• Open Access

Clinical Features and Prognostic Variables in 109 Horses with Esophageal Obstruction (1992–2009)

Authors


  • All data collection was obtained from medical records at the Equine Hospital at Colorado State University, Fort Collins, CO.

Corresponding author: Diana M. Hassel, Department of Clinical Sciences, College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, CO 80523; e-mail: dhassel@colostate.edu.

Abstract

Background: Esophageal obstruction is common in horses and can result in life-threatening complications. Previous studies have described clinical findings in horses with esophageal obstruction, but there are no reports that attempt to make correlations of clinical findings with outcome.

Hypothesis: Specific clinical features of horses with esophageal obstruction are associated with increased likelihood of complications.

Animals: One hundred and nine horses with esophageal obstruction.

Methods: Retrospective cross-sectional study. All clinical records of horses admitted between April 1992 and February 2009 for esophageal obstruction were reviewed. The association among 24 clinical, hematological, biochemical, therapeutic variables and the likelihood of complications was investigated by a univariable logistic regression model, followed by multivariable analysis.

Results: Multiple logistic regression analysis revealed that intact males (P= .02), age >15 years (P < .01), and a need for general anesthesia (P < .01) were associated with the development of complications after an episode of esophageal obstruction. Increased respiratory rate (>22 breaths/min) and moderate or severe tracheal contamination, although not associated with complications as a whole, significantly increased the risk of developing aspiration pneumonia (P≤ .01).

Conclusions and Clinical Importance: Signalment, clinical variables, and endoscopic findings were confirmed as important tools in assessing the severity of the esophageal lesion and pulmonary involvement. Knowledge of risk factors for the development of complications will aid in making informed decisions to optimize treatment and assist in the assessment of prognosis.

Abbreviations:
HR

heart rate

NSAID

nonsteroidal anti-inflammatory drugs

OR

odds ratio

RR

respiratory rate

TP

total protein

Esophageal obstruction or “choke” is a common clinical presentation in the horse; most frequently it is caused by feed impaction. Suggested causes for simple obstruction include ingestion of inadequately soaked sugar beet pulp, ingestion of apples or carrots, excessively rapid ingestion of dry fibrous, pelleted, or cubed feedstuffs, inadequate mastication from poor dentition, or the swallowing of a foreign body.1,2 Other reported causes include inadequate water intake, eating when heavily sedated, and esophageal disease including functional neuromuscular disorders, stenosis/strictures, diverticula, neoplasia, abscesses, and cysts.3,4

Horses with persistent esophageal obstruction often become compromised as a consequence of dehydration, acid-base and electrolyte imbalances, and aspiration pneumonia.5 Complications are common and may include esophageal mucosal ulceration, stricture formation, esophageal perforation, aspiration pneumonia, chronic recurrent obstruction, postoperative infection, pleuritis, laminitis, laryngeal paralysis, and Horner's syndrome.6

There are few published studies describing the clinical findings in horses with esophageal obstruction and these mainly consist of smaller numbers of horses. They also fail to make correlations of clinical findings with outcome. The purpose of the present retrospective cross-sectional study was to describe clinical features of a population of horses presented to a referral institution with esophageal obstruction, and to assess the association between analyzed variables and the likelihood of complications.

Materials and Methods

The clinical records of horses admitted for esophageal obstruction to the Equine Hospital at Colorado State University between April 1992 and February 2009 were analyzed. Variables included breed, sex, age, temperature, heart rate (HR), and respiratory rate (RR) at admission, whether or not thoracic or esophageal radiographs were performed, and radiographic findings. Radiographic evidence of alveolar or bronchoalveolar densities in the caudoventral region of the lungs was considered a sign compatible with aspiration pneumonia secondary to esophageal obstruction. When endoscopy was performed, tracheal contamination with food material was subjectively graded as described previously2 as absent, when no visible particles were detected, mild, when there were single food particles in the trachea, moderate, when small amounts of aspirated fluid were detected on the tracheal floor, or severe, when large amounts of aspirated fluid were detected on the tracheal floor, food particles, or both were detected in the dorsal aspect along the trachea. Visible esophageal lesions were classified based on the endoscopic record as absence of irritation or swelling (normal), mild to moderate inflammation or erosion, or severe erosion, ulceration, or morphological abnormalities. The location of the obstruction was identified as proximal (<60 cm), middle (61–90 cm), or distal (>90 cm) third of the esophagus as determined endoscopically, measured as centimeters from the nares, and via clinical examination, taking as a reference an adult Warmblood and proportionally adapting the measurement to smaller horses or foals.

