Materials and Methods

Study population—The following cases were included in the study: horses referred to the Equine Medical Center of Ocala with colic that had ≥ 2 abdominocenteses performed, including one at admission and another within 1 to 6 hours after admission. The timing for collection of the second abdominal fluid sample was selected on the basis of the attending clinician's impression that the patient had deteriorated clinically and was not associated with the passage of a specific amount of time. Horses were excluded from the study if they were < 6 months of age or considered to have diffuse peritonitis or primary kidney or liver disease. Horses were also excluded from the study if they had ≥ 2 of the following conditions: diarrhea (manure that was not supported by straw bedding); marked leukopenia (WBC count, < 3,000 WBCs/μL); or pyrexia (ie, rectal temperature > 39.2°C [102.6°F]).

Case definitions—An SL was diagnosed when there was gross evidence of the simultaneous occlusion of the intestinal lumen and its blood supply. An NSL (simple intestinal obstruction) was diagnosed when there was gross evidence of intraluminal occlusion without vascular obstruction. Abdominal lesions were classified as being either SLs or NSLs on the basis of findings obtained at either surgery or necropsy. Horses with signs of colic that responded to medical treatment were assumed to have an NSL.

Data collection—Data were collected retrospectively from the medical records of horses admitted with colic between September 2006 and February 2010. All horses had an initial physical examination that included abdominal auscultation and percussion, rectal examination, nasogastric intubation, and abdominal ultrasonography. Information on duration of colic was not reliable or consistently available, so it was not included in analyses. Venous blood was collected aseptically and submitted for hematologic evaluation and serum biochemical analysis. Peritoneal fluid was collected as follows: a 2-cm² area was clipped of hair immediately caudal to the xiphoid and to the right of the linea alba with the horse standing, the area was scrubbed alternately with chlorhexidine and alcohol solutions, and then a sterile stainless steel teat cannula was inserted through a small stab incision. The second and subsequent peritoneal fluid samples were obtained through the stab incision by use of a sterile teat cannula following scrubbing. Venous blood and peritoneal fluid samples were collected in an evacuated glass tube^a containing EDTA; peritoneal fluid was obtained by gravity flow from the teat cannula. A 2-cm² area was clipped of hair over the left jugular vein, the area was scrubbed alternately with chlorhexidine and alcohol solutions, and a silastic catheter was inserted IV and sutured in place. The horse was taken to the stall and administered a 20-L bolus of lactated Ringer's solution IV within 1 hour.^b

Clinicopathologic data were determined for all blood and peritoneal fluid samples within 10 minutes after collection. Other values recorded for peritoneal fluid included gross appearance, nucleated cell count, and total protein and lactate concentrations. Hematology values were determined by use of an automated system,^c and differential diagnoses were determined following a Gram stain of a blood smear and microscopic review by a trained veterinary technician. Serum biochemistry values were determined by use of an automated multiple profile system.^d All BL and PFL measurements were determined by use of a point-of-care portable lactate monitor.^e The reliability of this handheld lactate analyzer and the details of its operation have been reported previously.^19,20,29

The following variables were recorded for each horse in the study: date of initial evaluation; age; breed (categorized as warmblood, Thoroughbred, Quarter Horse, Paso Fino, Arabian, or other breed); sex (female or gelding or sexually intact male); heart rate (beats/min); findings of rectal examination or abdominal ultrasonography (no abnormalities detected, cecal distension, small intestinal distension, colonic impaction, colonic displacement, colonic distension, or other); signs of pain (absent, mild, moderate, or severe or the horse was dull or lethargic); peripheral blood and peritoneal fluid WBC count, PCV, and total protein concentration; BL and PFL (at admission and subsequently); whether surgery was performed (no vs yes); whether there was an SL (no vs yes); whether an intestinal resection or a pelvic flexure enterotomy was performed; diagnosis (spasmodic or gas colic, small intestinal volvulus, strangulating lipoma, mesenteric rent with small intestinal incarceration, ileal impaction, anterior enteritis, large colon volvulus, large colon displacement, large colon impaction, cecal impaction, gastric rupture, renosplenic [nephrosplenic ligament] entrapment, gastric ulcers, and other cause of colic); outcome (discharged, died, or euthanized); and days hospitalized.

