Introduction

Septic peritonitis is a commonly treated condition in the veterinary critical care unit with a classically high mortality rate, although newer reports have shown a more favorable prognosis with 64–85% survival.^1–6 Negative prognostic indicators include hypoalbuminemia, intraoperative hypotension, and failure of lactate clearance.^2,5,6 Animals with septic peritonitis often have prolonged anorexia postoperatively and have frequent gastrointestinal signs, including vomiting or regurgitation.⁷ Nutritional support is often considered in patients with septic peritonitis, due to the high incidence of anorexia in this population as well as the association between hypoalbuminemia and poor outcome.

The current human literature broadly suggests a benefit to enteral nutrition in the critically ill population across a variety of disease processes, especially when initiated early.^8–10 In the human and veterinary literature, early enteral nutrition (EEN) is defined as initiation of nutritional support less than 24 hr from the time of admission or onset of a hypermetabolic insult.^8–15 When compared to initiation of feeding more than 72 hr after admission, early initiation has been associated with less gut permeability and gastrointestinal protein loss, diminished activation and release of cytokines, and reduced systemic endotoxemia in people.^12,16

Postoperative feeding in humans is recommended within 24 hr by the European Society for Parenteral and Enteral Nutrition.¹⁷ Furthermore, early postoperative enteral nutrition (EN) has been shown to increase wound strength and collagen content in a canine model of enterotomy healing.¹⁸ EEN specifically for postoperative human patients is associated with a trend towards decreases in dehiscence, mortality rates, and length of hospitalization.^19–21 A recent veterinary retrospective study confirmed a decrease in hospitalization by 1.6 days in dogs with septic peritonitis that voluntarily ate, were given nutrition through a feeding tube, or received parenteral nutrition within the first 24 hr after surgery.⁷

Hesitation in initiating EEN generally is due to concern for increasing the incidence of gastrointestinal complications (GICs), aspiration pneumonia, ischemic bowel, and dehiscence of a gastrointestinal surgical site. The most common GICs noted are vomiting, regurgitation, and high gastric residual volume.^22,23

The objective of this study was to evaluate the safety of initiation of EEN for canine postoperative septic peritonitis patients compared to those with late enteral nutrition (LEN) or no enteral nutrition (NEN). Our hypothesis was that EEN would not cause additional gastrointestinal complications as compared to LEN and NEN.

Materials and Methods

The medical records database of the Michigan State University Veterinary Medical Center was searched for all dogs diagnosed with septic peritonitis between January 2004 and September 2011 that underwent an exploratory laparotomy. Septic peritonitis was diagnosed when intracellular bacteria were identified by a boarded pathologist on microscopic evaluation of abdominal fluid or when there was bacterial growth on aerobic or anaerobic culture of the abdominal fluid. Dogs were excluded if they did not have a definitive diagnosis of septic peritonitis, survived less than 24 hr postoperatively, or had incomplete medical records.

Initiation of enteral nutrition was considered as the time a dog began to eat on its own or began tube feedings and will be called “initiation of nutrition” for the remainder of the paper. To be considered significant oral intake, a dog had to ingest food multiple times on its own, although caloric intake was not able to be determined in the medical records. Initiation of tube feeding was considered the time that a diet was first given through the feeding tube, whether as a bolus or continuous infusion, and continued for at least 6 hr or two bolus doses. Syringe force feeding was not included as either oral intake or tube feeding. EEN was defined as initiation of nutrition within 24 hr from the start of surgery. LEN was defined as initiation of nutrition greater than 24 hr from the start of surgery. NEN was defined as those dogs that did not receive nutrition during hospitalization.

Data collected from the medical records included signalment (age and sex), body weight, and physical exam findings (temperature and pulse) on presentation. Pre-admission clinical signs of vomiting, regurgitation, diarrhea, and anorexia were also recorded, and the duration of signs prior to presentation was identified. The surgery time and date were documented, as well as the surgery type and source of sepsis. Information on EN support was collected, including time of tube placement, type of tube, and time of initiation. GICs of vomiting/regurgitation and diarrhea were recorded for each animal. Any signs consistent with aspiration pneumonia were noted. The date the animal began to eat on its own consistently was recorded as well. The use of parenteral nutrition and prokinetics was noted. Outcome measures of length of hospitalization of survivors, suspected dehiscence, and survival to discharge from the hospital were assessed. Dehiscence was suspected if new intracellular bacteria were identified on microscopic evaluation of postoperative abdominal fluid cytology or confirmed if dehiscence was identified at a subsequent surgery or necropsy. Non-survivors were animals that were euthanized or died in hospital.

