Introduction

Sepsis in veterinary medicine is currently defined as infection with a systemic inflammatory response.¹ Mortality rates of dogs and cats with sepsis have been reported to be between 20 and 68%, with increased mortality associated with progressive organ dysfunction, hypotension, tachypnea, and altered mentation.^2–5 The 2021 International Guidelines for Management of Sepsis and Septic Shock in humans emphasize the importance of early identification and management of septic patients.⁶ These guidelines further outline the pinnacles of treatment for septic patients: fluid resuscitation with crystalloids, the early use of vasopressors, rapid source control, and early antimicrobial administration, ideally within 1 hr of sepsis recognition when septic shock is present and potentially later in patients without shock.⁶

The importance of timely antimicrobial administration in cases of sepsis has repeatedly been demonstrated in human medicine, with mortality decreasing from 38.5 to 19% when appropriate antimicrobials were administered within 1 hr of triage.⁷ A retrospective study found evidence that time to antimicrobial administration significantly affects mortality rates in human patients with sepsis, with an increased odds of mortality by 0.3% with each hourly delay of antimicrobial administration; even higher increases in mortality were found for patients with septic shock.⁸ Numerous other studies have reported that the time to antimicrobial administration impacts outcome in humans with septic shock.^7,9,10 Another nationwide cohort study also found strong evidence that a subset of patients with septic shock demonstrated a 35% increased risk of mortality for every 1 hr delay in antimicrobial administration.¹¹ Although multiple retrospective studies advocate that timely antimicrobial administration decreases mortality rates in human patients with sepsis and shock, other individuals have questioned the strength of validity of these findings and suggest that more robust and better controlled experimental studies are needed to more accurately evaluate the effect of time to antimicrobial administration on mortality rates in cases of sepsis.^12–14

Few studies have been completed regarding the timing of antimicrobial therapy in septic dogs; thus, current treatment recommendations in veterinary patients are largely based on human medical guidelines.¹⁵ One retrospective study found that appropriate empirical antimicrobial therapy did not impact the survival rates in a population of 84 dogs with septic peritonitis.¹⁶ An additional retrospective study identified time to antimicrobial therapy as a potential predictor of mortality in dogs with septic shock.¹⁷ A retrospective study completed in a population of 82 cats with septic peritonitis found that appropriate antimicrobial therapy at admission was associated with a 4.4-fold increased likelihood in survival when compared with cats that did not receive appropriate antimicrobials, although this association was not investigated in the context of culture results.¹⁸

The objective of this study was to determine if the time to any antimicrobial administration, as well as time to appropriate antimicrobial administration, impacted hospitalization time or mortality of dogs diagnosed with sepsis. Source control was also evaluated as a variable. We hypothesized that septic dogs with a shorter time to any antimicrobial administration, as well as shorter time to appropriate antimicrobial administration, and received source control, would have a lower mortality and shorter hospitalization time.

Materials and Methods

Medical records from a tertiary referral teaching institution were searched from 2012 to 2022 for dogs diagnosed with sepsis. The electronic medical record was searched for “sepsis-bacterial,” “septic shock,” “septicemia,” “pneumonia-aspiration,” “septic peritonitis,” “septic abdomen,” “septic arthritis,” “pneumonia,” “pyothorax,” “pyometra,” “bacterial pneumonia,” “pyelonephritis,” “prostatitis,” “necrosis-skin,” “metritis,” “meningitis-bacterial,” “mastitis,” “mastitis-abscess,” “endocarditis,” “cholangiohepatitis-bacterial,” “arthritis-septic,” “abscess-prostate,” “abscess-liver,” and “abscess-pancreas.”