Data analysis—The association between the dichotomous outcome (SL or NSL) and individual variables of interest was analyzed by use of univariable logistic regression analysis. The magnitude of these associations was expressed as the OR, derived by exponentiation of the coefficients of univariable logistic regression analyses. The 95% CI were determined by use of maximum likelihood estimation; the P value for testing the null hypothesis that the OR was equal to 1 (ie, equal odds of SL being present or absent for a given variable) also was reported. Some continuous variables were recoded as categorical variables on the basis of findings of exploratory data analysis. Variables significantly associated with SLs by use of univariable logistic regression analysis were included in multivariable logistic regression analysis; both forward and purposeful regression modeling was performed that included all variables associated with being in the SL group with a value of P ≤ 0.20. The specificity and sensitivity for various test measurements were determined from 2 × 2 contingency tables, with sensitivity defined as the proportion of horses with SL that tested positive and specificity defined as the proportion of horses with NSL that tested negative. For all analyses, values of P ≤ 0.05 were considered significant; analyses were performed by use of statistical software.^f

Results

Among 804 horses admitted to the Equine Medical Center of Ocala during the study period, 94 (11.7%) met the inclusion criteria. Of the 94 horses, 26 (28%) had an SL (SL group). These horses included 10 horses with small intestinal volvulus, 8 with large colon volvulus, 5 with SLs associated with pedunculated lipomas, and 3 with mesenteric rents. The remaining 68 horses had NSLs (NSL group) and included 15 horses with ileal impactions, 18 with large colon impaction, 12 with large colon displacement other than renosplenic entrapment, 11 with spasmodic or gas colic, 5 with renosplenic entrapment, 2 with intraintestinal foreign bodies, and 5 with duodenitis or proximal jejunitis.

Signalment—Horses in the SL group were significantly (P = 0.004) older than horses in the NSL group (Table 1). There was no significant difference in the distribution of breed of horses between the SL and NSL groups. Among the SL group, the breeds represented were Thoroughbred (n = 8 horses), warmblood (5), Arabian (5), Quarter Horse (3), Paso Fino (2), and other breeds (3). Among the NSL group, breeds represented included Thoroughbred (n = 25), Quarter Horse (13), warmblood (11), Arabian (6), Paso Fino (2), and other breeds (11). Although the proportion of Arabians was greater among horses in the SL group (5/26 [19%]) than in the NSL group (6/68 [9%]), the odds of being Arabian were not significantly (P = 0.173) different between groups. The distribution of sexes was similar among the SL and NSL groups; in the SL group, there were 5 (19%) geldings, 6 (23%) sexually intact males, and 15 (58%) females, and in the NSL group, there were 18 (26%) geldings, 12 (18%) sexually intact males, and 38 (56%) females. Relative to male horses (geldings or sexually intact males), the odds of being female were not significantly (P = 0.803) greater for the SL group. Horses in the SL group were less likely to survive to discharge than were horses in the NSL group: the proportion of survivors among the SL group (10/26 [38%]) was significantly (P < 0.001) less than that of the NSL group (62/68 [91%]). The odds of survival were 16-fold greater (95% CI, 5.2 to 52.2; P < 0.001) for the NSL group relative to the SL group.

Physical examination findings—Heart rate was considered as a categorical variable for 2 reasons. First, the effects of heart rate did not appear to be linear in the logistic scale and the distribution of heart rates for the SL group was skewed toward higher values (mean, 60 beats/min; median, 51 beats/min). Thus, heart rate data were considered as a categorical variable (< 60 beats/min or ≥ 60 beats/min). Although the proportion of horses with heart rates > 60 beats/min in the SL group (14/26 [54%]) was greater than that in the NSL group (31/68 [46%]), this difference was not significant (P = 0.171). The median heart rate in the NSL group was 48 beats/min, similar to that of the SL group (51 beats/min).

