Reason for performing study: Abdominal ultrasonography has become a part of the diagnostic investigation for the acute abdomen in many equine clinics. There is limited information on the correlation between abnormalities detected on abdominal ultrasonography and the specific category of small intestine (SI) and large intestine (LI) diseases.
Objectives: To investigate the relationship between abdominal ultrasonographic findings and disease categories that cause abdominal pain requiring surgery.
Methods: Medical records were reviewed for horses undergoing surgery or post mortem examination for colic. The ultrasound examination was performed to assess free peritoneal fluid, the left kidney, stomach, appearance and motility of the duodenum, identification, appearance, motility and thickness of small intestine loops, and the appearance and motility of the colon. Logistic regression analysis was used to identify associations between disease categories and ultrasonographic findings; a Chi-squared test was used to test for associations between each variable and disease categories.
Results: The study included 158 horses. Distended and nonmotile SI loops were associated with strangulated obstruction (n = 45); increased free peritoneal fluid, completely distended SI loops with abnormal motility and thickened loops were associated with definitive diagnosis involving SI (n = 58). Failure to visualise the left kidney was associated with renosplenic entrapment (n = 16); thickened large colon (LC) was associated with LC strangulating volvulus (n = 9).
Conclusion: The use of abdominal ultrasonography can be used for the accurate definitive diagnosis involving SI and LI diseases.
Potential relevance: This retrospective study may be used as a basis for prospective studies to assess the ultrasonographic findings in horses with medical colic and to compare these with surgical findings.
Colic is a common emergency in equine practice and is reported to be the highest cause of mortality in the horse (Tinker et al. 1997).
A quick and accurate assessment of horses with acute abdominal pain is important for a rapid diagnosis. In order to facilitate early treatment, diagnosis is usually restricted to simple discrimination between horses that require surgical vs. medical treatment. In this process, individual clinical and laboratory variables play an important role.
An algorithm based on clinical data was developed as a valuable discriminatory method to identify a need for surgery (Ducharme et al. 1989; Reeves et al. 1991). More recently, a prospective survey examined the reliability of degree of pain, peritoneal fluid colour and rectal temperature as possible predictors of the type of treatment needed (Thoefner et al. 2000). The authors recognised the presence of a high percentage of false positive cases that appeared to be in need of immediate surgery and suggested abdominal ultrasonographic evaluation as an alternative diagnostic tool.
Despite the limitations caused by the size and depth of the abdomen, the partial enclosure of the viscera by ribs and the degree of gaseous distension (Scharner et al. 2002), US provides information that could not be obtained by other means; moreover it allows access to regions of the abdomen that are inaccessible by other diagnostic methods (Fischer 1997; Freeman 2002a).
Recently, a protocol for fast localised abdominal sonography of horses (Flash) for the investigation of colic has been suggested as a technique that can be used in an emergency scenario by veterinarians without extensive US experience (Busoni et al. 2010).
The purpose of this study was to evaluate any correlation between the US examination findings and specific categories of SI and LI diseases obtained surgically or by post mortem examination.
Materials and methods
Medical records of horses admitted for investigation of an acute abdomen at the Veterinary Teaching Hospital between January 2006 and November 2010 were reviewed. Inclusion criteria were that the medical records were complete, there was a complete abdominal US examination (at least 10/13 parameters evaluated) (Busoni et al. 2010) and a definitive diagnosis was obtained by exploratory celiotomy and/or post mortem examination. Horses were selected for exploratory celiotomy on the basis of physical examination findings, clinical pathology, rectal palpation, abdominal US findings and persistent or recurrent pain. Horses were subjected to euthanasia if the owner did not consent to surgery and the clinical status suggested a poor prognosis for life; subsequently they were submitted to post mortem examination. The details obtained from the medical records included: sex, age, breed, abdominal US findings and definitive diagnosis.
Acquisition of ultrasonographic images
An ultrasonographic machine (Sonosite 8000SE)1, equipped with a 3.5 MHz curvilinear transducer, was used. Ultrasound evaluations were made in a systematic fashion as described previously (Fischer 1997; Reef 1998a), using alcohol and without clipping. The left kidney (Rantanen 1990; Reef 1991), stomach (Cannon and Andrews 1995; Reef 1998a), duodenum (Kirkberger et al. 1995) and colon (Reef 1998a,b) were examined and the amount of free peritoneal fluid was assessed. Loops of SI were evaluated only when visible. When horses did not allow the systematic approach because of severe abdominal pain, US evaluation was limited to the assessment of these abdominal structures in a limited number of windows (Busoni et al. 2010).