Data recorded from the medical records included age, breed, sex, weight, reproductive status, final diagnosis (i.e., septic peritonitis, pyometra, etc.), and vitals upon admission (heart rate, respiratory rate, temperature, mentation, and systolic blood pressure). Hematological and biochemical parameters measured at admission were also recorded, including white blood cell count, platelet count, hematocrit, lactate, glucose, and albumin. Dogs were included in the study as septic if they had a clinical diagnosis of a bacterial infection and met two or more of the systemic inflammatory response syndrome (SIRS) criteria from Table 1. An acute patient physiological and laboratory evaluation fast (APPLE_fast) score was calculated based on biochemical parameters and mentation at the time of admission.¹⁹ Dogs were excluded if an APPLE_fast score could not be calculated, an antimicrobial was administered before presentation, the time of antimicrobial administration could not be determined, or if an antimicrobial was not administered while the patient was hospitalized.

TABLE 1 Criteria for Diagnosis of SIRS in Dogs

View Fullscreen

If a culture was performed, the culture source and the result of the culture were recorded, as were the identification of the organism that was grown and the corresponding antimicrobial susceptibility, if applicable. None of these patients had repeat cultures performed. Patient outcome and hospitalization length were also collected. Additional data recorded included the date and time of admission, the date and time of initial antimicrobial administration, and the date and time of appropriate antimicrobial administration.

Time to antimicrobial administration was defined as the time to the first antimicrobial administered during the patient’s hospitalization. Time to appropriate antimicrobial administration was defined as time to administration of an antimicrobial to which all cultured bacteria were susceptible. For time to appropriate antimicrobial administration, only patients that had a positive culture result were included in the analysis. The time between initial antimicrobial administration and admission, as well as appropriate antimicrobial administration and admission, were calculated and recorded.

Results

Records from 323 canine patients were reviewed. A total of 148 dogs were excluded from the study because of their records missing key data points, such as criteria to calculate an APPLE_fast score and documentation regarding any antimicrobial administration while hospitalized. Subsequently, 175 dogs were included in the study (Figure 1). The median age of dogs was 8 yr (range: 2 mo to 17.5 yr), and the mean weight was 20 kg (range: 1.3 kg to 69.7 kg). There were 52 neutered males (30%), 20 intact males (11%), 54 spayed females (31%), and 48 intact females (27%). The sources of sepsis for dogs can be found in Table 2 and the APPLE_fast and SIRS criteria for dogs can be found in Table 3. Of the dogs included in this study, 44% had accessible culture results and 36% had susceptibility testing performed. For the culture results without susceptibility testing, the antibiotic was determined to be an appropriate choice if the organism cultured was found to historically be susceptible to the prescribed antibiotic from antimicrobial susceptibility standards. Of the 78 dogs with culture results, 19 dogs had no growth on the submitted cultures. The most common infections were caused by gram-negative bacteria (n = 35), followed by gram-positive bacteria (n = 15) and mixed infections (n = 9). Four dogs that had bacterial infections also had culture results that tested positive for other, nonbacterial causal agents (i.e., viral, protozoal, fungal, etc.) and were included in analysis. Only one of the dogs with positive culture results had an anaerobic infection (Clostridium difficile). For all dogs, the initial antimicrobial prescriptions can be seen in Table 4. Of those treated with inappropriate antimicrobials (n = 29), 21 dogs were prescribed ampicillin/sulbactam, 3 dogs were prescribed ampicillin/sulbactam with enrofloxacin, 1 dog was prescribed enrofloxacin, 1 dog was prescribed ampicillin/sulbactam with amikacin, and 3 dogs were prescribed no antibiotics. The overall survival to discharge was 80%. Of the dogs that died, 28 were euthanized and 7 arrested while hospitalized. The median hospital stay was 2 days (range: 0 days to 25 days).

View Figure

• Download as Powerpoint
• Download Figure

TABLE 2 Quantification of Disease States of Dogs Included in the Study

View Fullscreen

TABLE 3 Quantification of APPLE_fast and Patient Illness Severity Scores

View Fullscreen

TABLE 4 Initial Antimicrobial Prescriptions in Septic Dogs

View Fullscreen

No association between time to antimicrobial administration and length of hospitalization was identified (P = .11; Figure 2); this finding remained insignificant when evaluating only dogs that survived (P = .44).