Results

One hundred and twenty dogs that were presented to the facility during the study period had a diagnosis of septic peritonitis. Of these, 52 did not go to surgery, seven were euthanized on the surgery table, and five did not survive at least 24 hr postoperatively. Of the 56 dogs that had an abdominal exploratory performed and survived 24 hr postoperatively, 16 received EEN, 27 dogs received LEN, and 13 had NEN.

The median age on presentation was 8.63 yr (range 0.25–15.25), which was not significantly different between groups (P = 0.267). Thirty-six dogs were male (25 neutered, 11 intact) and 20 were female (16 spayed, 4 intact). Dogs' weights were similar between groups, with a median weight of 39.8 kg (5.45–81.8). Admission temperature was not different between groups (P = .127), with a median of 102.5°F (99.2–107.7). Similarly, admission pulse rate was not different between groups (P= .567), with an overall median pulse of 156 beats per minute (range 60–260).

Seventy-five percent (42/56) of cases were septic due to a gastrointestinal perforation, with 35.7% (15/42) of these from dehiscence of a previous gastrointestinal (GI) surgery, 23.8% (10/42) from a perforated gastrointestinal foreign body, 23.8% (10/42) from unclassified GI disease, 9.5% (4/42) associated with GI neoplasia, and 7.1% (3/42) associated with the use of non-steroidal anti-inflammatory drugs. Fourteen percent of cases (8/56) were secondary to a urogenital source, with 25% of these each from a pyometra (2/8), bladder rupture (2/8), renal abscess (2/8), or prostatic abscess (2/8). One dog in the EEN group had a splenic abscess as the source. Additionally, in the LEN group alone, two dogs had hepatic abscesses, two dogs had penetrating abdominal trauma, and one dog had a pancreatic abscess. There was no statistical difference in the occurrence of sources of septic peritonitis between groups. Gastrointestinal signs were common presenting complaints among all groups, with vomiting and regurgitation present in 66% (37/56) of dogs and diarrhea in 25% (14/56). Within the EEN group, 75% (12/16) had historical vomiting/regurgitation and 25% (4/16) had diarrhea, compared to 55.2% (16/27) with vomiting/regurgitation and 13.8% (4/27) diarrhea in the LEN group and 69.2% (9/13) vomiting/regurgitation and 46.2% (6/13) diarrhea in the NEN groups. The pre-admission occurrence of vomiting/regurgitation and diarrhea was not statistically different between groups. Ninety-six percent of dogs were anorexic on presentation (93.8% [15/16] of EEN, 90% [26/27] of LEN, and 100% [13/13] NEN. This was not different between groups. Dogs were anorexic for a median of 5.5 days prior to presentation (range 0–17 days), which was not significantly different between groups.

Fifty-seven percent (32/56) of dogs had feedings tubes placed at some point during hospitalization, and 68.8% (22/32) of those tubes were placed at the time of surgery. All dogs in the EEN group that had feeding tubes placed had them placed at the time of surgery. For those that had a tube placed at the time of surgery, significantly more were in the EEN group (13/16) than in the LEN group (7/27) (P = .0006) and the NEN group (2/13) (P = .0007). Eleven dogs had feeding tubes placed that were not used in hospital (three dogs in EEN, two dogs in LEN, and six dogs in NEN). Of the 21 feeding tubes that were used, 16/21 (76.2%) were nasoenteric (nasoesophageal or nasogastric), including 7/10 in the EEN group and 9/11 in the LEN group. The remaining five tubes were surgically placed, including two jejunostomy tubes (both in the EEN group) and three gastrotomy tubes (one EEN and two LEN). These numbers were not different between groups.

Gastrointestinal signs (vomiting/regurgitation or diarrhea) were present in all groups postoperatively. Vomiting or regurgitation was present in 27/56 (48%) of dogs (8/16 [50%] EEN, 10/27 [37%] LEN, 9/13 [69.2%] NEN), which was not different between groups. Diarrhea was also observed in 27/56 (48%) of dogs (9/16 [56.3%] of EEN, 12/27 [44.4%] of LEN, and 6/13 [46.2%] of NEN), which was similar between groups. When comparing preoperative gastrointestinal signs to postoperative gastrointestinal signs within each group, the LEN group had more diarrhea in the postoperative period compared to preoperatively (13.8% versus 44.8%). There were no differences in any other comparison of preoperative to postoperative gastrointestinal signs.

