The manuscript represents part of a thesis submitted by the 1st author at the Universitat Autònoma de Barcelona as partial fulfillment of the requirements for a PhD program supervised by the corresponding author. This study was partially presented at the 9th International Equine Colic Research Symposium in Liverpool (UK), June 2008.
Corresponding author: Luis Monreal, Departament de Medicina i Cirurgia Animals, Servei de Medicina Interna Equina, Facultat de Veterinària, Universitat Autònoma de Barcelona, 08193-Bellaterra, Barcelona, Spain; e-mail: firstname.lastname@example.org
Background: Peritoneal D-Dimer concentration can be determined to assess peritoneal fibrinolysis activity in horses with gastrointestinal disorders. However, blood contamination of peritoneal fluid may occur during collection and could alter peritoneal D-Dimer concentration.
Hypothesis/Objectives: Blood contamination in peritoneal fluid does not affect interpretation of peritoneal D-Dimer concentration in horses with colic.
Animals: Thirty-four horses with colic and 4 healthy horses.
Methods: Peritoneal fluid and blood samples were simultaneously collected upon admission. Then, peritoneal fluid was serially contaminated with the horse's own blood; final contaminations corresponded to 1, 5, 10, and 20% of blood in peritoneal fluid. D-Dimer concentration was determined in blood, peritoneal fluid, and contaminated peritoneal fluid samples. Data were analyzed using a longitudinal linear model and a generalized estimating equations analysis to assess the quantitative and qualitative variations of the effect of blood contamination on peritoneal D-Dimer concentration.
Results: Peritoneal D-Dimer concentration was only quantitatively affected when peritoneal fluid was contaminated at 20% of blood. However, when using increasing cut-off values of peritoneal D-Dimer concentration (100, 2,000, 8,000, and 16,000 ng/mL), this effect disappeared at the highest cut-off values (8,000 and 16,000 ng/mL). When peritoneal fluid contamination was grouped as “minimally contaminated” (≤1% of blood) and “highly contaminated” (≥5% of blood), no significant differences on D-Dimer concentration between both groups at each cut-off value were observed.
Conclusions and Clinical Importance: Although quantitative results of peritoneal D-Dimer concentration could be affected by high levels of blood contamination (≥20%), interpretation of increased peritoneal fibrinolytic activity was not significantly affected.
Peritoneal fluid analysis is a very useful and simple technique that contributes to the diagnosis of gastrointestinal (GI) disorders in horses with colic. Gross appearance, cytological evaluation, and total protein concentration of the peritoneal fluid have been demonstrated to have diagnostic usefulness.1,2 Moreover, other variables determined in peritoneal fluid (eg, lactate, glucose, alkaline phosphatase, pH, and chloride) also have been reported to be useful in the diagnosis of GI disorders in emergency settings.1–5 Recently, D-Dimer concentration in peritoneal fluid of horses with colic has been evaluated by our group and it was confirmed that D-Dimer also is a useful test for the assessment of the peritoneal fibrinolysis activity and the diagnosis of severe GI disorders, especially in inflammatory conditions (eg, acute enteritis, peritonitis) and ischemic lesions.6 Considering that intraperitoneal fibrin formation should be associated with a proportionate increase in peritoneal fibrinolysis activity sufficient to breakdown this fibrin, a decreased peritoneal fibrinolysis response would allow the formation of permanent adhesions. Therefore, this test can be used to assess peritoneal fibrinolysis activity because of peritoneal fibrin formation in GI disorders and, consequently, to monitor the risk of peritoneal adhesions.