Hematologic and biochemical variables at the time of admission included PCV, total protein (TP), blood glucose, bicarbonate, and anion gap.

Recorded treatments included whether antibiotics were administered and the route of administration (oral, parenteral, or both). Additional drug therapies documented included the use of nonsteroidal anti-inflammatory drugs (NSAIDs), α2 agonists, opioids, and oxytocin. Whether or not general anesthesia was required for resolving the obstruction was recorded.

The duration of the esophageal obstruction before admission (<3 hours, between 3 and 6.3 hours, between 6.4 and 12.3 hours, between 12.4 and 48 hours, or more than 48 hours in duration) was established. The time between admission and resolution of the obstruction was classified as (a) relieved spontaneously by admission, (b) during initial treatment, (c) within 24 hours, (d) more than 24 hours, or (e) never resolved.

For the outcome variable, an animal showing any complication related directly to the esophageal obstruction was considered as “having complications” and these were specifically recorded.

A descriptive analysis of all the recorded variables and their association with the outcome was undertaken. The association between these variables and the outcome was assessed by logistic regression analysis. Continuous variables were evaluated for linear relation with the log odds of the outcome by the Lowess smoothing function and whenever the assumption was not met they were categorized based upon biological reasoning and distribution of the population. Variables with a univariable P≤ .25 were included into the multivariable model, which was constructed manually. Significance was set at P≤ .05. If a variable affected more than 20% of the coefficient of another variable, it was considered a confounder and was forced into the model. We checked all biologically plausible 2-way interactions. The Hosmer-Lemshow statistic was used to assess the overall fit of the model. All the analyses were conducted with Stata version 10.a

Results

The dataset consisted of 109 records from horses admitted to the Equine Hospital at Colorado State University for esophageal obstruction between April 1992 and February 2009. Only 5 records contained a complete collection of all variables of interest. Of these 109 horses, 56 (51.4%) developed complications including aspiration pneumonia (39), esophageal stenosis/stricture (8), fever (4), esophageal diverticula (4) or rupture (2), kidney failure (2), diarrhea (2), mild esophagitis (1), laryngeal hemiplegia (1), esophageal necrosis (1), laminitis (1) or a combination of these, or the obstruction was not resolvable because of undetermined causes (7). Thirteen horses (11.9%) died or were euthanized at the hospital as a consequence of the obstruction. Descriptive statistics for variables significantly associated with the outcome as determined via univariable analysis are summarized in Table 1. Because pneumonia alone represented almost 70% of the overall complications, its association with each variable of interest was further investigated and results are summarized in Table 2.

Table 1.   Descriptive statistics and univariable analysis of variables showing an association (P≤ .25) with the likelihood of developing complications after esophageal obstruction.
VariableN (%)N Outcomes (%)OR95% CILR χ2P-Value
  1. N, number of horses; OR, odd ratio; CI, confidence interval; LR χ2, likelihood ratio chi-squared; Ref., reference value; RR, respiratory rate; radiographs AP, radiolographic signs of aspiration pneumonia; TP, total protein.