Findings on abdominal palpation per rectum or on percutaneous abdominal ultrasonography performed at admission (Table 2) indicated that horses in the SL group were more likely than those in the NSL group to have either small intestinal distension or cecal distension detected by either method. Although the odds of having small intestinal distension detected at admission were greater for horses in the SL group (OR, 1.6; 95% CI, 0.9 to 2.7), the difference was not significant (P = 0.086). Small intestinal distension was detected in 8 of 26 (31%) horses in the SL group and 10 of 68 (15%) horses in the NSL group. Although the odds of having cecal distension at admission were greater for horses in the SL group relative to the NSL group (OR, 1.5; 95% CI, 0.9 to 2.6), the difference was not significant (P = 0.124); cecal distension was detected in 8 of 26 (31%) horses in the SL group and 11 of 68 (16%) horses in the NSL group.

Signs of pain—Signs of pain were categorized as absent, mild, moderate, or severe, or the horse was assessed as dull or lethargic. For purposes of analysis, signs of pain were considered as either moderate to severe or not moderate to severe. The odds of observing moderate to severe signs of pain were greater for the SL group relative to the NSL group (OR, 1.5; 95% CI, < 1.0 to 2.5), but this difference was not significant (P = 0.079). Eleven of 26 (42%) horses in the SL group and 16 of 68 (24%) horses in the NSL group had moderate to severe signs of pain observed at admission.

Clinicopathologic data at admission—The WBC count in blood at admission did not differ significantly (P = 0.295) between the SL group (mean ± SD, 9.9 ± 3.7 × 10³ cells/μL) and the NSL group (mean, 9.1 ± 2.9 × 10³ cells/μL). The values of PCV determined at admission did not differ significantly (P = 0.510) between the SL group (mean, 38 ± 7.7%) and the NSL group (mean, 39 ± 7.3%). The blood total protein concentration at admission did not differ significantly (P = 0.614) between the SL group (mean, 6.9 ± 0.74 g/dL) and the NSL group (mean, 6.8 ± 0.80 g/dL). Although not significant (P = 0.068), the BL at admission of horses in the SL group (mean, 3.3 ± 2.17 mmol/L) was higher than for those in the NSL group (mean, 2.6 ± 1.43 mmol/L). By use of various cutpoints suggested by exploratory data analysis to create dichotomous variables, BL was not significantly associated with the SL group (data not shown).

The distribution of peritoneal fluid WBC count at admission was skewed toward higher values for the SL (median, 1,700 WBCs/μL; mean, 6,235 WBCs/μL) and NSL (median, 1,150 WBCs/μL; mean, 9,228 WBCs/μL) groups. Consequently, data were transformed by use of the logarithm₁₀ function for purposes of analysis. There was no significant (P = 0.656) association of peritoneal fluid WBC count with being in the SL group. Although the mean peritoneal fluid total protein concentration at admission was higher in the SL group (mean, 3.0 ± 1.2 g/dL) than in the NSL group (mean, 2.7 ± 1.1 g/dL), there was no significant (P = 0.274) association of peritoneal fluid total protein concentration with being in the SL group.

Higher PFL at admission was significantly associated with being in the SL group, whether considered as a continuous variable or as a categorical variable (< 4 or ≥ 4 mmol/L; Table 1). The specificity and sensitivity of a PFL at admission of > 4 mmol/L for an SL was 91% and 27%, respectively.

The association of SLs with the PFL:BL ratio at admission was examined. Because exploratory data analysis revealed that the distribution of this ratio did not appear Gaussian, the data were transformed by use of the logarithm₁₀ function. Although values of the logarithm₁₀ of PFL:BL ratio at admission were higher for the SL group (mean, −0.03 ± 0.26) than for the NSL group (mean, −0.13 ± 0.27), this difference was not significant (P = 0.147). However, when the ratio was considered as a categorical variable (≤ 1 or > 1), horses in the SL group had significantly (P = 0.027) higher odds of having a ratio > 1 (Table 1).