D-Dimer is a specific marker of fibrinolysis activity that has been demonstrated to be a sensitive and useful test in assessing blood hypercoagulation and hyperfibrinolysis in humans, dogs, and horses.7–17,a,b Moreover, several studies using D-Dimer on peritoneal fluid have been performed in humans, showing that this marker also is a reliable indicator of fibrinolysis in the peritoneum.18–20
Because of the blind nature of the peritoneal fluid sampling technique in horses and the likelihood of animal movement during collection, contamination of peritoneal samples with blood coming from vessels in the skin and musculature, but also from mesenteric vessels or spleen, is not unusual.21 Malark et al22 reported that up to 17% of blood contamination of peritoneal fluid in clinically normal horses did not significantly alter interpretation of the nucleated cell count or protein concentration. The effect of this blood contamination on other peritoneal components such as D-Dimers has not yet been evaluated. In addition, marked variations have been demonstrated to occur between peritoneal and intravascular coagulation and fibrinolysis activities in humans and horses,6,19,20,23,24 and blood contamination of peritoneal fluid could substantially alter the peritoneal D-Dimer concentration. Thus, the aim of the present study was to determine whether blood contamination can cause significant changes on peritoneal D-Dimer concentration in horses with different types of GI disorders. Horses that may have low and high peritoneal D-Dimer concentrations were included, together with horses that may have low and high plasma D-Dimer concentrations.
Materials and Methods
In this prospective study, horses with colic admitted to the Equine Teaching Hospital of Barcelona between June 2006 and June 2007, from which an apparently nonblood-contaminated sample was collected for diagnostic purposes, were included. Other admitted horses with colic, in which peritoneal fluid was suspected to be contaminated or clearly contaminated with blood during collection, were excluded. In addition, during this period of time some other horses with different diseases not affecting the GI system were included with the owners' consent, in order to incorporate animals with low peritoneal and plasma D-Dimer concentrations. In these control horses, physical examination, diagnostic tests, peritoneal fluid analysis, and postmortem examination (in those in which it was performed) did not reveal any abdominal lesion.
Peritoneal fluid was collected aseptically, by a sterile blunt teat cannula, 2 cm, just to the right of the midline at the most dependent area of the ventral abdomen, using a standard technique.6,21,25 After that, a blood sample was obtained by direct jugular venipuncture using vacutainerc system. Peritoneal fluid and blood samples were collected in 5 mL tubes containing 3.8% sodium citrate. Within 1 hour of sampling, part of the peritoneal fluid sample was serially contaminated with the horse's own blood using both citrated samples. Final contaminations corresponded to 1, 5, 10, and 20% of blood in peritoneal fluid (Fig 1). Blood, peritoneal fluid, and the contaminated samples were immediately analyzed by a semiautomated cell blood counter.d PCV was determined by the microhematocrit system, and total plasma or peritoneal protein concentration was measured by refractometry. After that, all citrated samples were immediately centrifuged at 1,000 × g for 15 minutes, separated from the sediment, and frozen at −70°C until assayed for D-Dimer determination.
D-Dimer Concentration (ng/mL)
D-Dimer concentration was determined in plasma, peritoneal fluid, and the contaminated samples by commercial reagents and controlse by a quantitative immuno-turbidimetric latex agglutination assay,f according to the instructions provided by the manufacturer. This assay has been used in other studies for equine samples6,17,a,b and utilizes antibody-coated latex particles that aggregate in the presence of D-Dimer increasing turbidity. All tests were performed blindly, in duplicate, and by the same investigator (M.A.D.). When values were above the detection limit of the machine, dilutions were made.
Commercial software (SPSS, 15.0 version)g was used for all statistical analyses. Results for the quantitative effect of blood contamination were presented as mean and 95% confidence interval (CI) for each estimation between different cut-off values of D-Dimer quantification (100, 2,000, 8,000, and 16,000 ng/mL) and for each level of contamination (noncontaminated, 1, 5, 10, 20%, or blood). For this analysis, a longitudinal linear model analysis was used, which considered the intrasubject correlation for each horse and values of D-Dimer concentration on noncontaminated peritoneal fluid, different contaminations, and blood (considering blood as 100% of contamination), as evaluable prognostic factors.