Sex10956 (51.4)  5.3.07
 Female36 (33.1)17 (47.2)Ref.
 Gelding56 (51.4)26 (46.4)10.4–2.2 .90
 Male17 (15.6)13 (76.5)3.61–13.3 .05
Age (years)10956 (51.4)  13.5<.01
 <117 (15.6)12 (70.6)5.81.7–20.1 <.01
 1–1541 (37.6)12 (29.3)Ref.
 >1551 (46.8)32 (62.7)4.11.7–9.8 <.01
Temperature (C°)9152 (57.1)  2.9.23
 ≤3714 (15.4)7 (50.0)Ref.
 37.1–3835 (38.5)17 (48.6)0.90.3–3.3 .90
 >3842 (46.1)28 (66.7)20.6–6.8 .30
RR (bpm)9152 (57.1)  3.9.14
 ≤1234 (37.4)15 (44.1)Ref.
 13–2225 (27.5)17 (68.0)2.70.9–7.9 .07
 >2232 (35.2)20 (62.5)2.10.8–5.6 .13
Radiographs AP3828 (73.7)  3.9.05
 No13 (34.2)7 (53.8)Ref.
 Yes25 (65.8)21 (84.0)4.51–20.7  
Tracheal contamination4528 (62.22)  6.9.07
 None17 (37.8)7 (41.2)Ref.
 Mild5 (11.1)3 (60)2.10.3–16.4 .46
 Moderate18 (40)15 (83.3)7.11.4–34.4 .01
 Severe5 (11.1)3 (60)2.10.3–16.4 .46
Grade esophageal lesion7040 (57.14)  12.7<.01
 Absent16 (22.9)4 (25.0)Ref.
 Moderate29 (41.4)16 (55.2)3.70.9–14.2 .06
 Severe25 (35.7)20 (80.0)122.7–53.6 <.01
TP (g/L)6441 (64.06)  5.5<.01
 ≤7024 (37.5)11(45.8)Ref.
 >7040 (62.5)30 (75.0)1.21–1.4  
Antibiotics7137 (52.11)  6.5.04
 Oral13 (18.3)3 (23.1)Ref.
 Parenteral20 (28.2)10 (50.0)3.30.7–15.8 .13
 Both38 (53.5)24 (63.2)5.71.3–24.3 .02
General anesthesia10956 (51.4)  14.6<.01
 No69 (63.3)26 (37.7)Ref.
 Yes40 (36.7)30 (75.0)5.02.1–11.8  
Duration (hours)7843 (55.13)  9.7.05
 <311 (14.1)3 (27.3)Ref.
 3–6.39 (11.5)4 (44.4)2.10.3–13.8 .43
 6.4–12.318 (23.1)8 (44.4)2.10.4–10.8 .36
 12.4–4818 (23.1)11 (61.1)4.20.8–21.4 .08
 >4822 (28.2)17 (77.3)9.01.7–47.7 <.01
Table 2.   Results of descriptive statistics and univariable analysis of variables showing an association (P≤ .05) with the likelihood of developing pneumonia after esophageal obstruction.
VariableN (%)N Outcomes (%)OR95% CILR χ2P-Value
  1. OR, odd ratio; CI, confidence interval; LR χ2, likelihood ratio chi-squared; Ref., reference value.

Sex10939 (35.8)  7.9.02
 Female36 (33.1)9 (25.0)Ref.
 Gelding56 (51.4)19 (33.9)1.50.6–3.9 .36
 Male17 (15.6)11 (64.7)5.51.6–19.2 <.01
Age (years)10939 (35.8)  8.4.02
 <117 (15.6)9 (23.1)4.61.4–15.8 .01
 1–1541 (37.6)8 (20.5)Ref.
 >1551 (46.8)22 (56.4)3.11.2–8.1 .02
RR (bpm)9136 (39.6)  10.7<.01
 ≤1234 (37.4)7 (20.6)Ref.
 13–2225 (27.5)10 (40)2.60.8–8.1 .12
 >2232 (35.2)19 (59.4)5.61.9–16.8 <.01
Tracheal contamination4521 (46.7)  10.7.01
 None17 (37.8)4 (23.5)Ref.
 Mild5 (11.1)1 (20.0)0.80.1–9.5 .80
 Moderate18 (40)12 (66.7)6.51.5–28.8 .01
 Severe5 (11.1)4 (80.0)131.1–152.3 .04
TP (g/L)6431 (48.4)  8.7<.01
 ≤7024 (37.5)6 (25.0)Ref.
 >7040 (62.5)25 (62.5)5.01.6–15.4  
General anesthesia10939 (35.8)  10.1<.01
 No69 (63.3)17 (24.6)Ref.
 Yes40 (36.7)22 (55.0)3.71.6–8.6 <.01

Animal Level Variables

The cases ranged between 2 months and 35 years of age (median 13 years) and included 36 females (33.1%), 56 geldings (51.4%), and 17 intact males (15.6%). There was no significant difference among sex between the hospital population and our sample population over the same time period (χ2= 4.51, df= 3, P > .1). Breeds consisted of 45 Quarter Horses (41.3%), 17 Arabians (15.6%), 5 Thoroughbreds (4.6%), 8 ponies (7.3%), and 34 other breeds (31.2%). This breed distribution differed significantly from the hospital population over the same time period, with Thoroughbreds being significantly underrepresented, and Arabians and ponies being significantly overrepresented (χ2= 26.11, df= 4, P < .001).