Interpretation of ultrasound images
The veterinary clinicians on call, who had trained for at least 6 months in both investigation of colic and abdominal ultrasonography, performed and gave an interpretation of the US examination; only one person performed the examination in each horse and gave an interpretation of US findings. A total of 5 veterinary clinicians performed the clinical and abdominal US examinations.
All examinations were performed to assess the amount of free peritoneal fluid (absent or normal, increased), and for visualisation of the left kidney (yes or no). The free peritoneal fluid was considered ‘normal’ if a small amount of fluid was observed ventrally (Freeman 2002b). The stomach was considered ‘distended’ if visible in >5 intercostal spaces (Rabba and Busoni 2009). The duodenum was assessed for appearance (normal, partially distended, completely distended) and motility (normal, reduced, absent). Appearance of the duodenum was defined as ‘partially distended’ if the duodenum had fluid, or fluid and gas-dilated lumen with a square shape, and as ‘completely distended’ if the duodenum had fluid, or fluid and gas-distended lumen with a round shape. Motility of duodenum was defined as ‘normal’ if duodenum had >3 distensions/min, ‘reduced’ if it had <3 distensions/min and ‘absent’ if it was amotile (Reef 1998a). The SI loops were assessed for the ability to be identified (yes or no), appearance (normal, partially distended [Fig 1], completely distended [Fig 2]) and motility (normal, reduced, absent). The duodenum and SI loops were assessed for the presence of a thickened wall (yes >3 mm or no <3 mm). Appearance of SI loops was defined as ‘partially distended’ if they had fluid, or fluid and gas-dilated lumen with a square shape, and as ‘completely distended’ if they had fluid, or fluid and gas-distended lumen with a round shape. Motility of the SI loops was defined as ‘normal’ if SI had continuous contraction (Klohnen et al. 1996), ‘reduced’ if SI had <6 contractions/min (Rabba and Busoni 2009) and ‘absent’ if SI was amotile. The LC was assessed for appearance (normal, distended, thickened [>5 mm] with or without distension]) and motility (normal, reduced, absent). Motility of the LC was considered ‘normal’ if LC had more than 2 contractions/min (Rabba and Busoni 2009), ‘reduced’ if LC had <1 contraction/min and ‘absent’ if LC was amotile. Large colon was defined as ‘distended’ when there was loss of sacculations in the ventral portions and gas or ingesta reflected the ultrasound beam, limiting the ability to visualise other structures (Scharner et al. 2002; Freeman 2003).
The underlying definitive diagnoses were classified primarily according to the viscera involved (stomach, SI, LI).
The definitive diagnosis was classified secondarily as nonstrangulating obstruction of the SI, strangulating obstruction of the SI, intussusception, diseases of the caecum, LC impaction, right dorsal displacement and renosplenic entrapment of the LC, LC strangulating volvulus, LC nonstrangulating volvulus, small colon impaction, inflammatory bowel diseases, peritonitis and adhesions.
Logistic regression analysis was used to determine whether various US findings were associated with the portion of the intestine affected (SI [yes vs. no]) and with each disease category (yes vs. no). Screening of all variables was performed using a univariable logistic regression model with the data for the primary classification and then each disease category as the dependent variable. Specific factors analysed for an association with the primary classification (SI) and with the disease categories included increased free peritoneal fluid (no vs. yes), visualisation of the left kidney (yes vs. no), distension of the stomach (no vs. yes), appearance of the duodenum, motility of the duodenum, loops of SI visible (yes vs. no), appearance and motility of SI loops, presence of thickened wall loop (no vs. yes), and appearance and motility of the LC. Variables with a univariable P value <0.3 were considered for subsequent inclusion in a multivariable model. Modelling was performed subsequently by forward-stepwise procedures, and the best models for each specific disease were indentified. For each model, odds ratios (OR) and their 95% confidence intervals (CI) were calculated. A Chi-squared or Fisher's exact test was used, when appropriate, to test the associations between the primary classification and the variables, and between each disease category and the variables. Values of P<0.05 were considered significant. For each significant variables the sensitivity, specificity, and positive (PPV) and negative predictive value (NPV) were calculated. The statistical package R2 was used for data analysis.