For qualitative data analysis, samples were grouped as “minimally contaminated” (noncontaminated peritoneal fluid and 1% contamination) and “highly contaminated” (between 5 and 20% contamination). The same D-Dimer cut-off values (100, 2,000, 8,000, and 16,000 ng/mL) that were used for the quantitative analysis were used for sensitivity analysis. Odds ratio (OR) and 95% CI were then calculated for each cut-off value, in order to evaluate the increment of positive samples for D-Dimer concentration (defined as the quantitative value that is above the cut-off value) from blood versus “highly contaminated” samples, and from blood versus “minimally contaminated” samples by means the generalized estimating equations (GEE) analysis. This longitudinal analysis uses an appropriate distribution for the dependent variable; in this case, a binomial distribution (for the proportion of horses above the cut-off value at each level of contamination) was assumed. In the present study, the analysis includes level of contamination as the evaluable factor, and the “horse” as intrasubject factor (considering correlations within each horse). Additionally, the proportion of horses above the cut-off value and their 95% CI for each cut-off value of contamination were estimated by means of this GEE model. In all cases, tests were considered to be significant if P≤ .05.
A total of 38 horses were included in this study, and another 12 horses were excluded during the period of the study because of suspected blood contamination of the peritoneal fluid sample during collection. Eight of 38 horses were stallions (21%), 15/38 geldings (39.5%), and 15/38 mares (39.5%), aged between 0.7 and 26 years (mean ± SD, 8.8 ± 5.0 years). Breed distribution reflected the hospital's referral population, with 8 Andalusians (21%), 5 Warmbloods (13%), 3 Thoroughbreds (8%), 3 Draft horses (8%), 2 Arabians (5%), 1 donkey (3%), 9 cross-bred horses (24%), and 7 horses representing other breeds (18%).
Thirty-four of 38 horses were admitted because of abdominal pain. Diagnosis was based on clinical history, complete physical examination and results of diagnostic tests (CBC, plasma biochemistry, blood gas analysis, abdominal ultrasonography, and peritoneal fluid analysis). Findings of abdominal radiology, laparotomy, or postmortem examination were used for diagnostic purposes whenever these were performed. Seventeen horses had a diagnosis of colon obstruction, 9 horses had an acute inflammatory disease, 6 horses had an ischemic lesion, and 2 horses had hepatopathy. The remaining 4 of 38 horses were included in the study in order to incorporate horses without abdominal disease and with low D-Dimer concentration in peritoneal fluid. Their diagnoses were pneumonia in 2 horses, humeral fracture in 1 horse, and a dermatologic problem in another horse.
Assessment of Peritoneal Fluid Contamination
Cell counts, PCV, and total protein concentration demonstrated progressive increases in their values when increasing the percentage of blood contamination, compared with noncontaminated peritoneal fluid (Table 1).
Table 1. Effects of the ex-vivo blood contamination in peritoneal fluid samples (n = 38).
Erythrocytes (× 106/μL)
Total Nucleated Cells (× 103/μL)
Total Protein (g/dL)
Mean ± SD values of PCV, cell counts (erythrocytes and total nucleated cells), and total protein concentration in noncontaminated peritoneal fluid, contaminated peritoneal fluid (1, 5, 10, and 20% of blood contamination), and blood samples are shown.
0.00 ± 0.00
0.08 ± 0.03
1.42 ± 2.67
1.72 ± 1.37
0.29 ± 0.40
0.17 ± 0.27
1.50 ± 2.69
1.73 ± 1.37
1.48 ± 0.58
0.38 ± 0.10
1.68 ± 2.62
1.86 ± 1.34
3.01 ± 0.88
0.72 ± 0.18
1.88 ± 2.47
1.99 ± 1.31
6.23 ± 1.77
1.42 ± 0.36
2.42 ± 2.33
2.29 ± 1.20
31.91 ± 7.84
6.87 ± 1.70
6.78 ± 3.39
5.63 ± 0.87
Plasma and Peritoneal D-Dimer Concentrations
The mean (and 95% CI) D-Dimer concentrations of the uncontaminated peritoneal fluid of control and colic horses were 38.4 ng/mL (range, 0–82.0) and 7,430 ng/mL (range, 4,813–10,047), respectively. The mean (and 95% CI) D-Dimer concentrations of the blood of control and colic horses were 847 ng/mL (range, 0–2,236) and 1,799 ng/mL (range, 710–2,887), respectively.
