Materials and Methods

Case selection criteria—The medical records database of the University of California-Davis Veterinary Medical Teaching Hospital was retrospectively searched from 2005 to 2010. Hemoperitoneum was defined as an accumulation of blood within the peritoneal cavity^1,2,5; a diagnosis was established on the basis of either grossly visible blood within the peritoneal cavity at the time of surgery or a fluid analysis performed following abdominocentesis. Traumatic hemoperitoneum was defined as an accumulation of blood within the peritoneal cavity that occurred after a known traumatic event, and spontaneous hemoperitoneum was defined as an accumulation of blood within the peritoneal cavity that occurred when no recent (< 1 month) traumatic event was noted. To be included, the etiology of hemoperitoneum was required to be known for patients with both traumatic and spontaneous hemoperitoneum. Additionally, patients with spontaneous hemoperitoneum required a histologic evaluation by a board-certified pathologist of the organ or organs that were bleeding. Lastly, only dogs that underwent surgical intervention were included. Exclusion criteria included hemoperitoneum of unknown etiology or source and cases that did not include surgical intervention as part of the treatment.

Medical records review—Data recorded from the medical records included signalment (age, sex, and breed), body weight, clinical signs, physical examination findings, and clinical laboratory findings (CBC, serum biochemical analysis, coagulation profiles, blood gas analysis, and peritoneal fluid analysis of a sample obtained via abdominocentesis). Other data recorded included radiographic and ultrasonographic findings, treatment including surgical intervention pursued, etiology of hemoperitoneum (traumatic vs spontaneous), histopathologic diagnosis, and perioperative outcome. The perioperative period was defined as the time from admission to the hospital for treatment of hemoperitoneum until the time of discharge from the hospital or euthanasia (within the same hospital visit).

Exact binary logistic regression with a statistical software program^a was used to determine the association between the variables recorded from the medical records and survival to discharge from the hospital. Results are reported as OR and 95% CIs. Values of P < 0.05 were considered significant.

Results

The initial medical record search identified 432 possible cases of hemoperitoneum in dogs. All but 86 dogs were removed because of euthanasia or death prior to any surgical intervention or because the hemorrhagic abdominal effusion was not severe enough to warrant surgical intervention. An additional 3 dogs were not included because of the lack of histopathologic diagnosis from samples obtained during surgery. Eighty-three dogs met the inclusion criteria during the study period.

Of the 83 included case animals, 6 dogs had a traumatic historical event leading to hemoperitoneum and 77 dogs developed hemoperitoneum spontaneously. Male dogs represented 43 of 83 (52%) case animals (36 castrated and 7 sexually intact males), and female dogs represented 40 of 83 (48%) case animals (37 spayed and 3 sexually intact females). Case animals included mixed-breed dogs (n = 24), Labrador Retrievers (11), German Shepherd Dogs (9), Golden Retrievers (5), Boxers (3), English Bulldogs (2), and Belgian Malinois (2). There were 26 other breeds represented, including one each of Airedale Terrier, American Eskimo, Bassett Hound, Belgian Tervuren, Bouvier des Flandres, Bullmastiff, Cairn Terrier, Pembroke Welsh Corgi, Dachshund, Dalmatian, Doberman Pinscher, English Setter, German Shorthair Pointer, German Wirehair Pointer, Great Dane, Siberian Husky, Irish Setter, Irish Terrier, Rhodesian Ridgeback, Saluki, Samoyed, Schipperke, Shetland Sheepdog, Skye Terrier, Springer Spaniel, and Weimaraner. The median age of patients in the study was 10 years (range, 3 to 16 years). The median body weight was 31.3 kg (68.9 lb; range, 5.5 to 66 kg [12.1 to 145.2 lb]). On the basis of a body condition score scale of 1 to 5 (1 = cachectic and 5 = obese), 32 of 71 (45%) dogs were considered overweight or obese (body condition score ≥ 4) and only 6 of 71 (8%) dogs were considered thin or cachectic (body condition score ≤ 2).

The most common clinical signs were lethargy (69/83 [83%]), anorexia (33/83 [40%]), and vomiting (12/83 [14%]); 13 (16%) patients had acute collapse prior to initial evaluation at the clinic. No historical abnormality or clinical sign significantly affected discharge from the hospital. The median number of days from the time the dog was perceived to have been last clinically normal to evaluation for clinical signs was 2 days (range, 0.5 to 60 days).

