Criteria for Selection of Cases

The study population was derived from medical records of horses > 1 month of age that were admitted to the UGA-VTH from April 2000 through March 2003. Records were retrieved for horses with SCr concentrations of ≥ 3.0 mg/dL (reference interval, 0.7 to 2.2 mg/dL) at the time of admission. A record was excluded if the horse was not evaluated for a primary complaint of colic or colitis, if the duration of treatment at the UGA-VTH did not exceed 24 hours, or if the SCr concentration was not reevaluated prior to discharge. From these records, 79 horses met the study criteria and were analyzed as the study population. Within the study population, horses with high SCr concentrations (≥ 3.0 mg/dL) on initial evaluation that did not decrease to reference limits (ie, SCr concentration ≤ 2.2 mg/dL) after receiving fluid therapy during the first 72 hours of hospitalization were identified; there were 26 horses with persistent azotemia (PA group). Additionally, horses in the study population in which high SCr concentrations on initial evaluation did decrease to reference limits after receiving fluid therapy during the first 72 hours of hospitalization were also identified; there were 53 horses in which azotemia resolved after treatment (AR group). Because horses that did not have a second SCr concentration measurement performed were excluded from the study, the study population may have been more likely to include horses with severe gastrointestinal illness or other predisposing conditions than the overall UGAVTH population of horses that were evaluated for colic or colitis. Consequently, the distribution of a subset of study variables in the study population was compared with that in a sample of the UGA-VTH population of horses that were evaluated for gastrointestinal disease during the same time period. This sample included the randomly selected records of 88 adult horses that were evaluated for colic or colitis during the study period and is referred to as the hospital population.

Procedures

Records of study-population horses were analyzed for information regarding signalment, primary complaint, diagnosis, medication administration prior to referral, physical examination findings, diagnostic test results, clinicopathologic data, treatment, and outcome. Physical examination parameters included rectal temperature, pulse and respiratory rates, mucous membrane color, capillary refill time, the presence or absence of diarrhea or nasogastric reflux, and rectal examination findings. Diagnostic imaging data included results of abdominal or renal ultrasonographic examination. Clinical laboratory data included Hct; serum albumin, total protein, bicarbonate, sodium, chloride, potassium, and calcium concentrations; and urinalysis results, if available. Serum creatinine concentrations at initial evaluation and after discontinuation of treatment were also assessed, as was the interval (hours) to the first recheck evaluation of the SCr concentration. Following initiation of fluid therapy, the number of days to resolution of azotemia was recorded. Treatments administered prior to admission, including nonsteroidal anti-inflammatory drugs, aminoglycoside antimicrobial drugs, and IV administration of fluids, were assessed. Information regarding treatments administered following admission, including fluid therapy rate, occurrence of surgical exploration, medications administered, and specific treatments for acute renal failure, was also collected. Records from the hospital population were reviewed for signalment; primary complaint; initial SCr concentration; and the presence of gastric reflux, diarrhea, or an abnormal finding on rectal examination at initial evaluation. All data were entered into a commercially available computer spreadsheet.^a

Statistical analysis—All statistical analyses were performed by use of commercially available statistical software.^b Variable distributions were examined initially in the study and hospital populations to determine how well the study population reflected the comparable hospital caseload. A value of P ≤ 0.05 was considered significant. Significant differences among the independent variables between the study population and hospital population were evaluated by use of Mantel-Haenszel χ² tests or the Fisher exact test for categoric variables or t tests for continuous variables. Odds ratios were calculated by use of logistic regression to compare differences in aspects of signalment, history, and clinical findings between the study and hospital populations and between the PA and AR groups.⁷