Clinical Variables

Body temperature on admission ranged between 35.6 and 40.6°C (median 37.9°C). The HR and RR ranged between 30 and 100 beats/min and 8–66 breaths/min, respectively (median 56 beats/min and 24 breaths/min, respectively). Signs consistent with aspiration pneumonia were found in 25 of the 38 horses (65.8%) that underwent thoracic radiography at admission. Tracheoscopy was performed on 45 horses (41.3%). Esophagoscopy was performed on 70 horses (64.2%). Anatomical abnormalities included 8 cases of esophageal stricture. Two strictures were located in the proximal one third of the esophagus, 2 in the middle third, 2 in the distal third, and 2 had an undocumented location. Two of these were associated with poststenotic diverticula, 2 with esophageal scars, and 1 with a paraesophageal abscess in the caudal retropharyngeal region. One half of all cases of stricture were foals (4/8) <1 year of age. Other esophageal disorders associated with the esophageal obstruction ranged from moderate circumferential erosion (15) to esophageal ulcers (8), ranging in length from 5 cm to the entire esophagus. Two cases of distal esophageal hyperplasia were diagnosed via subjective assessment at endoscopy or during postmortem examination. Fifteen out of 59 obstructions (25%) were located in the middle third of the esophagus, 22 (37%) in the proximal and distal thirds, respectively (χ2= 1.66, df= 2, P > .1). For the other 11 horses undergoing endoscopy, the location of the obstruction was unrecorded.

Hematological and Biochemical Variables

On admission, the PCV ranged between 23 and 59% (median 34%). The mean TP was 73 ± 10.4 g/L (mean ± SD, range 34–99). Glucose measured at admission ranged between 3.3 and 16.3 mmol/L (median 7.8 mmol/L). Bicarbonate ranged between 13.7 and 30.6 mmol/L (median 25.6 mmol/L) and the anion gap between 9 and 35 mmol/L (median 15 mmol/L).

Treatment-Related Variables

Of the 71 horses for which antibiotic therapy was recorded, 13 (18%) received oral antibiotics only, 20 (28%) received parenteral antibiotics only, and 38 (54%) received both oral and parenteral antibiotics. Sixty-one out of 107 horses (57%) were treated with NSAIDs, 87 (81%) received α2 agonists, and 59 (55%) received opioids. All but one horse received opioids in association with an α2 agonist. Twenty-six (24%) were treated with acepromazine and 14 (13%) horses received one or more doses of oxytocin (0.11–0.22 IU/kg IV).

Other Variables

Out of 78 horses for which the duration of the choke was recorded, 11 (14%) were referred to the hospital within 3 hours from recognition of the choke episode, 9 (12%) between 3.1 and 6.3 hours, 18 (23%) between 6.4 and 12.3 hours, 18 (23%) within 24 hours, whereas 22 horses (28%) suffered chronic obstruction (more than 48 hours in duration). Eleven (50%) of these 22 horses showed severe esophageal lesions or anatomical abnormalities endoscopically compared with 9 (16%) out of the overall 56 referred with a shorter duration of obstruction (χ2= 5.5, df= 1, P= .02). The majority (32/61) of obstructions were resolved during initial attendance. Thirteen cases (21%) were not resolvable and were euthanized. Because euthanasia was inevitably performed on all horses with irresolvable obstructive lesions, these horses were categorized as “never resolved” and removed from further analysis because it was lacking in statistical variability.

Univariable and Multivariable Logistic Regression Analysis

Age, presence of radiographic signs of aspiration pneumonia, grade of esophageal lesion as detected endoscopically, total plasma protein, type of antibiotics used, whether general anesthesia was required for resolution of the obstruction, and duration of the obstruction before referral had significant associations with the development of complications (P≤ .05) via univariable analyses (Table 1). Breed, temperature, HR, whether the obstruction was in the proximal, middle, or distal third of the esophagus, PCV, glucose, bicarbonate, anion gap, and the time elapsed between admission and resolution were not associated with outcome. RR and grade of tracheal contamination, although not associated with complications as a whole, were significantly associated with the specific risk of developing aspiration pneumonia (Table 2). Sex, age, and whether or not general anesthesia was needed in order to solve the obstruction had complete observations and were thus included in the multivariable model.