Tachycardia (heart rate > 140 beats/min) was documented in 40 of 83 (48%) dogs, and a heart murmur was diagnosed in 10 of 83 (12%). The median heart rate of dogs in this study was 146 beats/min (range, 80 to 225 beats/min). Dogs that were tachycardic at the time of initial evaluation at the clinic were determined to be significantly (P = 0.049) less likely to be discharged from the hospital (OR, 0.98; 95% CI, 0.963 to 1). No other physical examination finding was significant for discharge from the hospital (Table 1). The most common respiratory abnormality noted on physical examination was tachypnea, which was diagnosed in 14 of 54 (26%) dogs. Abnormal abdominal palpation findings included abdominal distention (36/83 [43%]) and a palpable abdominal mass (11/83 [13%]). The median body temperature was 38.44°C (101.2°F; range, 36.27° to 40.06°C [97.3°F to 104.1°F]; reference range, 38.06° to 39.17°C [100.5°F to 102.5°F]). Hypothermia and pyrexia were noted in 31% and 28% of dogs, respectively. The median systolic arterial blood pressure was 128 mm Hg (range, 50 to 180 mm Hg). Blood pressure was recorded in only 21 of 83 dogs, with 5% of dogs being hypotensive (systolic arterial blood pressure < 70 mm Hg) and 14% of dogs being hypertensive (systolic arterial blood pressure > 150 mm Hg).

Results of a CBC and serum biochemical analysis were not available for all dogs (Table 2). The median WBC count was 17,500 WBCs/μL (range, 4,000 to 39,900 WBCs/μL; reference range, 6,000 to 13,000 WBCs/μL); 35 of 47 (74%) dogs had leukocytosis, and 4 of 47 (8.5%) dogs were leukopenic. Forty of 56 (71%) dogs were thrombocytopenic at the time of admission, and the median platelet count was 113,500 platelets/μL (range, 8,000 to 774,000 platelets/μL; reference range, 150,000 to 400,000 platelets/μL). The median peripheral PCV and abdominal fluid PCV were 29% (range, 9% to 71%; reference range, 40% to 55%) and 30% (range, 8% to 61%), respectively. The median TS concentration of the peripheral blood was 6 g/dL (range, 2.6 to 9.3 g/dL; reference range, 5.4 to 6.9 g/dL), and the median TS concentration of the abdominal fluid was 5 g/dL (range, 3.0 to 6.8 g/dL). The median peripheral albumin concentration was 2.8 g/dL (range, 0.8 to 4.2 g/dL; reference range, 3.4 to 4.3 g/dL), with 39 of 50 (78%) dogs being hypoalbuminemic. Sixty-nine of 80 (86%) dogs were anemic at initial evaluation, and 30 of 72 (42%) dogs were hypoproteinemic.

The median prothrombin time was 9.8 seconds (range, 7.2 to 21.5 seconds; reference range, 7 to 9.3 seconds), with 18 of 31 (58%) having prolongation. The median activated partial thromboplastin time was 15.4 seconds (range, 9.7 to 126 seconds; reference range, 10.4 to 12.9 seconds), with 22 of 31 (71%) having prolongation. Hyperlactatemia was present in 54 of 66 (82%) dogs, and a median lactate concentration of 4 mmol/L (range, 0.7 to 12.9 mmol/L; reference range, 0 to 2 mmol/L) was noted. Twenty-nine of 53 (55%) dogs with hemangiosarcoma had elevated lactate concentration, and 25 of 30 (83%) dogs with an etiology other than hemangiosarcoma had elevated lactate concentration.

Thoracic radiographs were obtained in 73 of 83 (88%) dogs prior to surgical intervention. No abnormalities were noted in 37 of 73 (51%) dogs. Pleural effusion was diagnosed in 8 of 73 (11%) dogs; therefore these patients were considered to have bicavitary effusion. Seven of 73 (10%) dogs were determined to be hypovolemic on the basis of the presence of microcardia or a small vena cava. A pulmonary nodule or nodules were noted in 6 of 73 (8%) dogs. None of the pulmonary nodules were biopsied or aspirated. Aspiration pneumonia or signs of pulmonary interstitial infiltrates were noted in 4 of 73 (5%) dogs. Sternal lymphadenopathy was present in 8 of 73 (11%) dogs, and 3 (4%) dogs had evidence of mild cardiomegaly. The presence of pleural effusion on preoperative radiographs, (ie, bicavitary effusion), was a negative prognostic indicator for discharge from the hospital (OR, 0.13; 95% CI, 0.022 to 0.74; P = 0.018). All patients underwent abdominal ultrasonography performed either via a focused abdominal sonography⁸ scan by an emergency department veterinarian or a full abdominal ultrasonographic scan performed by a radiologist. In all cases, abdominal fluid was noted ultrasonographically and was confirmed to be blood after obtaining a sample via abdominocentesis.