Peritoneal fluid data collected 1 to 6 hours after admission—All horses included in the study also had a second peritoneal fluid sample collected. The distribution of WBC counts in these second peritoneal fluid samples was skewed toward higher values for the SL (median, 4,400 WBCs/μL; mean, 12,631 WBCs/μL) and NSL (median, 1,700 WBCs/μL; mean, 13,288 WBCs/μL) groups. Consequently, data were transformed by use of the logarithm₁₀ function for purposes of analysis. The log-transformed peritoneal fluid WBC counts were significantly higher in the SL group than in the NSL group (Table 1). Peritoneal fluid WBC count also was considered as a dichotomous variable with a cutpoint of 5,000 WBCs/μL. The odds of having the second peritoneal fluid WBC count ≥ 5,000 WBCs/μL were significantly (P = 0.003) greater for the SL group than for the NSL group. The specificity and sensitivity of a peritoneal WBC count of ≥ 5,000 WBCs/μL at 1 to 6 hours after admission for an SL was 85% and 46%, respectively. The total protein concentrations in the second peritoneal fluid sample collected were significantly higher in the SL group than in the NSL group. Peritoneal fluid total protein concentration was also considered as a categorical variable (< 3.5 or 3.5 g/dL). The odds of the total protein concentrations in the second peritoneal fluid sample being ≥ 3.5 g/dL were significantly greater for the SL group than for the NSL group. The specificity and sensitivity of a peritoneal fluid total protein concentration of ≥ 3.5 g/dL at 1 to 6 hours after admission for an SL was 79% and 58%, respectively. The PFL 1 to 6 hours after admission was significantly greater among the SL group than among the NSL group. The PFL was also considered as a categorical variable by use of various cutpoints suggested by exploratory data analysis to create dichotomous variables (< 4 or ≥ 4 mmol/L); horses in the SL group were significantly (P < 0.001) more likely to have a PFL ≥ 4 mmol/L than were horses in the NSL group. The specificity and sensitivity of a PFL of ≥ 4 mmol/L at 1 to 6 hours after admission for an SL was 90% and 55%, respectively.

Differences in peritoneal fluid values between admission and within 1 to 6 hours after admission—Although the peritoneal fluid WBC count had a larger increase (ie, higher positive difference between samples obtained within 1 to 6 hours after admission and at admission) among horses in the SL group (mean, 6,396 ± 14,471 WBCs/μL) than in the NSL group (mean, 4,060 ± 30,654 WBCs/μL), this difference was not significant (P = 0.712). However, when considered as a categorical variable (0 or > 0 WBCs/μL), the odds of having an increase in peritoneal fluid WBC count was significantly (P = 0.039) greater for the SL group than for the NSL group (Table 1). The specificity and sensitivity of an in increase in peritoneal fluid WBC count for an SL was 51% and 73%, respectively.

Although the increase in peritoneal fluid total protein concentrations was higher for the SL group (mean, 0.5 ± 1.2 g/dL) than for the NSL group (mean, 0.1 ± 0.7 g/dL), the difference was not significant (P = 0.056). Increase in total protein concentration in peritoneal fluid also was considered as a categorical variable. Although the odds of SL group horses having an increase in peritoneal fluid total protein concentration were increased relative to NSL group horses (OR, 2.1; 95% CI, 0.8 to 5.4), this finding was not significant (P = 0.108).

Values of the difference in PFL were significantly (P = 0.007) higher among horses in the SL group than in the NSL group (Table 1). When considered as a categorical variable, the odds of PFL increasing between admission and 1 to 6 hours after admission were significantly (P < 0.001) higher for the SL group relative to the NSL group. The specificity and sensitivity of an increase in PFL over time for an SL was 51% and 73%, respectively.

Multivariable logistic regression—Variables significantly associated with being in the SL group on the basis of univariable logistic regression were included by use of a forward stepwise scheme. This included all variables associated with being in the SL group (P ≤ 0.20). The best-fitting model included the following variables: PFL at admission (< 4 or ≥ 4 mmol/L), an increase in PFL (PFL 1 to 6 hours after admission – PFL at admission; > 0 mmol/L), and the interaction of these 2 terms (Table 3). After accounting for these variables, no other factors were significantly associated with the SL group. Although the P value for age (< 10 or ≥ 10 years of age) was not significant, a second model was purposely fit that included this term. Age did not appear to be an important confounding variable, and all trivariate and bivariate interaction terms including age were not significant.