Quantitative Analysis Results
Using the longitudinal linear model, a significantly lower D-Dimer concentration was observed in the 5% contamination (P= .031), 20% contamination (P= .004), and blood (P < .001) samples for a global comparisons approach when compared with the noncontaminated peritoneal fluid samples. However, mean differences of D-Dimer concentration (difference between the noncontaminated peritoneal fluid sample and each contaminated sample) and their 95% CI showed that values of the 5% contamination group were overlapped by values of the 1 and 10% contamination groups, although having a significant P-value (Table 2).
Table 2. Global quantitative results of D-Dimer concentration (ng/mL) in noncontaminated peritoneal fluid, contaminated peritoneal fluid (1, 5, 10, and 20% of blood contamination), and blood samples.
Differences of D-Dimer Concentration
Data are expressed as means (95% CI).
Differences of D-Dimer concentration were calculated as the difference between D-Dimer concentration of noncontaminated peritoneal fluid and D-Dimer concentration of each contaminated sample or blood.
0.0 [referent value]
−126.5 (−419.2 to 166.2)
−345.8 (−658.6 to −32.9)
−371.6 (−883.9 to 140.8)
−1,410.1 (−2,340.8 to −479.4)
−4,954.2 (−6,953.6 to −2,954.9)
A significant (P < .001) interaction between the mean difference of D-Dimer concentration in noncontaminated peritoneal fluid and the mean difference obtained in other contaminated samples or blood was also found such that horses with low D-Dimer concentrations (between 100 and 2,000 ng/mL) in the noncontaminated peritoneal fluid were more affected by high levels of blood contamination, whereas horses with high D-Dimer concentrations (above 8,000 and 16,000 ng/mL) showed higher variability, thus obscuring significant differences between horses with ≤8,000 or >8,000 ng/mL, and horses with ≤16,000 or >16,000 ng/mL. As shown in Figure 2, horses with D-Dimer concentration ≤100 ng/mL in the noncontaminated peritoneal fluid sample were hardly affected by blood contamination, although plasma D-Dimer concentrations usually were higher than peritoneal D-Dimer concentrations in these horses. On the other hand, in horses with peritoneal D-Dimer concentrations >100 ng/mL, increasing levels of blood contamination underestimated the peritoneal D-Dimer concentration at all levels of contamination, except for 20% contamination. When comparing horses with ≤2,000 or >2,000 ng/mL in the noncontaminated peritoneal fluid, differences only appeared in 20% contamination peritoneal samples and in blood samples. At higher D-Dimer cut-off values (8,000 and 16,000 ng/mL), statistical differences were only found in blood samples in the case of the 8,000 ng/mL cut-off value. This occurred because blood contamination in horses with either ≤8,000 or ≤16,000 ng/mL produced a decrease in peritoneal D-Dimer concentration (not observed at lower cut-off values), which approximated those observed in horses with >8,000 and >16,000 ng/mL, respectively. On the other hand, and as previously stated, the 95% CI was wider in horses with D-Dimer concentration >8,000 and >16,000 ng/mL, showing increased variability (Fig 2).
The GEE model showed that blood contamination decreased the proportion of horses with D-Dimer concentrations above each cut-off value, except for the 100 ng/mL value in which blood contamination increased this proportion (Table 3). In the qualitative analysis and taking 2,000 ng/mL as the cut-off value, D-Dimer concentrations above 2,000 ng/mL were more frequently found in the “minimally contaminated” group (OR, 1.48; 95% CI, 1.27–1.73) and in the “highly contaminated” group (OR, 1.37; 95% CI, 1.18–1.59) than in blood samples. Pairwise comparisons between both groups versus blood confirmed that statistically significant (P < .001) differences existed in both cases. However, significant differences were not found between both groups of “minimally contaminated” versus “highly contaminated” (P= .071). Similar results were seen at the 100, 8,000, and 16,000 ng/mL cut-off values; although when the lowest cut-off value of D-Dimer concentration (100 ng/mL) was taken, concentrations above 100 ng/mL were more frequently found in blood (see Table 3 for details).