Every dog in this study received a blood product transfusion. Of dogs receiving packed RBCs (78/83 [94%]), the median volume administered during the perioperative time frame was 15.4 mL/kg (7 mL/lb; range, 4.8 to 290 mL/kg [2.2 to 131.8 mL/lb]). Dogs that received fresh-frozen plasma (59/83 [71%]) received a median volume of 11.1 mL/kg (5.05 mL/lb; range, 3.6 to 272 mL/kg [1.64 to 123.6 mL/lb]) in the perioperative time frame. Of dogs receiving fresh whole blood (12/83 [14%]), the median volume administered was 9.8 mL/kg (4.45 mL/lb; range, 5.5 to 12 mL/kg [2.5 to 5.45 mL/lb]) in the perioperative time frame. Four dogs met the criteria for having received a massive transfusion perioperatively as previously defined⁹ (a transfusion of a volume of whole blood or blood components that is greater than the patient's estimated blood volume within a 24-hour period, replacement of half the patient's estimated blood volume in 3 hours, or the replacement of 150% of the patient's blood volume irrespective of time). The requirement of a massive transfusion for hemodynamic stabilization was a negative prognostic indicator for discharge from the hospital (OR, 0.063; 95% CI, 0.0011 to 0.86; P = 0.035).

Surgical interventions that were performed included splenectomy (n = 67), liver lobectomy (11), adrenalectomy (3), intestinal resection and anastomosis (3), nephrectomy (2), uterine stump ligation (1), gastric derotation with gastropexy (1), partial prostatectomy (1), and liver, omental, or lymph node biopsies (31). The causes of traumatic hemoperitoneum (6/83 [7%]) were determined to be gastric dilatation and volvulus, vehicular trauma, previous surgery for an intestinal foreign body and ovariohysterectomy, penetrating trauma from an elk attack, and fine-needle aspiration of a hepatic tumor; the hepatic tumor was considered in the traumatic group because the mass was not bleeding prior to aspiration. Ten dogs had a spontaneous hemoperitoneum due to a nonneoplastic cause; 9 of those cases were secondary to a ruptured splenic hematoma, and 1 occurred because of rupture of a hepatic abscess. Of the dogs with spontaneous hemoperitoneum, neoplasia was the cause of hemoperitoneum in 67 of 77 (87%), with hemangiosarcoma being the histologic diagnosis in 51 of 67 (76.1%) dogs. The locations of neoplastic lesions (primary and metastatic) included the spleen (n = 65), liver (39), omentum (10), mesentery (7), adrenal glands (5), kidneys (4), abdominal lymph nodes (4), intestines (3), and diaphragm (2). Compared with all dogs and all sources of hemorrhage, dogs with hemoperitoneum secondary to splenic rupture were significantly (P < 0.001) more likely to be discharged from the hospital (OR, 9.44; 95% CI, 2.38 to 41.35). No other source of hemorrhage was significantly associated with discharge from the hospital.

The most commonly encountered postoperative complication was cardiac arrhythmias. Ventricular premature contractions or paroxysmal ventricular tachycardia was documented in 16 of 83 (19%) dogs. An accelerated idioventricular rhythm was documented in 11 of 83 (13%) dogs. Overall, 27 of 83 (32%) dogs developed a ventricular arrhythmia after surgery. Although the difference between these types of arrhythmias was loosely defined in the medical records, there was no significant correlation between the presence of arrhythmias after surgery and discharge from the hospital. Four (4.8%) dogs were noted to have a coagulopathy, and incisional serosanguineous discharge was documented in 4 (4.8%). Neither the presence of a coagulopathy nor the presence of an incisional complication was significantly correlated with discharge from the hospital. Suspected aspiration pneumonia was documented on thoracic radiographs in 4 (4.8%) dogs after surgery. Presence of respiratory disease after surgery secondary to suspected pulmonary thromboembolic disease or ARDS was shown to negatively affect discharge from the hospital (OR, 0.051; 95% CI, < 0.01 to 0.49; P = 0.01). Acute respiratory distress syndrome was defined by unresponsive severe hypoxemia, dyspnea, and pulmonary infiltrates that were noted on thoracic radiographs.^10,11

