One of the most common diseases in intact bitches is pyometra affecting approximately 25% before 10 years of age [1]. Differences in incidence rates between breeds have been described [2–4]. The diagnosis is based on case history, physical examination, and laboratory analyses, often combined with radiography and/or ultrasonography of the uterus and ovaries. Clinical signs vary depending on severity of disease [5, 6]. Leucocytosis, neutrophilia with left shift, anaemia, monocytosis, hypoalbuminemia as well as affected liver or kidney function are common findings [7]. Pyometra has deadly consequences if left untreated and despite modern treatment routines the mortality is 3-4% [1]. The safest and most effective treatment is ovariohysterectomy (OHE) but purely medical treatment can be used in selected cases [8]. Though OHE is a routine procedure, anesthesia and surgery in bitches suffering from severe systemic disease or/and organ malfunctions may be hazardous [9]. The majority of bitches with pyometra suffer from systemic inflammatory response syndrome, which previously has been associated with increased hospitalization and mortality rates [10]. It is important, but clinically difficult, to predict outcome which is why indicators for survival, complications and morbidity are wanted. Prognosis or mortality prediction by investigating different physical status and laboratory parameters is currently in demand in both human and veterinary medicine. Guidelines for performing anesthesia and assessing anesthetic risk based on different criteria are also being developed [11, 12]. Most clinical variables are, however, unspecific which is why current research focuses on identifying clinically valuable biomarkers with high sensitivity and specificity. In diseases with low mortality, such as pyometra, duration of postoperative hospitalization has been used as a measure for morbidity [5, 13–18]. The present study explored clinical and laboratory parameters as indicators of morbidity, measured by duration of postoperative hospitalization and/or peritonitis, using analysis of multivariable associations.

Potentially life-threatening complications of pyometra, described after surgery, include sepsis, septic shock, disseminated bacterial infection, peritonitis and hemorrhage [19–21]. Clinical signs such as vomiting, diarrhea, abdominal distention, decreased appetite or abdominal pain may be observed in animals with septic peritonitis [22, 23]. These signs are, however, commonly encountered in bitches suffering from pyometra with or without peritonitis [24]. For this purpose, researchers have studied biomarkers as indicators of severity of disease and outcome [13, 25]. However, adding analyses that are not routinely performed in clinical work may be time consuming and the cost benefit must be considered. The value of clinical analyses and variables routinely available such as case history data, clinical signs, physical examination findings or laboratory biomarkers has not yet been fully explored as indicators of outcome after surgical treatment of pyometra.

The aims of the present study were to describe complications of pyometra and to investigate variables that may be useful as indicators of peritonitis and/or prolonged hospitalization after surgical treatment.

During the years 2006–2007, 356 bitches of 92 different breeds were diagnosed with pyometra. The age range was one to 15 years (mean ± SD, 9 ± 1.4 years). Seventy-two bitches were > 10 years old. All bitches were diagnosed within 2 months of the previous oestrus. Oestrus prevention by medroxyprogesterone acetate (MPA, the only registered treatment in Sweden) is rare in Sweden, and was only given to one bitch in the study material.

Identifying complications in bitches with pyometra is vital for selecting optimal monitoring routines and treatments and for determination of prognosis. In the present study, complications were observed in 25% of the bitches treated by OHE. Peritonitis was the most common complication and can be life-threatening [26]. Urinary tract infection (UTI) was the second most common complication (6%). Previously, subclinical UTI with the same bacterial strain as in the uterus has been demonstrated in 25% of bitches with pyometra [27]. Though subclinical UTI may resolve without intervention after OHE, proteinuria or clinical signs of disease should be monitored to prevent severe renal disease from developing [28]. In the present study, uveitis was diagnosed in six bitches. Uveitis has not previously been associated with pyometra but has been described in dogs and cats suffering from severe bacterial infection [29, 30]. Cardiac arrhythmia, as identified in five bitches, could have been induced by endotoxemia or myocardial injury [21, 31]. Peritonitis, uveitis and cardiac arrythmias are serious, but treatable, consequences of endotoxemia and sepsis, further supporting the importance for predicting such complications early [32–34].

In the present study, the total mortality including euthanasia was 10% which is higher than earlier reported [1]. Our previous study included bitches < 10 years whereas the present study incorporated data from bitches of all ages, hence the 20% that were > 10 years may contribute to some extent to explain this difference since the owners could be more likely to choose elective euthanasia instead of treatment in an older dog or that has concurrent diseases (which is more likely in older dogs). Four bitches died postoperatively, resulting in a mortality of 1% after surgery, which is lower than the 5-27% in previous reports [19, 31, 35]. Surgical treatment of pyometra had thus a very good prognosis, when performed in the selected cases, in our study. Thirty-two (9%) bitches were euthanized instead of treated due to old age and/or concurrent diseases (such as mammary tumor), all by request of the owner and in agreement with the veterinary surgeon in charge. These bitches were excluded from the analysis of predictors for increased postoperative hospitalization, peritonitis and mortality because they were not surgically treated and their death was not associated with a poor prognosis or severity of pyometra. Age and weight did not differ between the groups (without complications and hospitalization < 3 days compared to with complications and/or prolonged hospitalization), showing that peritonitis or prolonged hospitalization was not more common in older bitches. This indicates that age by itself is not a risk-factor for surgical treatment of pyometra [9].

