrinogen was expressed as a percentage of the extrinsic pathway, because fibrinogen has been measured after activation of the extrinsic pathway.

To define reference ranges for each thromboelastometric test, mean, SD, and 5th–95th percentile values were calculated. Results from racing and saddle horses were compared using a Mann–Whitney U-test. Correlations between the results of the CBC and thromboelastometry were investigated using the Spearman test. Hematologic and thromboelastometric results obtained before and after racing were compared using a Wilcoxon paired t-test. Differences for all tests were considered significant when P<.05.


Assessment of health status

None of the horses included in the study had any abnormalities based on physical examination. Physical examination findings after the race were also unremarkable. No signs of hemorrhage were found on the skin or in the respiratory tract by bronchoscopy, excluding the presence of EIPH. Moreover, hematology and serum biochemistry results were unremarkable in all horses.

Analytical validation of the thromboelastometer


Imprecision was very low for MCF, α, and AUC, with CVs ranging from 1.2% (α of fibrinogen) to 5.9% (AUC of fibrinogen). CVs were slightly higher for CT and CFT, ranging from 4.5% (CFT of fibrinolysis) to 11.9% (CT of native coagulation). CVs were considerably higher for Vmax, ranging from 9.0% (fibrinogen) to 21.7% (intrinsic pathway).

Influence of storage time and temperature

High interindividual variability was noted for CT and CFT results, especially at 20 hours post-sampling. For the intrinsic pathway, CT at 2 hours and MCF at 20 hours were significantly higher and α at 2 hours was significantly lower in refrigerated compared with nonrefrigerated samples (P<.05). The CT of both the extrinsic pathway and fibrinogen function were significantly lower at 20 hours than at 2 hours and the MCF of the extrinsic pathway was significantly lower at 4 hours than at 2 hours (P<.05). No other significant differences were found in other test results based on storage time or temperature (Table 1).

Table 1.   Thromboelastometric results for clinically healthy horses and the influence of pre-analytical factors of exercise, breed, and sample hemolysis and storage.*
nCT (second)CFT (second)MCF (mm)Alpha (°)Vmax (mm/min)AUC (mm2)
  • *

    Results are mean ± SD (median), 5th–95th percentiles. Arrows indicate significant differences: E, postexercise compared with pre-exercise; R, racing compared with saddle horse; H, with hemolysis; 4°C, with refrigeration; S, with storage.

  • NA indicates not applicable; —, no significant differences.

  • AUC, area under the curve; CFT, clot formation time; CT, coagulation time; MCF, maximum clot firmness.

Native coagulation30487.6 ± 282.9184.6 ± 85.850.0 ± 8.159.8 ± 10.99.5 ± 3.35025 ± 809
 ↓E↑R; ↓E↓H↑E; ↓R↓H
Intrinsic pathway36262.0 ± 56.3120.4 ± 31.646.8 ± 6.368.6 ± 5.712.7 ± 4.34703 ± 631
 ↑4°C↑4°C; ↓H, E↓4°C, R↓R↓H, E
Extrinsic pathway3652.1 ± 9.5133.7 ± 44.453.8 ± 6.271.5 ± 6.915.1 ± 6.05352 ± 616
 ↓S↑R, E↓H, S, E↓R, E↓R↓H, E
Fibrinogen function3649.6 ± 6.6NA17.1 ± 4.974.8 ± 5.517.7 ± 5.11652 ± 390
 (50.3) (16.0)(76.0)(16.8)(1562)
 8.5–65.0 11.6–29.961.0–81.610.9–29.31123–2582
 ↓S ↓R, E↓R↓R, E↓R, E
Fibrinolysis1854.1 ± 6.3148.1 ± 24.950.6 ± 5.071.0 ± 6.413.9 ± 4.15026 ± 508

Influence of hemolysis

The percentage of hemolysis induced by mechanical stress ranged from 5.1% to 27.7% (mean±SD, 15.6±8.1%). Platelet number appeared higher in hemolyzed samples, likely due to the artifactual counting of membranes and lysed cellular debris by the automated analyzer. Hemolysis caused decreased coagulability, as shown in Figure 2, with significant (P<.05) decreases in MCF and AUC in tests of native coagulation and of the intrinsic and extrinsic pathways (Table 1).