Table 3. Qualitative assessment of ex vivo blood contamination across 100, 2,000, 8,000, and 16,000 ng/mL D-Dimers cut-off values when they were grouped as “minimally contaminated” (noncontaminated peritoneal fluid to 1% of blood contamination) and “highly contaminated” (5% of blood contamination or higher).
OR (95% CI)
Estimated Proportion of Samples above the Cut-Off Value (95% CI)
Differences of Estimations (95%CI)
Results represents OR (95% CI) for the “minimally contaminated” and “highly contaminated” groups from blood samples, together with differences of estimations (95% CI) among “minimally contaminated,”“highly contaminated,” and blood samples, and their associated P-values.
0.03 (−0.02 to 0.08)
−0.08 (−0.16 to 0.01)
−0.39 (−0.55 to −0.24)
−0.32 (−0.46 to −0.17)
0 (−0.07 to 0.07)
0.03 (−0.02 to 0.08)
−0.29 (−0.43 to −0.15)
−0.29 (−0.43 to −0.15)
−0.05 (−0.12 to 0.02)
0.03 (−0.02 to 0.08)
−0.16 (−0.27 to −0.04)
−0.11 (−0.2 to −0.01)
Results of this study suggest that iatrogenic blood contamination of peritoneal fluid does not significantly alter the peritoneal D-Dimer concentration of horses with colic at low levels of blood contamination (<20% of blood). Although there are no reports about the effect of blood contamination on peritoneal D-Dimer analysis in horses, these results are not surprising on the basis that Malark et al22 reported that up to 17% blood contamination of peritoneal fluid in clinically normal horses did not significantly alter interpretation of the nucleated cell count or protein concentration.
In a previous study by our group, D-Dimer concentration in peritoneal fluid of horses with colic was found to be a useful test for the assessment of the peritoneal fibrinolytic activity and the diagnosis of severe GI disorders.6 In that study, peritoneal D-Dimer concentration of healthy horses was found to be very low (much lower than that obtained in plasma samples), whereas in horses suffering from GI disorders peritoneal D-Dimer concentrations considerably exceeded plasma concentrations. Moreover, the highest peritoneal D-Dimer concentrations were found in horses suffering from severe GI disorders (eg, peritonitis, enteritis, and ischemic disorders), with altered peritoneal fluid analysis (eg, modified transudate, exudate), and in nonsurvivors. Thus, to perform the study reported here, horses suffering from different GI processes and also some healthy horses were selected in order to evaluate the effect of blood contamination on peritoneal fluid in horses having low and high peritoneal D-Dimer concentrations together with horses having low and high plasma D-Dimer concentrations. However, because of the considerable differences between horses in D-Dimer concentrations, either in peritoneal fluid or plasma, comparisons between means of each contamination group were not valuable in this study. Otherwise, pairwise comparisons of the mean difference values (the difference between the noncontaminated peritoneal fluid and each contaminated sample or blood) of each horse were selected to evaluate this effect.
When evaluating the quantitative effect of blood contamination, a trend to decrease the peritoneal D-Dimer concentrations was observed as the blood contamination level increased. However, significant differences were only found when peritoneal fluid was contaminated with 5 and 20% blood. Although having a low P-value, the 5% contamination group was not considered to be relevant because of the low mean difference value, which was comparable to the 1 and 10% contamination values, and because its 95% CI was included within the results of the 1 and 10% contamination groups. Thus, this result could have been because of an excess of precision of the statistical analysis. A relationship was detected between the D-Dimer concentration of the noncontaminated peritoneal fluid and the mean difference obtained in the other samples. In fact, statistical analysis showed that horses with low D-Dimer concentrations in the noncontaminated peritoneal fluid (100 and 2,000 ng/mL) were more affected by blood contamination, whereas horses having higher concentrations (8,000 and 16,000 ng/mL) were expected to have a higher mean differences of D-Dimer and larger CIs, which obscured the differences among groups. Regarding the 100 ng/mL cut-off figure (Fig 2, top left), the 95% CI of the 1, 5, and 10% blood contamination groups overlapped, but this does not mean that statistically significant differences could not be found. It is usually true that when CIs overlap, they tend to result in P-values that are not significantly different. However, this is not always true because some other factors influence the evaluation of statistical significance, such as the type of statistical analysis, differences between the calculation method of the CI, and the calculation method of the P-values.