The significant interaction term in the final multi-variable model indicated that the positive association of an increase in PFL with presence of an SL depended on the value of PFL at admission (Table 4). When the PFL at admission was < 4 mmol/L, an increase in PFL was significantly (P < 0.001) associated with a 61.7-fold greater odds of having an SL (95% CI, 7.6 to 518.7). Among horses with a PFL at admission of < 4 mmol/L, the sensitivity for detecting members of the SL group from an increase in PFL was 95% and the specificity was 77%. When PFL at admission was ≥ 4 mmol/L, an increase in PFL was not significantly associated with having an SL. Among horses with a PFL at admission of > 4 mmol/L, the sensitivity for detecting members of the SL group was 29% and specificity was 67%.

Discussion

In clinically normal horses, it has been reported that PFL ranges from 0.22 to 1.47 mmol/L^9–11,21,29 and that BL is always higher than PFL.¹⁰ Among 20 control horses and 189 horses with signs of colic in a previous study,²⁹ mean ± SD PFLs were significantly (P < 0.001) higher in horses with signs of colic (4.00 ± 4.63 mmol/L) than in controls (0.6 ± 0.19 mmol/L). Of 126 horses with signs of colic in that study,²⁹ 39 (31%) were classified as having an SL and 87 (69%) were classified as having an NSL; the mean ± SD PFL was significantly (P < 0.05) higher in horses with SLs (8.45 ± 5.52 mmol/L) than in horses with NSLs (2.09 ± 2.09 mmol/L). Of the 94 horses with signs of colic in the present study, 26 (28%) were classified as having SLs and 68 (72%) were classified as having NSLs. The mean PFL at admission was significantly (P = 0.049) higher in horses with SLs (3.3 ± 2.7 mmol/L) than in horses with NSLs (2.2 ± 2.2 mmol/L), and the mean PFL 1 to 6 hours after admission was significantly (P < 0.001) higher in horses with SLs (4.7 ± 2.7 mmol/L) than in horses with NSLs (1.9 ± 2.3 mmol/L).

Under aerobic conditions, a single molecule of glucose is converted to pyruvate and 2 molecules of ATP, and following diffusion of pyruvate into the mitochondria, it is converted to CO₂ and H₂O, with a theoretic yield of 36 molecules of ATP via the Krebs cycle.^33,34 During anaerobic conditions, glucose is converted to pyruvate but cannot be processed in the mitochondria because the final event of the electron transport chain in the Krebs cycle is oxygen dependent.^33,34 Therefore, in oxygen-deficient conditions, pyruvate is converted to lactate and the total energy yield is reduced to 2 molecules of ATP. In most critically ill patients, hypoperfusion and hypoxia are responsible for the increases in lactate concentrations identified in blood and peritoneal fluid. In the present study, referred horses with signs of colic were reviewed to determine the value of serial measurements of PFL in identifying horses with SLs. The best-fitting multivariable logistic regression model included the following variables as significantly associated with SLs: PFL at admission (< 4 or ≥ 4 mmol/L), an increase in PFL (ie, PFL 1 to 6 hours after admission – PFL at admission of > 0 mmol/L), and the interaction of these 2 terms (when PFL at admission was < 4 mmol/L, an increase in PFL was associated with a 61.7-fold greater odds of having an SL).

In a study²⁰ of serial measurements of BL in adult horses treated on an emergency basis, it was demonstrated that BL at admission (> 4.1 mmol/L) and an increase in BL over time had the strongest associations with non-survival for horses with strangulating disease of the large intestine and colitis. Conflicting information was found in another study,²¹ in which a statistical relationship between the presence of ischemic intestine and lactate concentrations could not be demonstrated for BL but was demonstrated for PFL. The results of the present study are consistent with the report by Delesalle et al²¹: although the BL of horses in the SL group (mean, 3.3 mmol/L) was higher than that in the NSL group (mean, 2.6 mmol/L), the difference was not significant (P = 0.0675). In a study³⁵ that used 8 pigs, segments of the small intestine were isolated by a vascular pedicle containing 1 artery and 1 vein so that intestinal ischemia could be created by occluding the mesenteric vasculature. It was demonstrated that PFL started to increase within 1 hour after occlusion, but systemic BL did not increase for the entire experiment (250 minutes).³⁵ When looking at PFL and survival rate in the study reported here, horses in the SL group (10/26 [38%]) were significantly (P < 0.001; OR, 16) less likely to survive to discharge than were horses in the NSL group (62/68 [91%]). A previous study²¹ found that PFL is superior in its ability to predict the need for surgery or intestinal resection, probability of ileus development during hospitalization, and probability of death. It has been suggested that PFL, rather than BL, is more suitable and sensitive for early recognition of intestinal ischemia and concomitant prediction of outcome.^21,35

