Evaluation of the Correlation of Arterial and Venous Lactate and Glucose Concentrations in Anesthetized Critically Ill Dogs

ABSTRACT

Objective

To investigate the correlation between arterial and venous concentrations of lactate and glucose and to explore the relationship between hyperlactatemia and hyperglycemia in critically ill dogs.

Design

Prospective clinical study.

Setting

University veterinary teaching hospital.

Animals

Ninety client‐owned dogs presenting for emergency surgery between June 2015 and March 2024.

Interventions

None.

Measurements and Main Results

Arterial and venous blood samples were collected from the dorsal pedal artery and jugular vein, respectively, ≤5 min apart to measure lactate and glucose concentrations with portable devices. Dogs with known diabetes mellitus were excluded. Based on arterial blood, 35 of 90 dogs (38.9%) were hyperlactatemic, 49 (54.4%) were hyperglycemic, and 9 (10.0%) were hypoglycemic. Based on venous blood, 41 of 90 dogs (45.6%) were hyperlactatemic, 46 (51.1%) were hyperglycemic, and 8 (8.9%) were hypoglycemic. A very strong correlation was found between arterial and venous lactate and glucose concentrations (ρ = 0.850; 95% confidence interval [CI]: 0.766–0.905; p < 0.001 and ρ = 0.970; 95% CI: 0.950–0.982; p < 0.001, respectively). In dogs with arterial hyperlactatemia, a very strong correlation was found between arterial and venous samples (ρ = 0.835; 95% CI: 0.683–0.918; p < 0.001), whereas the correlation was fair in normolactatemic dogs (ρ = 0.488; 95% CI: 0.224–0.685; p < 0.001). The sensitivity and specificity of venous lactate to detect arterial hyperlactatemia were 81.8% and 88.6%, respectively. No correlation was found between arterial lactate and glucose concentrations (ρ = −0.57; 95% CI: −0.261 to −0.152; p = 0.593).

Conclusions

Venous lactate and glucose concentrations correlated well with arterial values; however, hyperlactatemia was not correlated with hyperglycemia. Venous blood sampling offers a practical and reliable method for assessing lactate and glucose concentrations in critically ill dogs, reducing the need for arterial sampling in emergency situations.

Abbreviations

CI

confidence interval

1. Introduction

Blood lactate concentrations are commonly used in critically ill patients to monitor tissue perfusion, to detect anaerobic energy production, and as prognostic indicators [1]. Lactate is derived from carbohydrate metabolism [1, 2]. Hyperlactatemia can be divided into two types [3]. Hyperlactatemia A, in which oxygen delivery is lower than oxygen consumption and can result from increased oxygen demand or decreased oxygen delivery, is the most common type. Hyperlactatemia B, which is most commonly associated with underlying diseases such as sepsis, systemic inflammatory response syndrome, neoplasia, or hepatic insufficiency, occurs when there is no evidence of oxygen deficiency [3]. Hyperlactatemia A is most commonly seen in patients in emergency settings [3, 4], but the two types can occur simultaneously [3]. Evaluation of blood lactate concentrations in critically ill dogs is important to help assess the overall clinical state and prognosis, and it can be used to evaluate response to therapy [1, 2, 5, 6, 7, 8, 9]. In people with sepsis, it has been documented that hyperlactatemia can promote hyperglycemia and vice versa [10, 11]. Hyperglycemia in nondiabetic critically ill dogs has been associated with increased hospitalization time, and dogs with more pronounced hyperglycemia have a higher mortality rate [12]. Similar findings were seen in nondiabetic critically ill people [13]; hence, glucose should be closely monitored in these patients, and therapy may be beneficial to maintain normoglycemia [14, 15].