View Figure

• Download as Powerpoint
• Download Figure

There was a significant relationship between time to appropriate antimicrobial administration and length of hospitalization (Figure 3); dogs that had longer times to appropriate antimicrobial administration from admission had significantly longer hospitalization stays (P = .004). This correlation remained significant when including data only from the dogs that survived to discharge (P < .05).

View Figure

• Download as Powerpoint
• Download Figure

Although a negative trend was identified, there was no significant difference in time to antimicrobial administration between dogs that survived and dogs that did not survive (P = .11). The mean time to antimicrobial administration in dogs that survived was ∼6 hr (standard deviation 0.25 hr), whereas the mean time to antimicrobial administration in dogs that did not survive was ∼8 hr (standard deviation 0.78 hr). There was no significant difference between time to appropriate antimicrobial administration and the outcome of dogs included in this study (P = .37).

A statistically significant relationship was found between survival and APPLE_fast score (χ² = 4.96, df = 1, P = .02) but not between APPLE_fast score and time to any or appropriate antimicrobial administration (P = .07). No significant difference was found between hospitalization time and APPLE_fast score (P = .073), even when evaluated independently in survivors (P = .18).

A statistically significant relationship was also found between source control and survival rate (P = .05), with patients significantly more likely to survive if source control was performed (survival rate 90%), versus patients that did not have source control performed (survival rate 74%).

Although the relationship between time to any antimicrobial administration and source control was not statistically significant (P = .77), the relationship between time to appropriate antimicrobial administration and source control was significant, with patients that had source control performed receiving significantly faster appropriate antimicrobial administration when compared with patients that did not receive source control (P < .01).

Discussion

This study found a positive association between the time to appropriate antimicrobial administration and hospitalization time, as well as between time to appropriate antimicrobial administration and source control. No association was identified between the time to any antimicrobial administration and hospitalization time or source control of dogs diagnosed with sepsis. Although a negative trend existed between time to any antimicrobial administration and outcome (survival or nonsurvival), the time to any or appropriate antimicrobial administration did not reach significance for dogs diagnosed with sepsis. This lack of significance could reflect the need for a larger sample population or be affected by illness severity. It is also possible that the diagnostic labels used to search records for septic patients may have included patients without sampling that were not truly septic, which would have affected results. These limitations should be further investigated in future studies.

Human studies have repeatedly shown that the time to appropriate antimicrobial administration significantly affects the length of hospitalization in septic patients. When evaluating 1,058 people with sepsis, researchers found a 0.095-day increase in hospitalization time per hour delay of appropriate antimicrobial delivery.²⁰ Van den Bosch et al. found similar results using a population of 1,890 adults with suspected bacterial infections.²¹ These findings agree with the association found in the present study between time of appropriate antimicrobial administration and hospitalization length; dogs that took longer to receive the appropriate antibiotic had a longer hospitalization. However, it is important to note that confounding variables may be present within this association. For example, a clinical diagnosis of bacterial infection without cytology and/or culture can be challenging, as inflammatory disease processes may be sterile and viral or fungal infections were not excluded. By including dogs in the study that lacked cytology and cultures, it is possible that not all included dogs truly had bacterial sepsis. Additionally, dogs with more severe forms of infection may not respond to empirical antimicrobial choices, thus requiring the results of culture and sensitivity before appropriate antimicrobial therapy can be administered. Although disease severity may also impact hospitalization time and antimicrobial administration, this study sought to reduce the impact of disease severity hospitalization time by controlling for APPLE_fast scores when evaluating time to antimicrobial administration. Predicting appropriate antimicrobials from the outset of disease, by using antibiograms or consultation with infectious disease specialists, may help reduce hospitalization time, and therefore client costs, in septic dogs. Only 36% of dogs included in this study had available antimicrobial susceptibility results, highlighting the need to obtain early cultures more consistently in septic patients. This low percentage of susceptibility results as well as the population of patients that received susceptibility testing may have further affected the interpretation of some of the data associated with susceptibility testing. Patients that present for surgical source control are more likely to have cultures performed than patients with pneumonia, which raises another source of bias within the study. Approximately 43% of the patient population in this study had pneumonia, which may have affected results in several ways. Additionally, many of these dogs were likely tachypneic from their primary disease process, increasing the likelihood of fulfilling sufficient SIRS criteria even if they were not experiencing a dysregulated response to infection. Future studies may consider excluding pneumonia from the study population to further evaluate pneumonia’s role as a source of bias in this context.