The final model was significant (P < .01) (Table 3). When considering all the variables in the model, intact males were almost 7 times more likely to develop complications than females (odds ratio [OR] = 6.6, 95% confidence interval [CI] 1.4–31.6, P= .02), senior horses (>15 years) were more than 6 times more likely to develop complications than younger adult horses (>1 year and <15 years) (OR = 6.2, 95% CI 2.2–17.6, P < .01), and the requirement of general anesthesia for solving the obstruction increased the likelihood of developing complications 5 times (OR = 5.1, 95% CI 1.9–13.4, P < .01). Overall, the model fits the data well based on the Hosmer-Lemeshow statistic (χ2= 3.2, df= 6, P= .78) and had a good predictive value (AUC ROC 0.79). The specificity and sensitivity of the model were 71.7 and 66.1%, respectively.

Table 3.   Results of multivariable analysis for the identification of risk factors for complications after esophageal obstruction.
VariableCoeff.ORSE95% CIPartial LR χ2P-Value
  1. Coeff., coefficient; OR, odd ratio; SE, standard error; CI, confidence interval; LR χ2, likelihood ratio chi-squared; Ref., reference value.

Intercept−2.0 
Sex    6.3.04
 FemaleRef. 
 Gelding0.31.40.70.5–3.7 .52
 Males1.96.65.31.4–31.6 .02
Age (years)    14.2<.01
 ≤11.44.03.00.9–17.6 .07
 1–15Ref. 
 >151.86.23.32.2–17.6 <.01
General anesthesia1.65.12.51.9–13.412.1<.01

Discussion

The short term survival rate and the complications associated with esophageal obstruction were in accordance with earlier findings.6 Previous studies reported that aspiration pneumonia is one of the most frequent complications after esophageal obstruction and a cause of mortality.6,7 This was confirmed by the present study, as aspiration pneumonia represented nearly 70% (39/56) of overall complications. Other complications were sporadic and often associated with pneumonia. According to our study, the risk of developing aspiration pneumonia was positively associated with RR at admission. Although an increased RR can occur because of several reasons such as pain or stress, this could also be an indicator of early pulmonary impairment, as suggested by our results, in which a RR over 22 breaths/min increased the risk (odds) of developing pneumonia nearly 6-fold, with respect to a rate of 12 breaths/min or less. Feige et al2 suggested that clinical and radiographic evaluation of the lungs is of value whenever respiratory impairment is suspected. In our study we demonstrated a significant association of radiographic signs compatible with aspiration pneumonia with the subsequent development of complications. Interestingly, radiographic signs were not directly associated with pneumonia. However, this may have been a spurious finding because of the smaller number of horses when only aspiration pneumonia was considered as an outcome. Unlike Feige et al,2 we found that the extent of tracheal contamination was proportionally associated with the subsequent development of pneumonia.

In our study, the grade of the esophageal lesion was strongly associated with the development of complications. This confirms the endoscopic exam as a valuable diagnostic aid for the diagnosis of mucosal impairment. Endoscopy and contrast radiography of the esophagus should be strongly considered in cases of chronic (>48 hours) or recurrent obstruction,2,6–8 as anatomic abnormalities are more likely to be present. The 3 main anatomic abnormalities encountered were ulcers, strictures, and esophageal diverticula. Half of all cases of stricture were foals, in accordance with earlier findings.6 The etiology of stricture in foals is not completely clear; it may be a congenital or developmental disorder, reflect a neurological or muscular developmental problem, or be the consequence of circumferential ulceration from a previous obstruction.9–11 Whereas some of these stenoses documented in foals have been reported to resolve after medical management, the survival rate is higher when treated surgically.6,9

Unlike a previous report,7 we did not find any significant difference in the prevalence of the location of the obstruction, nor could we identify any association between the location and the complication rate.

Between the hematological and biochemical variables, only TP > 70 g/L was associated with an increased likelihood of complications and pneumonia in particular. Perhaps this is related to either dehydration or hyperglobulinemia at presentation, possibly reflecting a more severe disease process.