Arterial lactate is considered a reliable indicator of systemic lactate concentration, whereas venous lactate may vary according to local tissue perfusion at the sampling site [3]. Arterial sampling is the gold standard for assessing blood lactate and glucose concentrations in people [16, 17]. However, it has several disadvantages, including the requirement for greater technical skill, difficulties obtaining blood in hypotensive patients, posing a risk of compromising distal blood flow, and causing more discomfort compared with venous sampling [18, 19]. In human medicine, multiple studies found a high level of agreement between arterial and venous lactate concentrations [18, 20, 21, 22, 23]. However, a systematic review found a poor correlation in people with hyperlactatemia, with venous lactate concentrations being higher than arterial lactate concentrations [24]. In veterinary medicine, only two studies, both performed in healthy dogs, evaluated the difference in lactate concentrations between different blood collection sites. Discordant results were found in the variation between arterial and venous lactate [25, 26]. The first study, conducted in 60 healthy dogs, revealed that the lactate concentrations obtained from the cephalic vein were higher than those obtained from the femoral artery, which were, in turn, higher than samples obtained from the jugular vein [25]. Conversely, the second study, in 23 dogs undergoing elective surgery, found no difference in lactate concentrations between samples obtained from the lingual vein, jugular vein, and dorsal pedal artery [26]. A prospective study in critically ill people revealed that glucose concentrations are underestimated in capillary blood and overestimated in central venous blood samples compared with arterial samples [17]. This is in contrast to a study conducted in 44 dogs after anesthesia that showed that glucose concentrations in arterial blood taken from the dorsal pedal artery were significantly higher than those in venous blood taken from the jugular vein [27].

The current study aimed to investigate the correlation between arterial and venous concentrations of lactate and glucose and the correlation between hyperlactatemia and hyperglycemia in critically ill dogs. Our first hypothesis was that venous lactate and glucose concentrations would correlate with arterial concentrations; our second hypothesis was that there would be a positive correlation between hyperlactatemia and hyperglycemia as is seen in nondiabetic, critically ill people [10, 11].

2. Materials and Methods

The current study consisted of a prospective, convenience sample of critically ill dogs presented for emergency surgery. Any dog that was systemically ill and anesthetized for an emergency surgery, such as dogs presenting with acute abdomen, was considered for inclusion. Dogs that underwent emergency surgery and were unlikely to have hyperlactatemia, such as dogs presenting for acute disc extrusions and dogs with known diabetes, were excluded. Formal institutional approval was granted by the local ethics and welfare committee (EC 2019–40).

All dogs were stabilized before being presented for anesthesia. This stabilization was based on the individual assessment of the attending emergency veterinary surgeon. Once the dogs were fully anesthetized, an arterial catheter was placed in the dorsal pedal artery for arterial blood pressure measurement during anesthesia, as is current practice in our institution. After placement of the catheter, 0.1 mL of blood was collected via the catheter, and lactate and glucose concentrations were immediately measured using portable laboratory devices. Just before or just after arterial blood sampling, 0.1 mL of blood was obtained from one of the jugular veins via direct venipuncture, and lactate and glucose concentrations were also immediately measured using the same devices.

Lactate concentration was measured using a handheld point‐of‐care monitoring device¹. Measurements range between 0.8 and 21.7 mmol/L for whole blood; lactate concentrations of <2.5 mmol/L were considered normal [24]. Glucose concentration was measured using a commercial monitoring system². Measurements range between 1.1 and 41.7 mmol/L for whole blood, and glucose concentrations between 4.4 and 6.6 mmol/L (79–119 mg/dL) were considered to be within normal limits [28]. Previous research showed that both hemodilution and hemoconcentration influence glucose concentrations obtained by the used portable device; therefore, corrected glucose concentrations were calculated as previously published using the following formula: corrected glucose concentration = measured glucose concentration + ([1.6 × hematocrit] − 81.3) [29]. The time required to perform the measurements with the respective portable devices was approximately 60 s for lactate and 15 s for glucose.

Data collection included signalment, indication for surgery, preoperative hematocrit, time between arterial and venous blood sampling, and lactate and glucose concentrations in both samples. Dogs in which arterial and venous samples were collected >5 min apart were excluded.