The 4 peritoneal cut-off values selected in this study were based on the results of our previous study.6 The medians of peritoneal D-Dimer concentration of horses suffering from obstructions, enteritis, and ischemic processes (2,000, 8,000, and 16,000 ng/mL, respectively) were used, as well as the higher reference value of the control group (100 ng/mL). Caution should be taken when considering the cut-off value of 100 ng/mL for classifying a peritoneal D-Dimer concentration as normal or altered because additional studies using larger numbers of healthy horses are needed to establish a normal reference range for peritoneal D-Dimer concentration. Moreover, additional studies with horses suffering from GI diseases and diseases not affecting the GI system should be conducted to compare peritoneal D-Dimer concentrations.
Although significant differences were found at 20% blood contamination, it does not necessarily mean that such differences would alter interpretation of the test. This suspicion was confirmed by the results of the GEE model. Blood contamination decreased the likelihood of finding increased D-Dimer concentrations at the 2,000, 8,000, and 16,000 ng/mL levels, and increased the likelihood at 100 ng/mL. Significant differences were not found in either of those cases between the groups of “minimally contaminated” (noncontaminated to 1% contamination) and “highly contaminated” (from 5 to 20% contamination) samples. These groups were selected on the basis of the gross appearance of the peritoneal fluid (see Fig 1), because starting from 5% blood contamination the appearance turned clearly reddish and somewhat difficult to differentiate among the more highly contaminated samples. Also, some clinicians commonly reject highly contaminated samples when evidence of iatrogenic vessel puncture exists during the abdominocentesis procedure, despite the demonstration by Malark et al22 that it has no effect on the interpretation of the nucleated cell count and total protein concentration.
In the study reported here, PCV, erythrocyte and total nucleated cell counts, and total protein concentration were determined in blood, noncontaminated peritoneal fluid, and in each contaminated sample. However, a statistical analysis of these data was not performed because these variables were determined as markers of increasing levels of contamination, in order to confirm that blood contamination induced changes in peritoneal fluid components. On the other hand, in the present study changes of total nucleated cells and protein concentration in peritoneal fluid because of blood contamination were similar to those previously reported.22
Although blood contamination of peritoneal fluid samples has not been studied widely, it also is a concern when collecting other body fluids, such as cerebrospinal fluid (CSF). In 1 study performed on CSF of dogs that was ex-vivo contaminated with 2% blood, CSF D-Dimer concentration did not show a significant increase.h Similarly, in another clinical study that evaluated the effects on total protein concentration and nucleated cell count of iatrogenic blood contamination of CSF in dogs with and without neurologic diseases, moderate blood contamination did not affect the diagnosis.26 Moreover, another study conducted on CSF of horses to assess the effect of blood contamination on a serologic test, concluded that the test was reliable only when blood contamination was moderate.27 These conclusions seem to be in agreement with those of the present study.
In conclusion, although quantitative analysis shows that peritoneal D-Dimer concentration could be affected by high levels of blood contamination (≥20%), the interpretation of the peritoneal D-Dimer concentration was not significantly affected by blood contamination up to a 20% level in horses with different GI diseases.
aArmengou L, Monreal L, Segura D, Navarro M. Plasma D-Dimers in horses with colic. Proceedings of the 8th International Equine Colic Research Symposium. Quebec, Canada; August 2005:1 (abstract)
bCesarini C, Monreal L, Segura D, et al. Hemostatic follow up of horses with medical and surgical colic. J Vet Intern Med 2009;23:434 (abstract)