Because our study population included cases collected over 17 years, the treatments differed widely and made it difficult to make inferences about the protocols used. Initial treatment for acute esophageal obstruction is often conservative, as many will resolve spontaneously or with medical management consisting of sedation, smooth muscle relaxants, and analgesics or anti-inflammatory drugs.1,4,12 Tranquilization and gentle pressure on the obstruction with a nasogastric tube is often necessary to promote passage of the obstruction into the stomach. General anesthesia may be necessary in some refractory cases, and surgical intervention is very rarely required.5 Oxytocin (0.11–0.22 IU/kg IV) has been reported to induce transient relaxation of esophageal musculature in healthy horses and resolution of choke in 8 of 10 horses.13 However, other investigators have not been able to corroborate these findings using manometric evaluations of the esophagus in the absence of balloon distention of the esophagus.14 In a previous study, detomidine, acepromazine, and a combination of xylazine and butorphanol have been reported to have potential detrimental effects on the resolution of the esophageal obstruction because of the reduction in swallowing and the change in normal peristaltic activity with increases in high-pressure events at the thoracic inlet.14 However, the combination of xylazine and butorphanol was commonly used in this population of horses. In vitro studies suggest oxytocin induces relaxation of only the smooth muscle portion of the esophagus.15 In our study the number of horses treated with acepromazine or oxytocin was too low and the number of those treated with α2 agonists too high to make any statistical inferences.

Only 13 out of 71 horses for which antibiotic therapy was recorded received oral antibiotics exclusively. Horses receiving antibiotic therapy seemed to have a trend toward developing pneumonia (P= .08) and those receiving both oral and parenteral antibiotics had a higher risk (odds) of developing complications (P= .02). This may have been related to the fact that those horses were considered at greater risk and subsequently received more aggressive and more prolonged antimicrobial therapy.

The duration of esophageal obstruction before admission is a significant risk factor for aspiration pneumonia, because the longer the obstruction, the longer the dysphagia, and consequently the risk of tracheal contamination and aspiration of saliva and food.2 Unlike earlier findings, we did not find any association between the duration of the obstruction and the development of aspiration pneumonia; however, horses with a history of chronic obstruction had 9 times higher odds (OR 9.0, 95% CI 1.7–47.7, P < .01) of developing complications compared with horses with an episode of obstruction <3 hours in duration.

The sex and the age of the animals, and the use of general anesthesia in order to resolve the obstruction were significant in the final multivariable logistic regression model. Although we found that intact males were more likely to develop complications compared with females, only 17 intact males were included in the sample population, so this increased risk for developing complications may have represented a spurious finding or may have been a result of unknown biologic differences. General anesthesia was highly associated with the development of complications (specifically with aspiration pneumonia) in both the univariable and multivariable models. General anesthesia increases the risk of postoperative fatalities,16,17 and it is likely that the most refractory cases needed general anesthesia for resolution of the obstruction.

Our findings may have been affected by changes in protocols over time and by the high number of variables excluded from the final model because of missing values. The model was not very accurate at predicting the future onset of complications in a horse experiencing an episode of esophageal obstruction because of the low sensitivity (66.1%) and specificity (71.7%) of the final model. Only sex, age, and whether or not the horse underwent general anesthesia fit the criteria for inclusion in the multivariable model. However, clinical variables along with radiographic and endoscopic results were confirmed as important tools in assessing the severity of the esophageal lesion and pulmonary involvement. These procedures should be applied in cases with recurrent esophageal obstruction because of the likelihood of preexisting anatomic or functional esophageal disorders. Two weaknesses in our study were the high number of missing values and the variability in procedures and treatments between different clinicians over the years. This is meant to be a preliminary study. A prospective study including all variables of interest, including clinical variables along with complete radiographic and endoscopic findings and the methods used to provide relief of the obstruction (eg, single tube, pressure lavage, gravity lavage, etc.), not recorded in our population, may provide a real benefit to practitioners and be desirable in the future.

Footnote

aStata/IC 10.1 for Windows, StataCorp LP, College Station, TX

Acknowledgment

The authors thank Dr Francisco Olea-Popelka for assistance with statistical analysis.

Ancillary