Thirteen (15.7%) dogs died or were euthanized during the perioperative period. Cardiopulmonary arrest occurred in 3 of 13 dogs (1 under anesthesia and 2 after surgery), all of which underwent unsuccessful resuscitation efforts. Intraoperative euthanasia was performed at the request of the owner in 6 of 13 dogs because of suspected metastatic disease or uncontrollable hemorrhage from hepatic or adrenal masses. The remaining 4 of 13 dogs were euthanized because of complications arising after surgery, including suspected pulmonary thromboembolic disease, aspiration pneumonia, and disseminated intravascular coagulopathy; this included 1 dog with a traumatic hemoperitoneum due to vehicular trauma. The median hospitalization time for the remaining dogs that survived to discharge was 2 days (range, 1 to 5 days).

Discussion

In the present study of perioperative hemoperitoneum in dogs, factors associated with death and failure to be discharged from the hospital included tachycardia, bicavitary effusion (ie, abdominal and pleural effusion), severe respiratory disease, and the requirement of massive blood product transfusions. The presence of disease within the spleen was positively associated with survival to discharge. In our series of 83 patients treated over a 5-year period, surgical intervention resulted in discharge from the hospital in most patients (84.3%), regardless of disease etiology.

When hemoperitoneum develops in a canine patient, rapid medical attention and quick decision making from owners are generally necessary. Acute traumatic cases of hemoperitoneum requiring surgical intervention are less often encountered,^5,7,12 compared with spontaneous cases; however, both forms of hemoperitoneum generally require aggressive diagnostic evaluation and treatment.

The presence of hemoperitoneum without a traumatic cause has historically been considered a poor prognostic indicator in canine patients because of the likelihood of a neoplastic etiology.² In previous reports^3,5 describing hemoperitoneum in dogs, bleeding from the spleen was the most common cause of both cases of traumatic and spontaneous hemoperitoneum. Traumatic events generally cause a fracture of the spleen or rupture of the blood supply to the spleen. Splenic neoplasia is often the cause of spontaneous hemoperitoneum, and hemangiosarcoma is the most common splenic tumor.¹ Recent reports^13,14 evaluating spontaneous hemoperitoneum document the prevalence of hemangiosarcoma to be as high as 70%. We found a similar prevalence of hemangiosarcoma in the present study; in dogs with neoplasia as the source of spontaneous hemoperitoneum, hemangiosarcoma represented 76% of all malignancies. The documented perioperative mortality rate of dogs with hemangiosarcoma is 39% to 59%.^1,2 The perioperative mortality rate for confirmed hemangiosarcoma in this study was much lower at 9.8%. However, this value may not reflect the true occurrence of hemangiosarcoma given that only those dogs that underwent surgical intervention were included in the present study.

Disease within the spleen causing hemoperitoneum was found to be a positive predictive indicator for discharge from the hospital among the dogs of this study, compared with all other etiologies. Clients are always confronted with a difficult decision when a splenic mass is noted preoperatively; thus, the importance of this finding lies in the impact that it may have on that decision-making process. Given that the current commonly used preoperative diagnostic tests (eg, abdominal ultrasonography, cytologic evaluation, and clinicopathologic tests) are inadequate to confidently confirm benign or malignant splenic disease, a decision of surgery versus euthanasia generally needs to be made. Perhaps this decision can be somewhat simplified in that dogs with the most common cause of hemoperitoneum (splenic bleeding) have a higher likelihood of surviving to discharge. Furthermore, we speculate that the treatment of splenic disease may be recommended more commonly, as splenectomy is generally considered a less complicated surgery than surgery involving other abdominal organs, such as the liver, adrenal glands, and kidneys.

Clinical signs and physical examination findings of dogs with hemoperitoneum can range from signs of severe hemorrhage such as collapse or weakness, obtundation, and poor perfusion to vague signs in dogs with normal perfusion parameters.³ The only physical examination or clinical sign factor determined to be significant in this study in predicting that a patient was less likely to be discharged from the hospital was tachycardia. Tachycardia can occur clinically for multiple reasons, of which the most probable causes in this group of dogs were signs of pain and hemorrhage-induced hypovolemia. An elevated pulse rate may signify a more severely compromised cardiovascular state in dogs because of either systemic effects from the underlying etiology or more severe blood loss and cardiovascular collapse.

