Are multiple blood cultures advantageous for canine patients?

Abstract

Successful treatment of bacteremic patients depends largely on timely detection of blood-borne pathogens. Failure to detect an infection and/or contamination of blood samples can substantially delay the proper treatment. To increase the detection rate of blood-borne pathogens, well-established guidelines on blood collection and processing have been practiced in human medicine. Investigations involving human blood cultures have shown that the multiple blood sample approach significantly improves the detection rate of bacterial pathogens in the blood. Unfortunately, veterinary-specific blood culture guidelines have not been defined. Therefore, we compared detection rates of blood-borne pathogens between single and multiple blood culture approaches in a retrospective study of the clinical data from canine blood culture cases. We analyzed the data that had been collected over ~6 y and 8 mo from 177 dogs admitted to a veterinary medical teaching hospital. The triple blood culture approach increased the detection rate of blood-borne pathogens by 19.5% compared to single sampling. The optimal timing between multiple sample collections remains to be determined.

Blood culture can be considered the reference (“gold”) standard in diagnosing bacteremia. ⁵ The success of treating bacteremic patients relies on timely detection and subsequent identification of bacteria; failure often results in life-threatening septicemia. False-negative (a failure to detect) or misleading (contamination) culture results can delay proper treatment and increase treatment costs. ⁴ To improve the detection rate, physicians follow well-established guidelines on how to collect and process samples for blood culture. For example, drawing multiple blood samples from a human patient for bacterial culture is the standard practice in the United States recommended by both the Clinical and Laboratory Standards Institute (CLSI) ² and the Infectious Diseases Society of America. ¹⁵ In contrast, the respective veterinary standards are not well-defined. Studies involving human blood cultures have demonstrated that the multiple blood sample approach improves the detection rates of bacterial pathogens in the blood.^2,12 Investigations that have examined bacteremia in dogs and cats have not addressed this knowledge gap.^{6–11,16–18}

There are probably at least 2 reasons for the common practice of veterinary clinicians submitting only a single blood sample per patient for bacterial culture. The first is the cost of bacterial culture, which would be lower for a single culture than for multiple blood cultures. The second is the lack of any veterinary-specific evidence (as opposed to human-specific data) regarding better detection rates achieved by the multiple culture approach. Therefore, our objective was to retrospectively analyze clinical blood culture data to determine if there was an advantage to canine patients in culturing multiple blood specimens over single samples.

We collected data from the electronic medical records of canine patients from March 2015 to November 2021 at the Texas A&M Veterinary Medical Teaching Hospital (VMTH; School of Veterinary Medicine & Biomedical Sciences, Texas A&M University, College Station, TX, USA). Over the period of ~6 y and 8 mo, 186 requests were submitted for blood culture (defined as cases) from 177 dogs. Bacteremia was suspected in all of the cases regardless of the number of samples drawn per patient. Decisions to submit single blood cultures reflected the practice historically established at the VMTH. A possible caveat of our study is that information was not included on whether antimicrobial therapy had been initiated prior to blood sampling. Of 303 blood culture bottles (BCBs) obtained, 235 BCBs were processed aerobically and 68 anaerobically. For 121 cases, a single blood sample was drawn per patient with a total of 85 and 36 requests submitted for aerobic and anaerobic culture (85, 36 hereafter), respectively. Of the remaining 65 cases, 13 cases were double blood sampled (21, 5) and 52 cases were triple blood sampled (129, 27). Double and triple samples were drawn 1–29 min (n = 45), 30–60 min (n = 29), > 1–3 h (n = 34), or > 3 to < 24 h (n = 9) apart.

Upon receipt (d 0), the BCBs submitted for aerobic culture were aerated by inserting sterile needles (Thermo Fisher) through the bottle tops. The BCBs were then incubated aerobically (under 5% CO₂) and anaerobically at 35°C for 10 d. Blood was plated on d 1, 2, and 7 post bottle incubation (pbi). Trypticase soy agar with 5% sheep blood, MacConkey II agar, and chocolate II agar plates (Thermo Fisher) were used for aerobic culture. For anaerobic culture, blood was plated on Brucella blood agar and Columbia CNA agar plates (Hardy). The respective plates were incubated aerobically (under 5% CO₂) and anaerobically at 35°C until bacterial growth was detected, or no longer than 10 d pbi. Bacterial isolates were identified via MALDI-TOF MS (Microflex LT; Bruker).

Based on their clinical significance, the bacteria isolated were categorized into 4 groups. Group I included those that were considered of veterinary clinical significance (Brucella canis, Clostridium perfringens, Enterococcus faecalis, Erysipelothrix spp., Pasteurella canis, Proteus mirabilis, Pseudomonas aeruginosa, Staphylococcus aureus, S. pseudintermedius, Streptococcus canis, S. dysgalactiae). Bacillus cereus, Cellulomonas spp., and Staphylococcus epidermidis were assigned to group II given that these bacteria had been isolated from animals but were considered to be contaminants or of questionable veterinary clinical significance for our study cases. Groups III and IV, respectively, represented bacteria that, to our knowledge, were either documented to have been isolated only from clinical specimens of human origin (Bacillus megaterium, B. pumilus, Devosia spp., Microbacterium spp.) or to have been isolated from neither humans nor animals (Bacillus altitudinis, B. marisflavi, Sphingobacterium daejeonense).