2.1. Statistical Analysis

Statistical analyses were performed using commercial software³. After recruitment of 20 dogs, a power analysis was performed using preliminary data. A paired two‐tailed t‐test with an α of 0.05 and a power of 80% indicated that 90 dogs were needed. The null hypothesis for power analysis was that no differences were present between arterial and venous lactate concentrations, whereas the alternative hypothesis was that a mean difference of 0.5 mmol/L would be present between arterial and venous lactate concentrations. This showed that a total of 90 dogs needed to be included. Normality of variables was assessed using Shapiro–Wilk tests. Wilcoxon matched‐paired signed rank test was used to compare arterial and venous concentrations of lactate and glucose, respectively. Correlations between arterial and venous lactate and glucose were evaluated using two‐tailed Spearman rho tests, and 95% confidence intervals (CIs) were estimated based on Bonett and Wright. Correlation coefficients were considered very strong if ρ > 0.8, moderately strong if ρ = 0.6–0.8, fair if ρ = 0.3–0.5, and poor if ρ < 0.3 [30]. Results were considered statistically significant if p ≤ 0.05.

Using arterial lactate concentrations as the gold standard, sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of the paired venous lactate concentrations were determined. Finally, Bland–Altman plots were created to evaluate differences between arterial and venous lactate and glucose concentrations. Bias (mean difference) and 95% (mean difference ± 1.96 SD) limits of agreement were calculated and plotted.

3. Results

A total of 93 dogs were enrolled in the current study, of which three dogs were excluded as the time interval between arterial and venous blood sampling was >5 min.

A total of 44 breeds were included; the most common breeds were German Shepherd Dog (n = 9), Labrador Retriever (n = 6), and Malinois (n = 6). Other breeds were represented by three or fewer dogs each. The median age was 77.5 months (range: 3–148 months), and the median body weight was 29.9 kg (range: 3.1–88.0 kg). There were 54 male dogs (21 neutered) and 36 female dogs (19 neutered). Dogs were presented for emergency surgery for diverse reasons, including septic peritonitis (n = 27), gastric dilatation–volvulus (n = 25), intestinal foreign body (n = 13), hemoabdomen (n = 6), bile peritonitis (n = 5), pyometra (n = 5), neoplasia with ruptured masses or causing sterile peritonitis (n = 3), mesenteric torsion (n = 3), and acute gastroenteritis, preeclampsia, and obstructive cholelithiasis (n = 1 each). The median time elapsed between arterial and venous blood sampling was 2 min (range: 1–5 min).

The median arterial and venous lactate concentrations were 1.6 mmol/L (range: <0.8–13.8 mmol/L) and 2.3 mmol/L (range: <0.8–13.3 mmol/L), respectively, which were not different (p = 0.407). Based on arterial blood, 35 of 90 dogs (38.9%) were hyperlactatemic compared with 41 of 90 dogs (45.6%) based on venous blood sampling. Ten hyperlactatemic dogs had arterial lactate concentrations ≥5.0 mmol/L. Hyperlactatemia was present in both arterial and venous blood in 31 of 90 dogs (34.4%); 4 of 90 dogs (4.4%) had hyperlactatemia based on arterial samples only, and 10 dogs (11.1%) had hyperlactatemia based on venous blood only (Table 1).

TABLE 1.

Simultaneous arterial and venous lactate and glucose concentrations in 90 anesthetized, critically ill dogs.

Arterial		Lactate concentration n = 90		Glucose concentration n = 90			Corrected glucose concentration n = 88
Venous
		<2.5 mmol/L	≥2.5 mmol/L	<4.4 mmol/L	4.4–6.6 mmol/L	>6.6 mmol/L	<4.4 mmol/L	4.4–6.6 mmol/L	>6.6 mmol/L
Lactate concentration n = 90	<2.5 mmol/L ≥2.5 mmol/L	45 10	4 31	0 8	24 12	25 21	4 8	24 14	19 19
Glucose concentration n =90	<4.4 mmol/L 4.4–6.6 mmol/L >6.6 mmol/L	1 23 31	9 8 18	8 1 0	0 30 2	0 5 44	6 5 1	2 24 12	0 5 33
Corrected glucose concentration n = 88	<4.4 mmol/L 4.4–6.6 mmol/L >6.6 mmol/L	7 27 19	10 12 13	8 1 0	6 21 3	3 17 29	11 1 0	6 31 1	0 7 31