Results

One hundred seventy-six horses that were evaluated at the UGA-VTH from April 2000 through March 2003 because of colic or colitis had a SCr concentration of ≥ 3.0 mg/dL. Seventy-nine of these horses met the criteria for inclusion in the study population. One thousand horses, including the study population, were evaluated at the UGA-VTH from April 2000 through March 2003 because of colic or colitis; 88 of these horses were randomly selected as a representative sample of the hospital population. The mean SCr concentration at initial evaluation in the horses included in the study population was significantly (P < 0.001) higher than that of the horses in the hospital population, confirming that the study population selection process was appropriate (Table 1). Grouping geldings and stallions together indicated that males were significantly (OR, 3.88; 95% CI, 1.98 to 7.61; P < 0.001) overrepresented in the study population, compared with the hospital population (61 [77%] males and 41 [47%] males, respectively). In regard to males, the proportion of stallions relative to geldings in the study population was higher than that in the hospital population (44 [72%] geldings and 17 [28%] stallions vs 36 [88%] geldings and 5 [12%] stallions), but this difference was not significant (P = 0.06). Quarter Horses composed a significantly (OR, 2.20; 95% CI, 1.17 to 4.15; P = 0.014) higher proportion of the study population (39/79 [49%] horses), compared with the hospital population (27/88 [31%] horses). Although Thoroughbreds were underrepresented in the study population (9/79 [11%] horses), compared with the hospital population (19/88 [22%] horses), the difference was not significant (P = 0.079).

In the study population, records indicated that 64 (81%) horses were evaluated at the UGA-VTH because of a primary complaint of colic and 15 (19%) horses were evaluated because of a primary complaint of colitis. In the hospital population, 84 (95%) horses were evaluated because of a primary complaint of colic and 4 (5%) horses were evaluated because of a primary complaint of colitis. Horses in the study population were significantly (P = 0.003) more likely to be evaluated because of a primary complaint of colitis than horses in the hospital population. Horses in the study population were also significantly more likely to have diarrhea (OR, 3.48; 95% CI, 1.44 to 8.45; P = 0.004) or nasogastric reflux (OR, 3.88; 95% CI, 1.97 to 7.67; P < 0.001) than horses in the hospital population. Regarding abnormal rectal examination findings at initial evaluation, there was no significant difference between the study and hospital populations.

The study population was allocated to 1 of 2 groups on the basis of whether the initial high SCr concentration returned to reference range (ie, azotemia resolved) within 3 days of hospitalization and fluid therapy. In 53 (67%) horses in the study population, azotemia resolved within 3 days of hospitalization and treatment for the primary complaint (AR group); in 26 (33%) horses, azotemia did not resolve within the first 3 days of hospitalization and treatment for the primary complaint (PA group). In the PA group, the mean initial SCr concentration was 5.2 mg/dL (range, 3.0 to 9.5 mg/dL); this value was significantly (P < 0.001) higher than that of the AR group (mean initial concentration, 3.9 mg/dL; range, 3.0 to 7.0 mg/dL). The mean final SCr concentration in the PA group was 3.0 mg/dL (range, 1.4 to 8.8 mg/dL); this value was significantly (P < 0.001) higher than that of the AR group (mean concentration after treatment, 1.9 mg/dL; range, 0.6 to 2.3 mg/dL). The mean number of days to resolution of azotemia in the PA group was 7.9 days (range, 4 to 22 days). Azotemia did not resolve in 15 of the 26 (58%) horses. Four horses received additional treatment for acute renal failure; all of these horses were in the PA group. Three of the 4 horses that received additional treatment for renal failure ultimately died or were euthanized.

The distribution and significance of bivariate associations for all variables associated with persistently high SCr concentration after 72 hours of fluid therapy were assessed (Table 2). Persistent azotemia was associated with a record of gastric reflux (OR, 1.72; 95% CI, 1.03 to 2.87; P = 0.05) and abnormal rectal examination findings (OR, 1.35; 95% CI, 1.08 to 1.68; P = 0.05). Low serum chloride concentration was also significantly associated with persistent azotemia (OR, 3.99; 95% CI, 1.41 to 11.26; P = 0.008). Eight (15%) horses in the AR group and 9 (35%) horses in the PA group had transabdominal ultrasonography performed that included examination of the kidneys. In the AR group, 4 of the 8 horses had abnormal findings in the renal parenchyma; in the PA group, 5 of the 9 horses had abnormal findings in the renal parenchyma. Horses in the PA group were 3 times as likely to die or be euthanized while hospitalized (OR, 3.11; 95% CI, 1.06 to 9.13; P = 0.03) as those in the AR group.