Because the bacterial isolates of groups II–IV were considered contaminants, the 14 cases that had contaminated sample(s) were discarded. Thus, we included 277 samples in our analysis, which represented 114, 22, and 141 single, double, and triple blood samples, respectively. Cultures were positive in 19 of 114 (16.7%) single blood culture cases, 1 of 11 (9.1%) double blood culture cases, and 17 of 47 (36.2%) triple blood culture cases (Table 1). To determine the probability of pathogen detection under the assumption of independence for each sample, contaminated samples (but not their respective cases) were excluded, with a total of 288 blood samples considered for the calculation of an optimal estimate of probability (Fig. 1). We found the observed probabilities of detecting a pathogen in single (16.7%) and triple (36.2%) canine blood cases to be between the optimal and conservative estimates (Fig. 1).

Table 1.

The culture results of 172 single, double, and triple canine blood culture cases.

	Blood culture cases
	Single sample	2 samples	3 samples
1 positive culture result per case	19	0	9
2 positive culture results per case	0	1	2
3 positive culture results per case	0	0	6
Total culture-positive cases	19	1	17
Total culture-negative cases	95	10	30
Total cases included after excluding “contamination” cases	114	11	47

Figure 1.

The probability curve of pathogen detection in canine single and triple blood cultures. The y-axis and x-axis represent probabilities of detecting a pathogen in at least 1 of 3 and 1 blood sample(s), respectively. The blue line represents an assumption in which the outcomes of culturing single and triple blood samples are identical. The closer the curve is to the y-axis, the more beneficial it is to culture multiple samples than a single sample. Under the assumption of independence for each sample cultured, the probability of pathogen detection in individual samples (n = 288) was calculated as the percentage of total positive cultures (n = 52) per total blood samples tested (52/288 × 100% = 18.1%). The optimal estimate of the probability of detecting a pathogen in at least 1 of a random 3-blood-sample set is represented by the solid black line. If a 3-blood-sample set from the same patient is assumed to be dependent, then the conservative probability of detecting a pathogen in the second sample, which is shown by the dashed black line, is decreased when the first sample has tested negative. We observed the probabilities of detecting a pathogen in single sample cases (vertical green line, 16.7%) and triple sample cases (horizontal green line, 36.2%).

A 2022 study showed that the detection rate for single blood samples taken from 45 dogs was 20%. ¹⁷ Investigations that had evaluated multiple human blood samples taken within a 24-h period found that the detection rate increased from 65.1–73.1% for the first blood samples to 80.4–89.7% and 95.7–98.2% for 2 and 3 blood cultures, respectively.^3,12,13 Our detection rate for 2 blood cultures was only 9.1%, which could be explained by a limited sample size (n = 11), and 36.2% for 3 blood cultures (n = 47). Although collecting multiple blood samples at 30–60-min intervals has been common practice, studies have demonstrated that multiple samples drawn (almost) simultaneously had similar detection rates, ² which has been explained by a cumulatively higher amount of blood cultured per human patient.^3,12 Future studies are warranted to provide evidence that the 30–60-min intervals are also unnecessary for veterinary patients.

To further increase detection rates, CLSI recommends that conventional blood cultures be incubated for 7 d despite the fact that 95–97% of pathogenic bacteria and yeasts are recovered within 3–4 d. ² A longer incubation is required for some slow-growing, fastidious bacteria and fungi (e.g., Brucella, mycobacteria, dimorphic fungi). ²

Although blood is considered a sterile sample type, interpretation of blood cultures can still be challenging. ¹ Depending on the bacteria recovered from the blood, it is sometimes impossible to attribute the bloodborne infection to a given isolate, even if the veterinary patient is systemically ill. We suggest that the following criteria be considered as rules of thumb for the interpretation of blood isolates for canine cases.

First, when bacterial isolates are obligate pathogens (e.g., Brucella canis), their clinical significance is obvious. Second, whenever bacterial isolates of blood cases have been associated solely with the environment (i.e., the bacteria are not documented to have been isolated from any clinical samples of animal origin), their interpretation as contaminants is also straightforward. Similarly, questionable clinical significance can also be reported with high confidence when isolates are considered to be common (known) contaminants (e.g., most Bacillus spp.).

However, the most challenging interpretation occurs when bacterial isolates are opportunistic pathogens—those that are present as part of the normal, healthy skin microbial flora but have the capacity to cause or contribute to disease when an opportunity arises (e.g., S. pseudintermedius). ¹⁴ To increase the confidence in assigning clinical significance to these opportunists, the multiple blood culture approach should be adopted, with the blood to be collected from different sampling sites. When a given opportunistic pathogen is recovered from ≥ 2 blood samples, the likelihood of clinical significance is higher.

A blood culture contamination rate of 1% is the target for best practices in human medicine, although a little higher rate (< 3%) is also acceptable. ² However, in veterinary practice, it would probably be challenging to achieve such a low contamination rate. We found contamination in 14 of 186 (7.5%) blood culture cases: 7 of 121 (5.8%) single culture cases, 2 of 13 (15.4%) double culture cases, and 5 of 52 (9.6%) triple culture cases.

Besides established skin disinfection procedures, the following steps could be performed to reduce contamination. First, it has been recommended that peripheral venipuncture be selected in preference to an intravascular catheter for blood collection. ² Second, when multiple blood specimens are collected per patient, the blood should be drawn from different sites. ² The latter procedure, as mentioned above, may also help to confirm the clinical significance of bacterial isolates.

Although we found that a greater number of blood samples collected for culture increased the detection rate of blood-borne pathogens, other knowledge gaps (e.g., timing between multiple sample collections, optimal ratios of blood-to-BCB volumes, a possible correlation between detection rates and blood collection during and after fever) need to be addressed.