To reestablish and maintain circulating volume, support oxygen-carrying capacity, or treat coagulation abnormalities, all of the dogs in this study received a transfusion of packed RBCs, fresh-frozen plasma, fresh whole blood, or a combination of these products. One study⁹ evaluated massive blood transfusions in a group of dogs, finding a high mortality rate among these patients (73%). There were 4 dogs in the present study meeting the criteria for a massive transfusion, 3 of which died or were euthanized. Patients requiring large quantities of blood products for stabilization often have severe illnesses but also can have complications related to the transfusions, such as electrolyte disturbances, acute lung injury, transfusion reactions, and alterations in acid-base status.⁹ We cannot conclude from the available data whether the higher mortality rate was the result of the transfusions or secondary to a presumed greater severity of hemodynamic compromise in these dogs. The data support that patients requiring massive transfusions have a poorer prognosis, and specifically dogs receiving massive transfusions were 16 times as likely to not survive the perioperative timeframe.

Thoracic radiographs were obtained in 88% of dogs in this study, and although 6 dogs were documented to have at least 1 soft tissue nodule, the clients elected to proceed with surgical intervention; all of those dogs were discharged from the hospital. This is possibly an underestimation of the true prevalence of pulmonary metastatic disease in these dogs, considering that a previous study¹⁴ has documented that there is a failure to diagnose pulmonary metastatic hemangiosarcoma on radiographs prior to necropsy. However, the purpose of this study was not to follow up on these dogs until death, so the long-term occurrence of pulmonary neoplasia is unknown in this population. Although only 11% of dogs were noted to have pleural effusion, and therefore bicavitary effusion, these dogs were 7.7 times less likely to be discharged from the hospital. The fluid was sampled in only 1 dog and was determined to be hemorrhagic effusion. In all other dogs, the fluid was scant and determined to be difficult to sample; therefore, a diagnosis was not obtained. Bicavitary effusion has been documented in many disease processes either from systemic effects of infectious or cardiovascular disease or metastatic spread of neoplasia.¹⁵ It has been suspected to be associated with advanced disease states and a higher patient morbidity rate in cats and dogs.^16,17 Bicavitary effusion, historically, has been most commonly observed in dogs with neoplastic or cardiovascular disease and carries a poor to grave prognosis.¹⁸

Multiple hematologic and serum biochemical variables were evaluated in the dogs of this study. Although not every variable was documented for each dog, none of the compared variables were significant for predicting survival to discharge from the hospital. One study determined that dogs with splenic hemangiosarcoma and hemoperitoneum were significantly more likely to have decreased TS concentration and thrombocytopenia.¹³ Although the long-term prognosis of hemangiosarcoma is poor, these values as well as the PCV, all of which can be considered indicators of blood loss, were evaluated in this group and had no effect on discharge from the hospital. Blood lactate concentration, which is an indicator of systemic perfusion, was elevated in most dogs in which it was measured. This is likely related to systemic hypoperfusion and hypoxemia. One study¹³ found that dogs with hemangiosarcoma seemed to have more elevated lactate concentration. Our data do not support this finding given that elevated lactate concentration was documented in a higher percentage of dogs that did not have hemangiosarcoma. Hyperlactatemia in the dogs of this study did not significantly affect survival to discharge, regardless of final diagnosis.

Of dogs described in a previous report¹² of traumatic causes of hemoperitoneum, only 5.6% required surgical intervention. On the basis of that previous report¹² and the present study, traumatic hemoperitoneum appears to rarely require surgical intervention. All of the dogs in this study population underwent surgery, and only 6 of 83 (7.2%) dogs had traumatic hemoperitoneum.

The most common surgical procedure reported in the present study was a splenectomy, which was expected given that 96% of the spontaneous hemoperitoneum cases were secondary to splenic pathological changes. Other procedures performed as a result of neoplasia included liver lobectomy, nephrectomy, and adrenalectomy; all of these procedures have the potential to be more involved and may be regarded as being technically more difficult than a splenectomy. This demonstrates the importance of attempting to identify the source of hemorrhage prior to surgery to better define the necessary skill and experience level of the surgeon, to determine the equipment needed to perform the procedures efficiently, and to decide whether more extensive postoperative monitoring will be warranted.