Although horses in the study population were more likely to have diarrhea or colitis than those in the hospital population, diarrhea or colitis was not significantly associated with prolonged azotemia. Other variables that were assessed but determined not to be associated with prolonged azotemia included signalment (age, breed, and sex); rectal temperature; pulse rate; respiratory rate; mucous membrane quality, including capillary refill time; administration of nonsteroidal anti-inflammatory drugs or aminoglycoside antimicrobials prior to referral; presence of diarrhea at initial evaluation; presence of a strangulating intestinal lesion; and abnormal serum bicarbonate, sodium, total protein, albumin, calcium, and potassium concentrations or Hct. Additionally, there was no difference in evidence of disease in other organ systems or in the number of horses that underwent surgical exploration between the PA and AR groups. The number of cases with urinalysis results was inadequate for comparisons of those data to be made between the PA and AR groups.

Discussion

The objective of the retrospective study reported here was to identify risk factors associated with renal insufficiency in horses with colic or colitis that were evaluated at a referral hospital. For the purposes of our study, renal insufficiency was defined as the failure of SCr concentration to return to the reference interval within 72 hours of initiation of therapy that included IV administration of fluids to correct dehydration and replace ongoing losses. Because measures of renal function such as urine specific gravity or urine fractional excretion of sodium are not typically measured prior to IV administration of fluids in horses admitted to the UGA-VTH for primary nonrenal problems and because these measurements are not reliable indicators of renal function once IV fluid therapy has been initiated, these tests could not be used to confirm renal dysfunction in the horses included in the study population. Failure of azotemia to resolve after appropriate fluid administration is an accepted means of differentiating prerenal from renal azotemia.^5,6,8,9 To our knowledge, this investigation is the first attempt to identify factors associated with the development of renal insufficiency in horses evaluated for a primary complaint of colic or colitis. Horses with renal insufficiency were 3 times as likely to die or be euthanized as horses with azotemia that resolved within 72 hours of treatment. Appropriate treatment for the primary gastrointestinal problem was likely an important factor in the recovery of horses with prerenal azotemia. However, we could not detect differences in the frequency of surgery, presence of other organ involvement, or fluid rates administered between horses in the PA and AR groups, so differences in treatments did not appear to influence whether azotemia resolved.

A drawback of designating the response to 72 hours of fluid therapy as a measure of renal dysfunction in horses evaluated for primary gastrointestinal disease is that 3 days have elapsed before the diagnosis of superimposed renal injury has been made. Confirmation of renal injury in these horses at initial evaluation would allow clinicians to provide a more informed prognosis to clients. Moreover, earlier diagnosis would also lead to earlier administration of specific treatments for renal failure such as furosemide, mannitol, or dopamine. Because persistent azotemia was associated with poor outcome in the present study, our findings suggest that efforts to diagnose renal dysfunction at the initial evaluation of horses with colic or colitis may be warranted.

Although the upper limit of the reference interval for SCr concentration in the UGA-VTH's clinical laboratory is 2.2 mg/dL, horses that were included in the study population were selected on the basis of initial SCr concentration of at least 3.0 mg/dL, as indicated in the medical records. This higher cutoff value was chosen to select a manageable number of cases; it was anticipated that the use of this cutoff value would ensure that the study population was likely to contain a substantial subset of horses with renal insufficiency. However, a weakness of this approach is that any horses with renal insufficiency and SCr concentration at initial evaluation of 2.2 to 2.9 mg/dL were excluded. Because analysis of the study data ultimately revealed that SCr concentration at initial evaluation was significantly lower in horses in the AR group, it seems likely that our selection process excluded relatively few cases of interest.