The details of 188 horses admitted for colic and requiring surgery were retrieved from the medical records database system. Of these, 30 were excluded because the US examination was considered insufficient. The remaining 158 cases were included in the study; the mean age at the time of clinical examination was 8.5 years (median 8 years; range one day to 23 years). There were 67 (42.5%) females, 53 (33.5%) geldings and 38 (24%) stallions. Fifty-four (34.5%) horses were Warmbloods, 29 (18.5%) Thoroughbreds, 15 (9.5%) Arabians, 12 (7.6%) Standardbreds, 8 (5.1%) Quarter Horses, 7 (4.5%) native breeds, and 5 (3.3%) ponies and 15 (9.5%) horses represented other breeds. The definitive diagnosis involved the LI in 91 (57.6%) horses, the SI in 64 (40.5%) and the stomach in 3 (1.9%) cases. The details of the definitive diagnosis are summarised in Table 1.
Table 1. Definitive diagnosis and disease category for 158 horses assessed by transcutaneous abdominal ultrasonography
Total number of horses
No. of horses
Epiploic foramen herniation
Gastrosplenic ligament incarceration
Large colon impaction
Right dorsal displacement of the large colon
Large colon volvulus
Small colon impaction
Inflammatory bowel disease
Abdominal ultrasonographic findings
The details of the abdominal US findings observed are summarised in Table 2. In 3 cases a target or ‘bulls-eye’ image was recorded during the US examination; in 7 horses distended loops of the SI were identified also inside the scrotum.
Table 2. List of the abdominal ultrasonographic findings evaluated in 158 colic horses and total number of horses in which they were assessed. Abnormal abdominal ultrasonographic findings evaluated and number of horses in which the abnormal ultrasonographic findings were obtained
No. horses assessed (%)
Number of horses showing each abnormality (%)
Peritoneal free fluid
Small intestine loops
The stomach (n = 3), peritonitis (n = 6), nonstrangulating obstruction (n = 7), intussusception (n = 6), large colon impaction (n = 7), caecal disease (n = 6), small colon impaction (n = 6), inflammatory bowel disease (n = 6) and adhesions were excluded from the statistical analysis because of the low number of cases in these categories.
Increased free peritoneal fluid (OR 2.42, 95% CI 0.08–1.85, P<0.05), reduced duodenum motility (OR 3.35, 95% CI 0.16–2.47, P<0.05), completely distended appearance (OR 6.3, 95% CI 1.10–3.80, P<0.0001) and absent motility (OR 4.61, 95% CI 1.1–3.32, P<0.0001) of SI loops were shown to be significantly associated with strangulating obstruction of the SI. Increased free peritoneal fluid (OR 3.16, 95% CI 0.43–1.98, P<0.001), reduced (OR 3.8, 95% CI 0.39–2.69, P<0.001) and absent (OR 3.28, 95% CI 0.51–3.47, P<0.01) motility of the duodenum, completely distended appearance of SI loops (OR 23.5, 95% CI 2.5–5.3, P<0.0001), reduced (OR 0.57, 95% CI 0.2–2.36, P<0.05) and absent motility (OR 21.2, 95% CI 2.29–4.6, P<0.0001) of SI loops were shown to be significantly associated with definitive diagnosis involving SI. Colon appearance, colon motility and visualisation of the left kidney were negatively associated with strangulating obstruction of the SI and with definitive diagnosis involving SI.
Increased free peritoneal fluid (OR 0.35, 95% CI 0.16–2.04, P<0.05) was shown to be significantly associated with right dorsal displacement of the LC. There was also a tendency to fail to identify SI loops in association with right dorsal displacement of the LC (P = 0.08). Appearance and motility of SI loops were negatively associated with right dorsal displacement of the LC. The distended appearance of LC (OR 4.10, 95% CI 0.12–2.73, P<0.05) and absent visualisation of the left kidney (OR 30.7, 95% CI 2.04–5.33, P<0.0001) were shown to be significantly associated with renosplenic entrapment. Thickened appearance (OR 12.1, 95% CI 0.61–4.24, P<0.01) and absent motility (OR 2.36, 95% CI 0.88–5.06, P<0.05) of the LC were shown to be significantly associated with strangulating LC volvulus; however, absent motility of LC (OR 1.84, 95% CI 0.98–2.07, P<0.05) and failure to visualise SI (OR 3, 95% CI 0.12–2.82, P<0.05) were shown to be significantly associated with nonstrangulating LC volvulus.