Figure 2.

 Influence of sample hemolysis on native coagulation. The sample was analyzed before (left) and after (right) hemolysis.

The increase in Vmax of native coagulation after lysis was negatively correlated (r<0.90) with the decrease in HCT and RBC count. In evaluation of the extrinsic pathway, the increased “platelet” count, which as previously mentioned likely included enumeration of cell debris rather than only platelets, was positively correlated (r=0.90) with the increase in CFT and negatively correlated with the decrease in MCF.

Overall, these results indicated that mechanical hemolysis induced a decrease in the results of parameters associated with the interaction of cells and soluble factors.

Definition of the normal thromboelastographic pattern of horses

The analysis of samples processed simultaneously revealed that both the CT and CFT were significantly lower (P<.001) for the intrinsic pathway compared with native coagulation, and were even lower (shorter) for the extrinsic pathway and fibrinogen function. The MCF and AUC were both higher in the extrinsic pathway compared with both native coagulation and the intrinsic pathway; all these values were significantly different from each other (P<.001). The MCFs of the extrinsic and intrinsic pathways were 110% and 94% of the MCF of native coagulation, respectively (Figure 3). The MCF and AUC of fibrinogen function were about one-third of those of the extrinsic pathway. Vmax was significantly higher (P<.001) in the fibrinogen than in the extrinsic pathway, intrinsic pathway, and native coagulation. Thromboelastograms obtained using Aptem (fibrinolysis) were almost identical to those of the extrinsic pathway (data not shown). Overall, activation of coagulation in horses was slower in native coagulation than in other thromboelastometric tests; moreover, the highest strength of the clot was found in the extrinsic pathway and the lowest was found in the fibrinogen tests.

Figure 3.

 Overlapping of the thromboelastograms of extrinsic (E) and intrinsic (I), native coagulation (N) and fibrinogen (F). Note the increase in CT from E to N and the wider amplitude of E compared with N and I.

Reference intervals for equine thromboelastograms

Resting conditions

Reference intervals for the whole population of horses in the study and a summary of differences between saddle and racing horses were tabulated (Table 1). Results for native coagulation had strong interindividual variability and consequently a wider range, and outliers were present. CT was shorter for the extrinsic pathway than for the intrinsic pathway and native coagulation

For all the thromboelastometric tests except native coagulation, RBC counts were negatively correlated (P<.001) with the angle and Vmax, with r2 values ranging from −0.53 to −0.81 (see an example of the extrinsic pathway in Figure 4). The MCF and AUC also tended to correlate negatively with RBC counts but the correlation was statistically significant only for fibrinogen (P<.001; r2=−0.64 for both). By contrast, CFT was positively correlated (P<.01) with RBC count for the extrinsic pathway (r2=0.57; Figure 4) and fibrinolysis (r2=0.59)

Figure 4.

 Extrinsic pathway thromboelastogram. The results of Spearman correlation between RBC count (on the y axis, expressed as 106 RBC/μL) and angle (°), Vmax (mm/s), and clot formation time (CFT; second).

Compared with saddle horses, racing horses had a longer CFT for native coagulation (P<.01), extrinsic pathway (P<.01), and fibrinolysis (P<.05) (Table 1). MCF and AUC were lower in racing horses compared with saddle horses for fibrinogen and fibrinolysis (P<.05 for both). Similarly, the angle and Vmax were lower in racing horses for native coagulation (P<.05), intrinsic pathway (P<.01), extrinsic pathway (P<.05), fibrinogen (P<.05), and fibrinolysis (P<.05). Taken together, these results indicate that racing horses have a slower activation of coagulation, slower clot formation, and reduced clot firmness compared with saddle horses. Racing horses, however, had also significantly higher (P<.001) RBC counts and HCT compared with saddle horses.