Peritoneal fluid lactate concentration and BL have been compared previously as the PFL:BL ratio. In healthy horses, this ratio is always < 1 because BL is always higher than PFL in clinically normal horses at rest.^10,21,29 Among horses with signs of colic in 1 study,²⁹ this ratio was < 1.0 in 52 of 54 (96%) horses with simple obstructions. When this ratio has been used as an indication of the need for surgery, there was decreased need for surgical treatment of horses with a PFL:BL ratio < 1.0.³⁶ The ability of the PFL:BL ratio to identify horses with SLs was evaluated in the study reported here. When the ratio was > 1, the odds of horses having an SL (42%) was significantly (P = 0.0266; OR, 3.1) greater than for horses having an NSL (19%); however, this variable was not retained in the final multivariable model. In other studies^10,11 in which this ratio was evaluated as a predictor of survival rate in horses with signs of colic, all horses with a value < 1 survived. In another study,²⁹ 47 of 54 (87%) horses with a PFL:BL ratio < 1.0 survived. In another report,²¹ it was suggested that the PFL:BL ratio had limited value because the ratio offered little additional information than when PFL and BL were evaluated individually.

Horses with colic can be categorized into 3 groups: those that need surgery, those that do not need surgery, and those for which, without further observation and monitoring of key clinical signs, the need for surgery is uncertain.³⁷ In the study reported here, 94 horses underwent ≥ 2 abdominal fluid evaluations because the necessity of surgery could not be determined. Of the key clinical signs used to evaluate such horses, persistent signs of pain and the ability of analgesics to eliminate signs of pain are the variables that have the most influence on a surgeon's decision to perform a celiotomy.³⁸ It has been demonstrated that the severity of signs of abdominal pain is related to the severity of the conditions causing colic, and the ability of analgesics to eliminate abdominal pain has been used to determine the need for surgery.³⁹ In that report,³⁹ it was proposed that signs of pain alone are not a reliable indicator of the need for surgery in individual horses because 10 of 22 (45%) horses that had constant signs of pain after treatment with analgesia did not require surgery. The results of the present study also demonstrate that the presence of moderate to severe signs of pain was not significantly (P = 0.079) different between horses with SLs and NSLs, suggesting that signs of pain alone are not an accurate indicator of an SL.

Intestinal obstruction is classified as either a simple obstruction or a strangulating obstruction. Simple intestinal obstruction is defined as an intraluminal occlusion of the intestine without vascular obstruction, and a strangulating intestinal obstruction is defined as the simultaneous occlusion of the intestinal lumen and its blood supply.⁴⁰ Strangulating obstructions may be further classified either as ischemic or hemorrhagic; ischemic obstructions occlude both the artery and the vein, and hemorrhagic obstructions occlude the vein alone so that, with continued arterial blood flow, retrograde hydrostatic congestion causes intestinal hypoperfusion and ischemia. In horses with simple intestinal obstruction, studies^41,42 have shown that there are critical intraluminal pressure levels beyond which intramural venous compression takes place, reducing tissue perfusion and ultimately producing mucosal lesions similar to those encountered during hemorrhagic strangulating obstruction. In the study reported here, in 3 of 15 horses with an ileal impaction and 5 of 18 horses with a large colon impaction (8 horses total), the PFL was < 4 mmol/L at admission and the PFL increased between admission and the second peritoneal fluid evaluation 1 to 6 hours after admission (data not shown). By study definition, because these 8 horses did not have gross evidence of vascular occlusion identified at surgery or necropsy, their lesions were classified as NSLs. It is plausible that a hemorrhagic strangulating obstruction was a possible explanation for increasing PFL identified in these 8 horses. True-negative results were identified in 48 of 62 (77%) horses and false-negative results were identified in 14 of 62 (23%) horses, which included these 8 horses (Table 4). Thus, the specificity for the test would have been higher if the study definition for SL included horses with a hemorrhagic strangulating obstruction.