In contrast, time to any antimicrobial therapy was not associated with hospitalization time in septic canine patients. This demonstrates the importance of identifying not just any antimicrobial at the time of hospitalization, but the most appropriate antimicrobial for likely infectious organisms based on the septic source and identifying risk factors for multidrug-resistant organisms.

Source control was found to positively impact survival and time to appropriate antimicrobial administration, although it was not significantly related to time to any antimicrobial administration. It is likely that source control helped facilitate culture collection that identified the bacterial agent of greatest concern in septic patients, which helped shorten the time to appropriate antimicrobial administration and significantly improve survival rates within this population. These findings further suggest the importance of source control as well as time to appropriate antimicrobial administration when treating veterinary patients with sepsis.

Current evidence regarding the importance of timing of antimicrobial administration on the outcome of veterinary patients with sepsis is lacking. Although a retrospective study in cats with septic peritonitis found that appropriate empirical antimicrobial administration at presentation was significantly associated with increased survival rates, this study did not investigate a relationship between survival and antimicrobial change after culture results were obtained.¹⁸ Whereas one study in dogs reported that time to antimicrobial administration predicted mortality in dogs with septic shock, another study found no association between appropriate antimicrobial choice and survival to discharge in dogs with septic peritonitis.^16,17 The current study also found no association between time to any or appropriate antimicrobial administration and outcome in septic dogs. The differences in these findings underline the importance of distinguishing the severity of sepsis in canine patients, with the implication that the timing of antimicrobial administration may be increasingly important in cases of more severe subsets of sepsis, such as septic shock. Another factor contributing to the difference in findings may also include the sources of sepsis in these studies. It is possible that patients with septic peritonitis may respond well to source control and lavage, as opposed to patients that cannot receive definitive control, which could confound significant associations identified between antimicrobial choice and survival. This prospect is reflected by Summers et al. finding that survivors were more likely to have gastrointestinal sepsis, versus other sources of sepsis.¹⁷

These trends are further emphasized in human studies, where the timing of antimicrobial administration has more often been found to impact the outcomes of patients with septic shock when compared with patients with sepsis.^7,9,11 Given that mean hospitalization time in our study was only 2 days and survival was 80%, our patient population may represent a less severe septic group than previous studies, although both of these factors likely skew the data away from the null. Mean hospitalization may have also been confounded by the clients’ ability to afford increasingly costly additional days, which could have resulted in patients returning home sooner than was recommended. It is additionally important to note that septic peritonitis is proportionately more common in veterinary medicine, which could lead to differences in findings when compared with human medicine. The studies in people are also larger, which may have affected the chance for error. Whereas Summers et al. focused their study on a population of dogs with septic shock, and Dickinson et al. used a sample of dogs with septic peritonitis, the present study included any dogs that had a diagnosis of sepsis, including dogs with and without shock.^16,17 Although these inclusion criteria allowed for a greater sample size, which helped reduce the chance of type 2 error, this may also explain the lack of a relationship for these parameters. New sepsis definitions in veterinary medicine may help prevent the inclusion of patients with infections who do not have significant organ dysfunction.