A final multivariable logistic regression model is shown in Table 3. The variables negatively associated with each disease category are summarised in Table 3.
Table 3. Results of multivariable logistic regression analysis of all abnormal abdominal ultrasonographic findings associated with disease categories with a P value <0.05 obtained from 158 horses with colic
Associations between disease categories and abdominal US findings
There was a significant difference between horses with strangulating obstruction of the SI and horses with other categories with respect to completely distended appearance of SI loops (P<0.0001). There was a significant difference in the presence of increased free peritoneal fluid (P<0.01) and distended appearance of the LC (P<0.001) between horses with a right dorsal displacement of the LC and horses with other categories. There was a significant difference between horses with renosplenic entrapment and horses with other categories with regard to the visualisation of the left kidney (P<0.0001). There was a tendency for a difference in the presence of increased free peritoneal fluid (P = 0.08) in horses with renosplenic entrapment and horses with other categories. There was a significant difference between horses with strangulating LC volvulus and horses with other categories with respect to thickened appearance (P<0.001) and absent motility (P<0.05) of the LC. There was a significant difference between horses with nonstrangulating LC volvulus and horses with other categories for the visualisation of SI loops (P<0.0001), but not for absent motility of the LC (P = 0.3). There was a significant difference between horses with definitive diagnosis affecting SI and those affecting LI with respect to increased free peritoneal fluid (P<0.001), reduced motility of the duodenum (P<0.001), completely distended appearance (P<0.0001) of SI loops, and thickened wall loops (P<0.05).
The sensitivity, specificity, and PPV and NPV for each positively significant US finding for each disease category are summarised in Table 4.
Table 4. Sensitivity, specificity, positive and negative predictive value of abnormal abdominal ultrasonographic findings for each disease category with a P value < 0.05 obtained from the multivariable models
Total number of horses
SI = small intestine; LC = large colon; PPV = positive predictive value; NPV = negative predictive value.
Completely distended SI loops
Right dorsal displacement
Increased peritoneal free fluid
Not visualised left kidney
Increased peritoneal free fluid
LC strangulating volvulus
Absent motility of LC
LC nonstrangulating volvulus
No visible SI loops
Absent motility of LC
Increased peritoneal free fluid
Completely distended SI loops
Reduced duodenum motility
Thickened wall loops
Abnormal motility of SI loops was 90% sensitive and 50% specific for SI disease. The PPV and NPV of this US signs were 60 and 86%, respectively.
In the last 15 years there have been major advances in ultrasound examination of the gastrointestinal tract in horses with colic. Despite considerable progress in this area, to our knowledge there are few studies that examine statistical correlations between abdominal US findings and the disease categories cause of colic identified at surgery or post mortem examination (Klohnen et al. 1996; Pease et al. 2004; Klohnen 2008; Busoni et al. 2010).
Although prospective studies are likely to generate more useful and scientifically strong data on the correlation between US and disease categories, retrospective studies, such as this one, are also helpful and provide valuable information on which future prospective studies can be based.
In accordance with the purpose of this study, the results indicate that for some disease categories it is possible to find positive correlations between preoperative US examination and the intraoperative or post mortem findings.
In the horse population included in this study, several SI problems were identified, such as strangulating lipoma, mesenteric volvulus, herniation in mesenteric tears, inguinal herniation and epiploic foramen entrapment. For each specific condition, there were relatively few cases; we therefore concentrated on strangulated obstructions, nonstrangulated obstructions and intussusceptions. Since the last 2 groups were represented by a small number of cases, they were excluded from the statistical analysis.