The final multivariable model in the present study identified a positive interaction term. The positive association of an increase in PFL with presence of an SL depended on the initial PFL being < 4 mmol/L and was significantly (P < 0.001; OR, 62) associated with the presence of an SL. When horses at admission had a PFL > 4 mmol/L (13/94 horses), an increase in PFL was not significantly associated with having an SL (Table 4). It is not clear why this occurred. The utility of an increase in lactate concentrations and a decrease in glucose concentration was evaluated at 30-minute intervals in 2 studies^35,43 that isolated the small intestine on a vascular pedicle containing a single artery and vein. With the initiation of vascular occlusion recorded as time zero, lactate concentrations increased and glucose concentrations decreased until hour 2, after which further increases in lactate and decreases in glucose concentrations were not identified as new steady-state concentrations were reached. It is likely that because the supply of glucose to the pedicle was exhausted, so too was the ability to produce lactate. In the present study, 5 of 7 horses with SLs had a PFL > 4 mmol/L at initial evaluation but the PFL did not increase over time. It is plausible that the supply of glucose and hence the ability to produce lactate was also exhausted in these patients. Recognizing that horses with a PFL > 4 mmol/L have disease that has progressed beyond what is found in horses with a PFL < 4 mmol/L, it has been previously speculated that the timing at which abdominocentesis is performed during the colic episode is important.²¹

Changes in PFL were evaluated every 1 to 2 hours in several horses in the present study for up to 6 evaluations during the first 24 hours (data not shown). In these same horses, evaluation of PFL was continued until the signs of colic resolved, an exploratory celiotomy was performed, or the horse was euthanized on the basis of financial constraints or a poor prognosis. The value of serial evaluations of PFL is supported by several reports. Fetal scalp BL during labor and delivery has been found to be a useful indicator of fetal distress due to lack of oxygen. Fetal scalp BL testing has been shown to be as predictive as fetal scalp pH testing, with the added benefits of requiring less blood, fewer scalp incisions, less time, and a lower sampling failure rate, compared with the pH testing.⁴⁴ In a recent study,⁴³ 12 pigs had their entire small intestine divided into 3 segments, with each segment isolated by a vascular pedicle containing 1 artery and 1 vein. Three study groups of 10 control segments, 10 segments with arterial ischemia, and 9 segments with venous ischemia were provided for lactate and glucose concentration evaluation every 30 minutes via microdialysis. Ischemia caused a significant (P = 0.001) increase in lactate and decrease in glucose concentrations that was different with respect to all groups at 1 hour, and the effect was even more pronounced 1.5 hours after the initiation of ischemia.⁴³

Ileal impaction is a leading cause of NSLs in horses. These impactions are more commonly diagnosed in the southeastern United States and in horses fed coastal Bermuda grass hay, a fact that can be useful when generating a list of differential diagnoses.⁴⁵ Differentiating an ileal impaction from an SL or anterior enteritis can be difficult and problematic, recognizing that an SL should be treated with an emergency celiotomy and an ileal impaction is initially treated medically with analgesics and IV fluids. It has been suggested that surgical treatment for ileal impactions should be pursued when horses are unresponsive to pain-reducing medications or that changes in the peritoneal fluid WBC count and total protein concentration suggest that the intestines are compromised.⁴⁶ In the present study, signs of pain and serial changes in peritoneal fluid WBC count were not included in the best-fitting model following multivariable logistic regression. On the basis of the results of the present study, serial evaluation of PFL could be used to differentiate ileal impactions that remain as NSLs versus those that become hemorrhagic strangulating obstructions.^42,43 When a rectal examination provides a definitive diagnosis of ileal impaction, serial evaluation of PFL would provide additional data to indicate whether these cases are likely to resolve with medical treatment.