Although not significant, the presence of a negative trend between the time to any antimicrobial administration and outcome suggests that significance may be found in future studies excluding healthier animals. The median time to any antimicrobial administration was 5 hr in survivors (range 0 hr to 53 hr) and 3.5 hr (range 1 hr to 4 hr) in nonsurvivors in this study. The median time to appropriate antimicrobial administration was approximately 7 hr in survivors (range 0.25 hr to 230 hr) and 6 hr in nonsurvivors (range 2.5 hr to 17 hr). Although the time to any and appropriate antimicrobial administration in both of these subsets is notably delayed for dogs ultimately diagnosed with sepsis, it can be speculated that the delayed administration can be attributed to low illness severities when the patients originally presented, which may have affected the primary outcomes. Additional multicenter studies evaluating time to antimicrobial administration for dogs with sepsis and septic shock by disease condition and with quicker antimicrobial administration are needed to further clarify this relationship. Studies that better define which patients benefit from early antimicrobial therapy, as well as which antimicrobials should be administered earliest, are needed.²² It is possible that this large delay in administration (5-7 hr later than human guidelines recommend) may have also eliminated any mortality benefit associated with earlier antimicrobial therapy. Although a prospective, blinded study regarding this relationship would provide further insight, it is unlikely to be performed owing to ethical concerns about delaying antibiotics and could only be performed observationally.

The retrospective design of this study leads to several limitations.¹² Recent prior antimicrobial administration was not controlled for in the present study, which may have affected the significance of associations between times to antimicrobial administration, outcomes, and lengths of hospitalization. Multiple studies in human medicine have found prior antimicrobial administration to affect such relationships, such as the retrospective cohort study by Johnson et al., which found that recent antimicrobial therapy within 90 days of hospitalization significantly increased mortality in cases of gram-negative severe sepsis or septic shock.²³ Additionally, recent antimicrobial exposure was also found to significantly increase lengths of hospitalization and hospitalization costs in human patients with gram-negative–associated severe sepsis or septic shock.²⁴ It is also possible that some of the bacteria cultured were not clinically important, potentially affecting the conclusions made from this study. However, the laboratory at the institution reports mixed, nonpathogenic flora; these do not have susceptibility testing performed and are considered negative cultures. The relationship between shorter hospitalization times and time to appropriate antimicrobials may also reflect the finding that resistant bacteria may be more common in sicker patients, suggesting that correct antibiotic administration may be more likely in healthier patients that make quicker recoveries. Research that further investigates the effect of prior antimicrobial therapy on variables such as outcome and hospitalization length, as well as the clinical importance of cultured bacteria and the presence of resistant bacteria in patients with varying disease severity, can help more accurately determine the importance of timing of antimicrobial therapy on septic veterinary patients. Another limitation of this study includes the finding that ∼80% of dogs included in this study that did not survive to discharge were euthanized, and causes for euthanasia, such as financial constraints or grave prognoses, were not controlled. Because so few dogs died naturally, comparison among those that survived, died, or were euthanized was not possible. Although euthanasia makes overall mortality difficult to assess, its exclusion in the present study would make generalizing these findings difficult, as euthanasia is a normalized procedure in veterinary medicine. Using the SIRS criteria to diagnose sepsis also has its limitations, as the current guidelines result in the inclusion of many patients who would not be considered septic in human medicine. Only 36% of dogs included in this study had available antimicrobial susceptibility results; submission of more prompt culture and susceptibility testing in septic patients, and research examining this larger population, is needed.

Conclusion

This study found a positive association between time to appropriate antimicrobial administration and hospitalization time, as well as between time to appropriate antimicrobial administration and source control. However, no association between timing to any or appropriate antimicrobial therapy and outcome or time to any antimicrobial administration and hospitalization length or source control was identified. Although further research regarding these associations is indicated, these findings propose that timely administration of appropriate antimicrobial treatment along with source control may decrease length of hospitalization, and presumably financial burdens for clients, for dogs with sepsis.

As this was a retrospective study that only reviewed records, no approval was necessary by the Institutional Animal Care and Use Committee.