After exercise, CT and CFT were significantly decreased (P<.05 for both) and the angle was significantly increased (P<.05) for native coagulation; MCF and AUC were significantly decreased (P<.01 for both) in the intrinsic pathway; and CFT was significantly increased (P<.05) and MCF (P<.001), angle (P<.05), and AUC (P<.001) were decreased in the extrinsic pathway (Table 1). MCF, Vmax, and AUC also were significantly lower (P<.05) in tests of fibrinogen function after the race, compared with resting conditions. RBC count and HCT were significantly higher (P<.05 and <.01, respectively) before the race than after the competition. Thus, most changes in thromboelastometric results after exercise suggested decreased coagulability.


The Rotem-gamma thromboelastometer had very good precision, as demonstrated by low intra-assay CVs, which were lower than those recorded in dogs.19 Only Vmax had significant intra-assay variability, likely due to the additive effect of the variability of the 2 parameters (MCF and time to reach the MCF) used for its calculation. Validation studies usually also investigate interassay imprecision. This was not carried out in the present study because it is known that thromboelastometry must be performed within a few hours of sampling.24 Nevertheless, we examined the influence of storage time and temperature. This might be superfluous, because the thromboelastometer is a patient-side portable instrument; however, in equine medicine, samples are commonly sent to a referral laboratory and it is thus essential to define the best transport and storage conditions. Although possibly influenced by the low number of samples and high interindividual variability, our results demonstrated that the intrinsic and extrinsic pathways and fibrinogen function are quite stable in nonrefrigerated samples. Refrigeration often influenced thromboelastometric results, likely due to damaged platelets.25 It is unlikely that repeated blood mixing before each analysis activated coagulation and masked subtle changes, because tubes were inverted very gently and the procedures were standardized. Based on these results, it is advisable not to refrigerate the blood and, in that eventuality it would be important to investigate native coagulation by immediately processing the samples. Unlike samples for native coagulation, samples for testing the intrinsic and extrinsic pathways and fibrinogen function can be processed up to 4 hours after collection, as in humans,24 and with rare exceptions, within 20 hours after collection. In healthy dogs, significant thromboelastographic changes occur after 2 hours of storage; however, thromboelastometric results from hyper- or hypocoagulable dogs are so intense they are often detectable after longer storage.19 Thus, even after prolonged storage, thromboelastometry may identify horses with coagulopathies.

In vivo studies have demonstrated that hemolysis can activate or alter coagulation,26,27 and that coagulation abnormalities are often associated with hemolysis associated with microangiopathic disorders, colic, and mechanical damage to RBCs caused by the impact of the hooves on the racetrack.5,8,28 Routine validation studies usually evaluate possible interference by hemolysis with spectrophotometric readings29; hemolysis in the cited study was simulated by adding a known amount of hemoglobin to the sample. In thromboelastometry, however, the simple presence of free hemoglobin would not simulate what likely occurs in vivo, where coagulation and platelet function can be activated by lysed RBC membranes with exposed phospholipids. On the contrary, the decreased concentration of RBCs in the sample decreases the probability of interaction between platelets and soluble coagulation factors. We thus induced mechanical hemolysis before performing thromboelastometry. This approach resulted in a substantial but widely variable amount of hemolysis in the samples. This variability might have been reduced by inducing hemolysis through the addition of chemicals or a hypotonic solution, or by freezing the sample, but these methods also influence coagulation.27 Again, the low number of animals could have affected the power of the statistical analysis, and more significant changes may be detected by examining a larger number of horses. Nevertheless, we demonstrated that after mechanical hemolysis, the blood began to clot later and had reduced clot firmness. This delayed clotting likely depends on the decreased concentration of RBCs in the samples, as demonstrated for native coagulation by the negative correlation between RBC count, HCT, and Vmax. The decreased concentration of RBCs thus seems to influence coagulation more than the presence of lysed membranes. However, the number of “platelets” increased after lysis in all samples, suggesting the impedance counter classified as platelets some cellular debris, likely composed of cell membranes, as already reported in horses.30 This partly influenced the results of the extrinsic pathway, as demonstrated by the positive correlation between platelet count and CFT and the negative correlation between platelet count and MCF. Independent of the mechanism responsible for these changes, these results suggest that hemolyzed samples should not be used for thromboelastometry.