Overall, the most common signs of disease, present in > 50% of the bitches, included vaginal discharge, anorexia, depression, polydipsia, and polyuria (Table 1), reflecting systemic involvement of the disease in the majority of bitches [19, 36, 37]. Vaginal discharge, which has been associated with more severe disease and is considered a characteristic sign of pyometra [6], was absent in 23% of the cases. Obscure signs of illness may make the disease more difficult to recognize and supports the need for diagnostic indicators for pyometra. Interestingly, lameness was present in 56 (16%) of the bitches, which has not been reported previously. It is possible that the activated immune response could, for example, trigger arthritis, as has been suggested in humans [38]. Regarding laboratory findings, leucocytosis with neutrophilia and left shift, monocytosis and anaemia were observed in the majority of bitches, as is common in the disease [19, 39].

One of 40 bitches with peritonitis died, resulting in a 3% mortality in pyometra with peritonitis which is comparatively low compared with mortality rates of 50% previously reported due to uterine rupture [26]. Importantly, none of the bitches with ruptured uterus died. We may speculate that this could depend on the disease being common in Sweden and possible complications recognized earlier compared with in countries where the majority of the bitches are neutered.

All surgically treated bitches were included in the analyses for identifying indicators for peritonitis and/or prolonged postoperative hospitalization. Two bitches were diagnosed with peritonitis postoperatively and since it could have been subclinical at the time of surgery or caused by uterine leakage during surgery, these bitches were included in the peritonitis group when analyzing for indicators. Moderate to severe depression and pale mucous membranes were associated with increased risk of prolonged postoperative hospitalization (Table 6). A striking result was that leucopenia was associated with a three-point-five-fold increased risk of having prolonged hospitalization compared to normal WBC and also an 18-fold increased risk of peritonitis. These results make leucopenia the most important clinical biomarker identified. Leucopenia could be caused by endotoxin-induced bone marrow depression in combination with more chronic inflammatory disease and loss of leucocytes to the uterine lumen. Increased mortality has also been demonstrated in animals and humans with leucopenia in other studies [40–42]. A WBC within the normal reference range was associated with increased risk for prolonged hospitalization and/or peritonitis as compared to leucocytosis. This could possibly reflect a transition from leucocytosis to leucopenia i.e. leucocytosis appearing earlier in the pathogenesis. Not only may the number of leucocytes be decreased in pyometra, but their function (phagocytic capacity and mitogen-driven lymphocyte proliferation) is also impaired, negatively affecting the combat against infection [43, 44]. Bitches with moderately to severely depressed general condition had a seven-fold increased risk for prolonged hospitalization. Pale mucous membranes, which might reflect anemia, were associated with a three-fold increased risk of prolonged hospitalization. In contrast, hyperemic mucous membranes were associated with decreased risk for prolonged hospitalization. Lactate levels, though only analyzed in 19 dogs, were associated with increased risk for prolonged hospitalization, indicating a predictive value.

Other variables than leucopenia were linked with presence of peritonitis. Fever, or hypothermia present in merely a third of the pyometra cases, was associated with a three-fold increased risk of peritonitis indicating a prognostic value for this variable. Uterine diameter was not associated with peritonitis or prolonged hospitalization, which would otherwise be plausible since a larger uterus could have indicated more severe local disease.

Some parameters were analyzed only in a few bitches hence missing data is a limitation in our study. Retrospectively collected data may also be less reliable than prospectively collected data. Because the bitches selected for medical treatment were already less severely affected by their disease than those subjected to surgical treatment, the data reported here cannot serve as a comparison of the two treatment methods. The results of the present study will increase the possibilities to predict prognosis and outcome after surgical treatment of pyometra. This will be valuable in early identification of cases with peritonitis and/or increased morbidity (hospitalization), which will in turn aid in treatment selection and thereby possibly also increase survival.

Complications such as peritonitis, uveitis, urinary tract infection, wound infection and cardiac arrhythmias were observed in bitches with pyometra. Several routine parameters that may be useful as indicators of peritonitis and/or prolonged hospitalization were identified. Leucopenia was associated with increased risk for peritonitis and prolonged hospitalization making leucopenia the most important biomarker to be aware of clinically. The results of the present study will be clinically valuable for identifying peritonitis in bitches with pyometra and the prediction of prolonged postoperative hospitalization after surgical treatment.