Overall, 30/56 (53.5%) dogs received prokinetics postoperatively, which was not different between groups (7/16 [43.8%] EEN, 15/27 [55.2%] LEN, 7/13 [53.9%] NEN). Additionally, 9/56 (16.1%) dogs received parenteral nutrition, with three dogs in each group. Parenteral nutrition use was not different between groups. No dogs were started on an antidiarrheal specifically for diarrhea, although many dogs received metronidazole as part of their empiric antimicrobial therapy.

Two dogs (one in the EEN group and one in the NEN group) had radiographs taken postoperatively, with findings consistent with aspiration pneumonia. The dog in the EEN group had radiographic changes consistent with acute respiratory distress syndrome, but aspiration pneumonia could not be ruled out as the inciting cause. No preoperative radiographs were obtained. The second dog with radiographic lung changes was in the NEN group and had a pattern described as consistent with aspiration pneumonia. Both of these cases arrested in hospital postoperatively, but no necropsies or other testing (endotracheal wash) were performed to definitively identify aspiration pneumonia or other pulmonary pathology.

The median time in hospital without eating voluntarily was statistically similar between the LEN and EEN groups. The EEN group did not voluntarily eat for 2.25 days (range 0–8), and the LEN group did not eat for 3 days (range 1.25–9). Overall, 36/56 dogs (64.3%) ate in hospital (13/16 [81.2%] of EEN and 24/27 [88.5%] of LEN). There was no significant difference in suspected dehiscence between groups with 19% (3/16) of EEN, 11% (3/27) of LEN, and 23% (3/13) of NEN having suspected or confirmed dehiscence. Of the nine dogs with suspected dehiscence, eight were confirmed on a second surgery and one was suspected based on intracellular bacteria, but the owner elected euthanasia without a necropsy.

The length of hospitalization for survivors was 5 days (3–15 days). The length of hospitalization was not significantly different between groups (EEN 5.5 days, 4–11; LEN 6 days, 3–13; NEN 4.5 days, 3–6).

Overall, 43/56 (76.7%) dogs survived to discharge. Dogs in the LEN group (23/27, 86.2%) were significantly more likely to survive to discharge compared to the NEN group (6/13, 46.2%; P = .0023). Dogs in the EEN group were not more likely to survive to discharge (12/16, 75%) compared to the other groups. Dogs that ate or were tube fed in hospital (EEN and LEN combined) were more likely to survive to discharge than those that did not (NEN; P = .0062).

Discussion

It has been shown that animals that eat in hospital have a better outcome.²⁴ Additionally, dogs with specific disease processes such as parvovirus, hemorrhagic gastroenteritis, and pancreatitis may benefit from early initiation of enteral nutrition.^11,15,25 However, due to the concern for GICs or a perceived lack of benefit, EEN is not commonly provided immediately postoperatively. In this study, no significant difference in gastrointestinal complications was found with EEN compared to LEN or NEN. Animals were more likely to survive to discharge if they had nutritional intake in hospital (EEN and LEN combined as compared to NEN). In contrast to a previous study, there was no difference in the length of hospitalization between groups. ⁷

In people, enteral nutrition initiation within the first 24 hr is known as the “window of opportunity,” as it is associated with better outcome.¹² Clinically, prospective randomized trials performed in human medicine looking at EEN for critically or acutely ill patients found a lower incidence of infections, mortality, and length of hospitalization.^8–10 It is apparent in this and previous studies that animals that eat in hospital have a better outcome.²⁴ Liu et al evaluated nutrition in postoperative septic peritonitis and found a decreased hospitalization time in dogs that either ate voluntarily or received parenteral nutrition or EN within 24 hours.⁷ The dogs in the present study that did not receive any nutrition, either voluntary or assisted, had a significantly higher mortality rate than those that received nutrition. Although we were not able to demonstrate statistical benefit to supported EEN, we were able to demonstrate a benefit to nutrition. Due to the case bias in this retrospective study, a causal relationship cannot be made between an animal not eating and a worse outcome. The LEN group had a significantly higher survival percentage than those that didn't eat. The EEN group also had a higher survival percentage than those that didn't eat, but it did not reach statistical significance. These groups were the smallest groups in case numbers, so it is possible this did not become significant due to type II error from the small case size.

GICs are common in critically ill patients receiving EN, with a majority of patients showing at least one GIC, but very few of these instances necessitate cessation of feeding.^23,26 Although vomiting and regurgitation were common in LEN and the EEN groups in this study, when compared to the postoperative GI signs seen in the NEN group or compared to the preoperative occurrence of vomiting/regurgitation, there were no significant differences. This comparison indicates that the gastrointestinal signs are likely more attributable to the underlying illness than to the provision of enteral nutrition. Because it is impossible to attribute gastrointestinal signs to the underlying disease, EN, or both, it may be inappropriate to stop EN solely due to the presence of GI signs, which may be present even without EN. Unfortunately, it was not possible to look at discontinuation of feeding based on GI signs in this retrospective study, as there was no set protocol and discontinuation was not consistently noted in the medical record.