Although risk factors associated with outcome in horses with primary gastrointestinal disease and concurrent renal injury may not have been previously investigated, renal insufficiency in horses with colic has been recognized for some time.² Moreover, high SCr concentration at initial evaluation has been associated with poor outcome in previous studies that describe clinical or laboratory findings associated with outcome in horses with colic¹⁰ or colitis.^3,11 Cohen and Woods¹¹ determined that an initial SCr concentration > 2.0 mg/dL was associated with failure to survive in horses with acute diarrhea. In horses with small intestinal volvulus, high SCr concentration had a tendency to be associated with nonsurvival.¹⁰ Findings of the present study extend these reports by indicating that, among horses presenting with primary gastrointestinal disease and high SCr concentration at initial evaluation, the initial SCr value was significantly higher in horses that ultimately failed to respond to 72 hours of fluid replacement. This information may aid clinicians in the development of an accurate prognosis for azotemic horses that are evaluated because of primary gastrointestinal disease. Also, in previous reports,^10,11 the influence of prerenal versus renal causes of azotemia on prognosis was not differentiated, as we have attempted to do by reporting outcome in horses with renal insufficiency.

The overrepresentation of male horses in the study population could not be easily explained. Serum creatinine concentrations are determined both by creatinine production and renal excretion. Our finding that male horses were not more likely to have persistently high SCr concentrations suggests that the SCr concentration was higher at initial evaluation in males because of increased creatinine production, possibly as a result of hormone-mediated influences on muscle metabolism. Likewise, Quarter Horses may also have been overrepresented in the study population as a result of increased muscle mass in this breed, compared with other breeds.

In the study population, more horses were evaluated because of a primary complaint of colitis and had a concurrent physical examination finding of diarrhea, compared with the hospital population; however, the presence of diarrhea or colitis was not significantly associated with prolonged azotemia. It was expected that colitis or diarrhea would be more common in the PA group because these conditions can be associated with hypovolemia and endotoxemia and the subsequent decreased renal perfusion and development of azotemia.³ It is not clear why horses with prolonged azotemia were more likely to have abnormal amounts of gastric reflux recorded.

On analysis of medical record data, hypochloremia was the only serum electrolyte abnormality associated with persistent azotemia in the study population. Because persistent azotemia was associated with a record of abnormal reflux, serum hypochloremia may have been linked to loss of chloride via nasogastric reflux. However, it is noteworthy that hypochloremia is the most common electrolyte abnormality in horses with renal failure.¹² It may be that hypochloremia at initial evaluation was an early indicator of renal dysfunction in horses in the study population. Diarrhea can also cause hypochloremia, and although diarrhea was not associated with persistent azotemia, it may have been a contributing cause of hypochloremia in some of the horses in the PA group.

Compared with the AR group, horses in the PA group were more likely to have received treatment for acute renal failure, including administration of furosemide, mannitol, or dopamine. However, of the 4 horses in the PA group that received treatment for acute renal failure, only 1 survived. Although the number of horses receiving treatment for acute renal failure was small, treatment for acute renal failure did not improve outcome in horses with persistent azotemia. The treatment may have been initiated too late to be successful; alternatively, horses may have been euthanized because of financial constraints once treatment for renal failure was recommended.

Transabdominal ultrasonographic examination, including evaluation of the kidneys, is often used to evaluate horses with gastrointestinal disease. In our study, although horses in the PA group were more likely to undergo transabdominal ultrasonographic examination than horses in the AR group, they did not have an increased incidence of renal ultrasonographic abnormalities. Collection and histologic examination of renal biopsy specimens were not performed in any of these horses; thus, we were unable to determine whether horses with renal abnormalities detected via ultrasonographic examination also had histologic evidence of renal disease.