One common complication associated with splenic disease is ventricular arrhythmias.^2,3 For this reason, it is often recommended that dogs have continuous ECG monitoring perioperatively.² Prevalence of ventricular arrhythmias in cases of splenic pathological lesions has been documented to be present in 39% to 86% of cases of splenic hemangiosarcoma, 31% to 87% of cases of splenic hematoma, and 33% to 100% of cases of splenic torsion.² Regardless of whether treatment was required to alleviate clinically relevant ventricular arrhythmias, all of the dogs treated in 1 study² survived to discharge. The prevalence of ventricular arrhythmias in the present study was 37% (24/65 dogs), and this diagnosis did not affect survival to discharge. The occurrence of ventricular arrhythmias can be variable in postoperative patients, and the true incidence may be underestimated in patients with subclinical arrhythmias in which an ECG was not continuously performed. Our data support previous studies^2,3 describing minimal clinical relevance of ventricular arrhythmias.

The presence of severe respiratory disease after surgery was significant for predicting failure to survive to discharge. The dogs in this study were all potentially exposed to known risk factors associated with pulmonary thromboembolic disease, including surgery, trauma, disseminated intravascular coagulation, and neoplasia.¹⁰ This disease process can be rapidly fatal and is difficult to diagnose, although 59% to 64% of dogs with pulmonary thromboembolism had > 1 predisposing condition in 2 studies.^10,19,20 Acute respiratory distress syndrome is a disease process that has been documented readily in human patients.¹¹ In dogs and cats, it occurs subsequent to pulmonary failure, often as the result of multiple organ dysfunction syndrome, and may carry a grave prognosis.²¹ Our data support that patients that develop suspected pulmonary thromboembolism or ARDS have a poorer prognosis, given that these dogs were 19.6 times less likely to be discharged from the hospital. However, it is a weakness of this study that necropsy was not performed in these cases to confirm the presence of these particular diseases; the suspected diagnoses were based on clinical findings. Thoracic radiographs were performed in these patients after clinical signs developed, and the findings were consistent with pulmonary interstitial infiltrates, suggesting the development of ARDS.

The median hospitalization time for patients with hemoperitoneum was short (2 days; range 1 to 5 days). It would appear that although these patients often require intensive care and monitoring immediately upon evaluation, the recovery time is relatively short.

There are several limitations of the present study. Although relatively large, compared with other hemoperitoneum studies,^1,2,14 the case numbers in this study are small; furthermore, only a limited number of dogs failed to survive to discharge. Also, given that the performance of euthanasia is based on the decision of an owner, this does not reflect a true death rate secondary to disease. Given the retrospective nature of this study, the population of dogs that did not undergo surgical intervention was not able to be evaluated. The reasons for the lack of surgical intervention may have been based on financial decisions, clinician counseling, or a perceived severe disease state. These factors make it difficult to evaluate the perioperative mortality rate in the group of dogs with hemangiosarcoma, even though it was lower than previously documented. An evaluation of these patients in the future may elucidate further risk factors for failure to survive the perioperative period. Retrospective studies have inherent weaknesses, including incomplete or poorly detailed medical records regarding surgical findings and postoperative care and recovery, inconsistent data obtained with regard to preoperative blood and diagnostic testing, and differences in resuscitation efforts and treatment for each dog. Although some important short-term prognostic factors were identified, a larger series of cases may identify more associations between the variables evaluated and short-term prognosis.

The present study identified a population of dogs with hemoperitoneum. Although most dogs had hemangiosarcoma and were likely given a poor long-term prognosis, 84% of the dogs in this study survived to discharge, with a median hospitalization time of 2 days. Owners of these dogs are expected to make immediate decisions regarding treatment and stabilization of their animals; additionally, consideration of the long-term treatment options and prognosis on the basis of the likely etiology is also necessary. Although multiple studies have named factors that may be suggestive of the diagnosis of hemangiosarcoma, no study to date has been able to elucidate any factor that is diagnostic for this type of neoplasia without histopathologic findings. It is the role of the treating clinician to counsel owners on the potential for malignant neoplasia, and the results of this study suggest that most dogs will survive the perioperative time frame to be discharged, despite the etiology of hemoperitoneum. Risk factors such as massive transfusions, the presence of pleural effusion documented on preoperative thoracic radiographs, and postoperative respiratory disease may help to guide owners regarding short-term prognosis for the possibility of discharge from the hospital. With proper education, and appropriate patient care, it appears reasonable to offer surgery as a means to obtain cardiovascular stabilization and to achieve a definitive diagnosis.