In agreement with other studies (Scharner et al. 2002; Klohnen 2008; Busoni et al. 2010), the presence of completely distended SI loops was highly related to definitive diagnosis involving SI as to strangulating obstruction. However, in the authors' experience, it is not uncommon to observe partial distension of SI loops in horses with LI disorders. This US finding might be explained by the compression exerted by the distended and/or displaced LC on the duodenum and by tension on the mesentery (Hardy 2008). However, there is a distinct difference in the US appearance of SI loops between surgical colics involving the LC and the appearance of the SI for nonstrangulated obstruction of the SI and strangulated obstruction of the SI; in nonstrangulating obstruction, prestenotic loops may appear to be dilated and with absent contractility, whereas post stenotic loops appear of normal aspect and contractility (Reef 1998a). In strangulated obstruction, SI loops appear completely distended with lack of motility and, finally, in cases of LC disease, SI loops appear partially distended with reduced or normal contractility. However, although values obtained could be considered acceptable, they are lower than those obtained by Klohnen et al. (1996). The lower sensitivity could be attributed to an early referral from the onset of clinical signs when a complete SI distension has not yet manifested. In contrast, lower PPV may be due to a relatively high number of false positive results, as reported by Busoni et al. (2010), in these cases evaluating the number of completely distended SI loops could help to reduce the number of false positive horses. In agreement with previous studies in strangulated obstruction, completely distended SI loops are associated with absent contractility of SI loops (Klohnen et al. 1996; Busoni et al. 2010), this is because intestinal distension and ischaemia induce congestion and inflammation of the bowel wall (Blikslager 2008). The absence of contractility could also be mediated by mechanoceptor or nociceptor stimulation of extrinsic afferent neurons decreasing gastrointestinal progressive motility (Rakestraw 2008). Strangulated obstruction generally causes a more rapid and serious alteration of perfusion, distension and contractility of the SI loops (Blikslager 2008). As reported by Klohnen et al. (1996) and Freeman (2002a), a lack of motility and thickened wall loops should be used to discriminate between surgical and nonsurgical cases in horses with SI diseases that show dilated and turgid loops. In our study, based only on surgical or post mortem cases, the results suggest that surgical treatment must not be excluded in the presence of dilated loops with reduced motility, but this should be interpreted in conjunction with clinical signs and clinical pathology, because, in the authors' experience, this US finding could represent an early stage of strangulated obstruction of SI. In agreement with previous studies (Klohnen et al. 1996; Fischer 1997), increased wall thickness has been associated with strangulated obstruction as well, as a consequence of the development of mural oedema or haemorrhage (Rowe and White 2008) so that the 5 US layers are lost (Reef 1998a), but the increase is related to the onset of the condition (Reef 1998a,b; Klohnen et al. 1996). This US finding has to be differentiated from horses with intussusception that may have thickened intestinal wall due to hypertrophy of the muscular wall (Freeman 2008) and from horses with infiltrative diseases, in which cellular infiltration of all layers is responsible for increased wall thickness (Reef 1998a; Scott et al. 1999; Dechant et al. 2008).
Reduced motility of the duodenum was also significantly associated with a definitive diagnosis involving SI in a multivariable analysis model. However, the duodenum, except in cases of proximal enteritis (Busoni et al. 2010), is rarely involved in simple or strangulated obstruction. When there is any other disease of the SI, the duodenum becomes involved later.
Finally, increased peritoneal free fluid was statistically associated with a definitive diagnosis involving SI as well as strangulating obstruction of SI. The presence of increased free peritoneal fluid is a consequence of vascular congestion, which results in increased endothelial permeability. If bowel wall ischaemia occurs, an exudative process develops and large quantities of protein and white blood cells move into the peritoneal cavity (Mair and Edwards 2003; Ross 2010). Peritoneal fluid changes associated with strangulating obstruction occur after approximately 1–2 h from the onset, depending on the type and severity of the lesion. Moreover, compartmentalisation can make identification of peritoneal fluid difficult (Mair 2002). Again, only horses in which the amount of free peritoneal fluid was subjectively considered to be increased were included in this study, without comparison with abdominocentesis. To our knowledge, the amount of free peritoneal fluid detectable in horses and the sensitivity and specificity of abdominal US for the detection of increased free peritoneal fluid have not been investigated in horses so far.
Interestingly, in this study SI loops were identified in 88.8% of horses, in agreement with Busoni et al. (2010), who identified SI loops in 75% of horses, when both medical and surgical cases were considered. In equine abdominal US examinations, the visualisation of the bowel in normal horses is considered to be limited by gas and ingesta in the LC (Norman et al. 2010). The frequency of visualisation of SI loops in the normal equine population has not been investigated yet, and it could be useful to investigate normal vs. colic-affected horses, as well as medical vs. surgical colic horses. Fasting, a consequence of colic (reduced appetite/withheld feed), could improve the ability to obtain images of the SI, as reported in a previous study of normal fasted horses (Norman et al. 2010).