When considering all measurements, the correlation between arterial and venous lactate concentrations was very strong (ρ = 0.850; 95% CI: 0.766–0.905; p < 0.001) (Figure 1). In dogs with arterial hyperlactatemia, a very strong correlation was found between arterial and venous samples (ρ = 0.835; 95% CI: 0.683–0.918; p < 0.001). However, in dogs with arterial normolactatemia, only a fair correlation was found between arterial and venous samples (ρ = 0.488; 95% CI: 0.224–0.685; p < 0.001). The Bland–Altman plot revealed that the agreement between arterial and venous lactate concentrations was high and identified only a very small systematic bias (0.03 mmol/L), meaning that only a small average difference was present (Figure 2). At lactate concentrations <5.0 mmol/L, arterial and venous lactate concentrations were equivalent. However, in the 10 samples in which arterial lactate concentrations were ≥5.0 mmol/L, more discordant results between arterial and venous values were apparent on the Bland–Altman plot (Figure 2). Nevertheless, in all those cases, hyperlactatemia was diagnosed based on arterial as well as venous blood samples. Using arterial lactate concentrations as the gold standard (i.e., presence of true normolactatemia or hyperlactatemia), the sensitivity of correctly classifying venous blood samples as normolactatemic or hyperlactatemic was 81.8%, whereas the specificity was 88.6%. The positive predictive value was 91.8%, the negative predictive value was 75.6%, and the accuracy 84.4%.

FIGURE 1.

Correlation between simultaneous arterial and venous lactate concentrations in 90 anesthetized critically ill dogs.

FIGURE 2.

Bland–Altman plot of the differences between simultaneous arterial and venous lactate concentrations from 90 anesthetized critically ill dogs. The upper and lower dashed lines represent the 95% limits of agreement, and the solid line indicates the mean difference (bias).

The median arterial and venous glucose concentrations were 6.8 mmol/L (122 mg/dL) (range: 2.8–25.8 mmol/L [50–464 mg/dL]) and 6.7 mmol/L (121 mg/dL) (range: 2.6–23.7 mmol/L [47–427 mg/dL]), respectively, which were significantly different (p < 0.001). Based on arterial blood, 49 of 90 cases (54.4%) were hyperglycemic, and 9 (10.0%) were hypoglycemic, whereas, based on venous blood, 46 of 90 dogs (51.1%) were hyperglycemic, and eight (8.9%) were hypoglycemic. Based on both arterial and venous blood samples, 44 of 90 dogs (48.9%) were hyperglycemic, and 8 (8.9%) were hypoglycemic. In 2 of 90 cases (2.2%), hyperglycemia was evident only in the venous sample and in five cases (5.6%) only on the arterial sample. One dog (1/90 [1.1%]) was hypoglycemic only on the arterial sample (Table 1).

Preoperative hematology was available for 88 dogs, and the mean hematocrit was 44.2 ± 13.0%. The corrected arterial and venous glucose concentrations were 6.4 mmol/L (115 mg/dL) (range: 2.9–22.9 mmol/L [52–412 mg/dL]) and 6.1 mmol/L (110 mg/dL) (range: 2.6–20.8 mmol/L [47–374 mg/dL]), respectively, and were significantly different (p < 0.001). Based on corrected arterial glucose concentrations, 38 of 88 dogs (43.2%) were hyperglycemic, and 12 (13.6%) were hypoglycemic, whereas, based on corrected venous glucose concentrations, 32 of 88 dogs (36.4%) were hyperglycemic, and 17 (19.3%) were hypoglycemic. Hyperglycemia was present based on both corrected arterial and venous samples in 31 of 88 dogs (35.2%), whereas 11 (12.5%) were hypoglycemic on both corrected arterial and venous samples. In 1 of 88 dogs (1.1%), hyperglycemia was only present on the venous sample based on corrected glucose concentrations, whereas seven dogs (8.0%) had hyperglycemia based only on corrected arterial samples. Hypoglycemia was only present in corrected venous samples from 6 of 88 dogs (6.8%) and in 1 (1.1%) corrected arterial sample (Table 1). A very strong correlation was found between arterial and venous glucose concentrations (ρ = 0.970; 95% CI: 0.950–0.982; p < 0.001; corrected glucose ρ = 0.959; 95% CI: 0.933–0.975; p < 0.001) (Figure 3). In dogs with arterial hyperglycemia, the correlation between arterial and venous glucose concentrations was very strong (ρ = 0.922; 95% CI: 0.850–0.960; p < 0.001; corrected glucose ρ = 0.868; 95% CI: 0.733–0.937; p < 0.001). Likewise, in dogs with arterial hypoglycemia, the correlation between arterial and venous glucose concentrations was very strong (ρ = 0.824; 95% CI: 0.237–0.970; p = 0.006; corrected glucose ρ = 0.869; 95% CI: 0.509–0.970; p < 0.001). In dogs with arterial normoglycemia, the correlation between arterial and venous concentrations was very strong if actual glucose concentrations were used; however, the correlation was only moderately strong if corrected glucose concentrations were used (ρ = 0.883; 95% CI: 0.744–0.949; p < 0.001; corrected glucose ρ = 0.705; 95% CI: 0.467–0.847; p < 0.001). The bias of the Bland–Altman plot between arterial and venous glucose concentrations was small (0.25 mmol/L [5 mg/dL]; corrected glucose 0.26 mmol/L [5 mg/dL]) (Figure 4). Although arterial and venous glucose concentrations were similar in most dogs, disagreement between both was substantial in some dogs with normoglycemia and hyperglycemia.