The occurrence of diarrhea was similar amongst all groups. Diarrhea is the most commonly reported GIC associated with EN in previous studies and in the current population.^22,23,27 Diarrhea occurs for many reasons, including, but not limited to, high fat content of diets, gastric hypersecretion, lactose intolerance, concomitant drug therapy, altered gastrointestinal flora secondary to antibiotics, and formula hyperosmolarity.²⁷ Dogs in the EEN group may be receiving other medications (opioids, lidocaine) early in the course of postoperative recovery when feeding is initiated that would be discontinued by a delayed feeding initiation time. These medications may alter gastrointestinal motility.²⁷ Although diarrhea was common in all groups, it did not occur more commonly in the EEN group compared to the others. When compared to preoperative signs, the only group that had significantly more diarrhea postoperatively was the NEN group. From these results, it is evident that diarrhea is an expected consequence in postoperative septic peritonitis patients, even in those with no nutritional intake. The diarrhea was self-limiting in all cases.

Early tube feeding in postoperative patients is not considered a risk factor for GICs or pneumonia.¹⁷ Although aspiration was not definitively identified after the start of feeding in these cases, it was not common in any of the study groups. The diagnosis of aspiration pneumonia proved difficult in this retrospective study, with only two dogs diagnosed with suspected aspiration pneumonia. The percentage of suspected aspiration pneumonia was low at 3.6%. Cases were considered suspected aspiration if they had radiographs performed that had findings consistent with aspiration pneumonia. There were likely additional dogs that did not have radiographs taken but may have had pneumonia.

Lack of baseline differences in signalment and clinical signs do not preclude the presence of unidentified group differences because of the non-randomized and retrospective nature of the study. It was not possible to compare groups with a grading system such as the acute patient physiologic and laboratory evaluation score, which classifies disease severity based on physical examination and bloodwork findings due to missing data for most patients.²⁸ It was not possible to retrospectively assess if the groups were truly similar or if the patients in which EN was initiated early were more systemically ill than those in which EN was withheld for a longer period of time. All of the dogs in the EEN group with feeding tubes had feeding tubes placed at the time of surgery, which was significantly more than the other two groups. At the time of surgery, the clinician may have had a suspicion that the animal would benefit from nutrition, potentially due to the presence of a more severe illness or suspicion of prolonged postoperative anorexia. It is impossible to identify the reason for tube placement timing retrospectively, and lack of placement of a tube at the time of surgery may have been due to a patient being more unstable or clinician preference and not based on a clinician's perception that a patient would not eat after surgery.

Limitations of this study include a relatively small number of dogs in each group. Mixing dogs that received tube feeding with those that ate voluntarily may have grouped dogs with different severity of illness. The definition of EEN and LEN hours made these groups similar. The definitions were chosen based on the standard definition of EEN as less than 24 hr. There is no standard definition of LEN, so LEN was defined as greater than 24 hr so as to avoid a large case dropout due to a longer time frame. Another difficulty due to the retrospective nature is that caloric intake was not able to be determined. This data was more available in the later years of the study; the actual intake was not generally noted in charts. For this reason, this study could not look at percentages of the resting energy requirement (RER) consumed. We defined voluntary intake as a dog eating on its own multiple times consecutively, but this may have included dogs that were only eating a small fraction of their RER. Prospective studies could evaluate the percentage RER achieved through tube feeding and through voluntary intake. Furthermore, a variety of clinicians managed these cases, and therefore differences in experience or beliefs may have influenced timing of feeding, surgical outcome, and mortality. Finally, common to retrospective studies, the clinical treatments and outcome may have been based on non-medical factors such as cost limitations as a reason to not initiate enteral nutrition or as a decision to euthanize.

Results of the present study support our hypothesis that EEN is safe in dogs who have had surgery for a septic abdomen and that EEN will not cause an increase in GICs, dehiscence, or aspiration pneumonia when compared to LEN.

Conclusion

The safety of postoperative feeding in addition to the results of improved survival to discharge with nutrition indicates dogs should not be fasted after surgery for septic peritonitis. Further prospective studies are needed to determine the potential benefits of early initiation of enteral nutrition in the postoperative septic peritonitis patient. Further studies should also be performed to look at the role of EEN in other critically ill patients.