The results of multivariable analysis of renosplenic entrapment demonstrate that failure to visualise the left kidney was highly significantly associated with this condition. In agreement with other authors (Santschi et al. 1993), in horses with renosplenic entrapment the presence of gas-filled colon dorsal to the spleen precluded imaging of the kidney. There are no published data on the sensitivity, specificity, PPV and NPV of this US finding, but in this study the values were 87, 83, 42 and 98%, respectively. The low PPV is a consequence of the number of false positives (18/143), and a similar finding was reported in a previous study (Busoni et al. 2010). In fact, a recent rectal examination often obscures visualisation of the left kidney (Reef 1998a); in our practice, rectal examination precedes abdominal US examination. Other authors recommend assessment of several ultrasonographic features for the diagnosis of left dorsal displacement of the LC (Santschi et al. 1993; Reef 1998a). Failure to visualise the left kidney was the only parameter assessed and analysed in this study. Even if increased free peritoneal fluid was also related to both renosplenic entrapment and right dorsal displacement, the OR was low in both conditions compared with the OR for SI. In fact, the sensitivity, specificity and PPV of increased free peritoneal fluid for these diseases were very low.
In the multivariable model for right dorsal displacement, a distended LC was significantly associated with this condition. In LC displacements, which are nonstrangulating conditions, because of the impaired flow of ingesta over time, gaseous distension becomes more significant and the distension of the LC is a common feature. With regard to renosplenic entrapment, it could be very interesting to evaluate whether there is a difference in US appearance of the LC in medical vs. surgical cases, since there are no published data. In agreement with Pease et al. (2004), a thickened LC was highly sensitive and specific, with good PPV and NPV. In contrast, the appearance of the LC was not correlated to nonstrangulating volvulus. An indicative alteration of strangulated LC volvulus is complete obstruction of venous drainage, which causes the LC wall to become oedematous and thickened (Snyder et al. 1989). In nonstrangulating obstruction (and in displacement) of the LC, the viscus could become distended with gas and ingesta but, because the vasculature is not compromised, the wall does not become thickened as in LC torsion (Hackett 1983). Absent motility of the LC had good sensitivity, specificity and high NPV for both strangulated and nonstrangulated volvulus, with better PPV for strangulated volvulus.
Limitation of the study
This study had some limitations. Only one trained veterinarian performed the US evaluation in each case and, because US findings are operator-dependent, this factor could have influenced the results. The localisation and the number of SI distended loops were not evaluated (Klohnen et al. 1996). On the standardised form, the presence of completely distended loops together with normal loops was not recorded, but these findings could be relevant in differentiating between simple and strangulating obstruction or intussusception.
The localisation of free peritoneal fluid was not recorded and this should be further investigated, since the fluid could be constrained in specific areas, especially in cases of LI diseases. Finally, with respect to the motility of both large and small intestine, sedation could have influenced the results; although sedation is not performed routinely, it was necessary in a few cases, and it was not considered in this study.
This study shows the role of ultrasonography in distinguishing between lesions of the small and large intestine. The study supports previous data obtained for definitive diagnosis involving SI, in particular strangulating obstruction (Klohnen et al. 1996; Klohnen 2008), and for strangulating LC volvulus (Pease et al. 2004). Thickened wall loops may be present in horses with obstruction, intussusception, and inflammatory bowel diseases. US examination is very sensitive for renosplenic entrapment, although false positive results were also identified. Finally, the appearance of the LC should be evaluated carefully: LC lesions could represent the primary disease even if abnormal SI loops are identified.
Authors' declaration of interest
None of the authors of this paper has a financial or personal relationship with other people or organisations that could inappropriately influence or bias the content of this paper.
The authors would like to thank Dr. Eleonora Lotto for her contribution with data collection.
All authors contributed to the initiation, conception, planning and execution of this study. The statistics were by F.B and the paper was written by F.B., M.P., M.C and S.N.
1 Medison, Clichy, France.
2 R Foundation for Statistical Computing, Vienna, Austria.