In most critically ill patients, anaerobic metabolism is the most common explanation for increases in BL and PFL. Other plausible explanations that create tissue hypoxia include pulmonary disease, circulatory failure, and anemia. In the present study, 20 L of lactated Ringer's solution was administered as an IV bolus prior to evaluating PFL 1 to 6 hours after admission, recognizing that dehydration is a common finding in horses with signs of colic.³² We propose that administering these fluids prior to collecting the second abdominal fluid sample was an essential part of the methods if the effect that nonintestinal sources of lactate had on the PFL was to be eliminated or minimized. The mean difference in PFL (PFL 1 to 6 hours after admission – PFL at admission) for horses with NSLs was negative (−0.3 ± 1.7 mmol/L) and was significantly (P = 0.007) different than that for horses with SLs (1.4 ± 3.5 mmol/L). This decrease in PFL illustrated an event called lactate clearance and was a clinical finding that instilled confidence in the decision to continue to treat patients with medical management. It has recently been reported that a 2-hour lactate clearance > 15% is highly predictive of positive outcome and may reassure cardiorespiratory clinicians that the therapeutic approach is appropriate.⁴⁷ In a multicenter, open-label, randomized controlled trial on 348 ICU patients, the objective for patients in the treatment group was to decrease lactate concentration by 20% every 2 hours for the initial 8 hours of ICU stay.⁴⁸ In the control group, attending clinicians had no knowledge of lactate concentrations other than the value at admission. When adjusted for predefined risk factors, lactate-guided treatment significantly (P = 0.006) reduced hospital mortality rate in the treatment group, sequential organ failure assessment scores were lower, inotropes and mechanical ventilation were stopped earlier, and discharge from the ICU occurred earlier.⁴⁸

An important limitation of the present study was that only a small fraction of horses with colic referred to our facility underwent 2 peritoneal fluid analyses. Because of this potential selection bias, it is unclear whether the specificity and, less likely, the sensitivity of the serial evaluation of PFL would have been different had all horses admitted to the clinic for colic and hospitalized for at least 6 hours been included. Moreover, the extent to which the findings of the present study can be generalized to other clinics or to geographic locations where ileal impactions are rare is unknown. Also, the duration of colic prior to initial evaluation was not known and was not included in analyses. Another limitation of the present study was that the timing for collection of the second peritoneal fluid was variable and occurred between 1 and 6 hours after admission. More consistent sampling or protocols for sampling might have resulted in different results.

Another limitation of the present study was that many of the factors examined were strongly correlated or coassociated. For example, PFL and BL were significantly correlated. Collinearity of variables could have impacted our multivariable modeling. To the authors’ knowledge, there is no satisfactory way to deal with this problem. In the present study, each of the variables selected in the final multivariable model was replaced by variables with which it was significantly correlated (data not shown); none of these iterations resulted in a model that increased the magnitude or significance of observed associations (indeed, the magnitude and significance were uniformly reduced for the substituted coassociated variables). For example, PFL was a stronger (ie, had a greater estimated OR) and more significant (ie, had a lower P value) predictor of being in the SL group and thus collinearity with serum lactate concentration was not an issue.

Public auction companies for Thoroughbreds require disclosure of the fact that any horse that has undergone abdominal surgery in the last 2 calendar years preceding the day of sale and disclosure of the fact that any horse that has had a resection of an abdominal organ (partial or complete) regardless of the day of surgery. Owners that sell Thoroughbreds at public auction are under financial pressure to avoid abdominal surgery because the resale value for these horses is substantially reduced. Anecdotally, equine surgeons that provide services to Thoroughbreds sold at these public auctions are frequently pressured to delay celiotomy until the necessity for surgery is absolute. Thus, a highly sensitive test or testing algorithm for the need for abdominal surgery is much needed.³⁷ Although no such test exists, the ideal screening test would detect every case of ischemia (100% sensitivity) and do so without false-positive results (100% specificity). Monitoring for intestinal ischemia is a challenge because lesions are obstructed from view and neglected intestinal ischemia can become fatal because of its location, a situation that is not the case with bone or muscle ischemia.⁴³ With these considerations in mind, it may be preferential to have a screening test for intestinal ischemia with a high sensitivity and an acceptable specificity.⁴³ The results of the study reported here suggest that PFL determined at admission and an increase in PFL, especially among horses without a markedly high PFL at admission, move us closer to this single distinguishing test suggesting when the requirement for surgery is absolute; however, a multicenter, prospective study needs to be conducted to more completely identify the utility of PFL in determining the need for emergency surgery in horses with signs of colic evaluated at referral facilities.