Once the validation study demonstrated the reliability of the instrument, we examined the characteristics of the equine thromboelastograms; the general pattern was similar to that in other species, with some differences.14,19,20 Specifically, CT and CFT were shorter for the intrinsic and extrinsic pathways and fibrinogen function than for native coagulation, likely because of the presence of activators that accelerate clot activation. However, the CT was shorter in horses than in other species for the extrinsic pathway and longer for the intrinsic pathway,19,20 suggesting that in horses the latter is less efficient than the extrinsic pathway. Although no correlation between coagulation times and thromboelastometric results has been found,14 the differences in CT between the 2 pathways correlates with the shorter PT and longer aPTT in horses compared with other species.8 Once clotting is activated, the time to obtain the MCF was similar to that reported for other species, but was higher for the extrinsic pathway than for native coagulation and the intrinsic pathway. The MCF represents the final result of the interaction between coagulation factors and platelets and thus should not depend on the presence of activators, which, on the contrary, might influence the first phases of clot formation (CT, CFT) and the velocity by which the MCF is reached (Vmax). As a consequence, as demonstrated in other species,14 the MCF should be similar for native coagulation and after activation of the intrinsic or extrinsic pathways. In horses, it seems in vitro activation of the extrinsic pathway increases clot firmness. This might depend, as previously mentioned, on increased efficiency of the extrinsic pathway typical of horses or on the type of activator used for its evaluation. In humans it has been demonstrated that different activators can alter clot firmness.31 Interestingly, in spite of higher fibrinogen concentration in equine plasma13 and lower platelet count,30 the MCF of fibrinogen function was about one-third that of native coagulation, similar to what occurs in other species.14

Most of the results for thromboelastometric parameters seem to depend on the characteristics of the blood sample. In this study, when RBC mass increased, the time necessary for activating coagulation and for achieving clot firmness decreased. This contrasted with what was reported above regarding the effect of hemolysis, where a decrease in RBC mass was associated with a hypocoagulable state. However, the artifactual decrease in RBC mass induced by hemolysis in this study represents an extreme condition found only in diseased horses, while the correlations reported in the study cited above were for RBC counts that remained within reference intervals. The presence of a negative correlation with the RBC mass, however, was responsible for the different ranges recorded in saddle vs racing horses or before and after exercise. Erythrocyte values, in fact, were significantly higher in racing horses, likely due both to young age and training, and were further increased after exercise. Again, these differences were often minimal, but it would be advisable to use appropriate reference intervals when examining horses of different aptitude or after racing.

In conclusion, the results of this study demonstrate that the Rotem-gamma thromboelastometer is precise and can be used for examining coagulation in horses. To our knowledge, this is the first study that describes the characteristics of the normal equine thromboelastogram, which differs in part from that of other species. This information will facilitate the comparison of results from diseased horses in the future. Additionally, we demonstrate that preanalytical factors, such as hemolysis and long storage or refrigeration, decrease the coagulability of a sample. Delayed analysis, analysis of hemolyzed samples, or analysis of samples stored at 4 °C must be avoided to prevent the false diagnosis of hypocoagulable states. Finally, racing horses had reduced coagulability compared with saddle horses, especially after physical exercise. Consequently it is important to use appropriate reference intervals.


The authors are grateful to Dr. Marco Salvadori who provided most of the samples and to Dr. Marco Fontana and Dr. Vincenzo Scala (DASIT S.p.A) for technical and interpretative support.