FIGURE 3.

Correlation between simultaneous corrected arterial and venous glucose concentrations in 88 anesthetized critically ill dogs (hematocrit was missing in two dogs, in which no corrected glucose could be calculated).

FIGURE 4.

Bland–Altman plot of the differences between corrected arterial and venous glucose concentrations from 88 anesthetized critically ill dogs (hematocrit was missing in two dogs in which no corrected glucose could be calculated). The upper and lower dashed lines represent the 95% limits of agreement, and the solid line indicates the mean difference (bias).

Finally, no correlation was found between arterial lactate and glucose concentrations (ρ = −0.57; 95% CI: −0.261 to 0.152; p = 0.593; corrected glucose ρ = −0.035; 95% CI: −0.243 to 0.175; p = 0.744) (Figure 5) or between venous lactate and glucose concentrations (ρ = −0.062; 95% CI: −0.266 to 0.147; p = 0.560; corrected glucose ρ = −0.028; 95% CI: −0.236 to 0.183; p = 0.799).

FIGURE 5.

Correlation between simultaneous arterial lactate and corrected arterial glucose concentrations in 88 anesthetized critically ill dogs (hematocrit was missing in two dogs in which no corrected glucose could be calculated).

4. Discussion

The current study showed a very strong correlation between arterial and venous lactate concentrations, as well as between arterial and venous glucose concentrations, supporting our first hypothesis. In contrast, no correlation was observed between hyperlactatemia and hyperglycemia, thereby rejecting our second hypothesis. In both critically ill human and veterinary patients, blood lactate concentration is considered a valuable prognostic indicator [1, 2, 5, 6, 7, 8, 9, 26]. In human medicine, serial lactate measurements are recommended to guide treatment over a single pretreatment concentration [22]. In people with sepsis, decreasing lactate concentrations over time are associated with a favorable response to treatment and, hence, a greater chance of survival [31]. A study in dogs with severe and complicated canine babesiosis showed that serial lactate measurements were useful to predict survival [2]. Similarly, good indications for survival in dogs after gastric dilatation–volvulus surgery were shown with repeat postoperative lactate concentrations [5].

Despite the fact that arterial lactate concentration measurements are considered the gold standard in people [16, 24], most studies have been performed using venous lactate concentration measurements [2, 5, 7, 8, 9, 32], or used arterial and venous lactate concentrations interchangeably [6]. It is currently unknown whether arterial lactate concentrations would provide more accurate information about tissue perfusion compared with venous lactate concentrations, or whether similar information regarding prognosis would be provided.

In the veterinary literature, there are two previous studies in healthy dogs that explicitly aimed to characterize the variations in arterial and venous lactate concentrations [25, 26]. The current study in critically ill dogs demonstrated only a fair correlation but a good agreement between lactate concentrations in arterial and venous samples for arterial normolactatemic samples, whereas a very strong correlation was observed in cases of arterial hyperlactatemia; however, the agreement was far less whenever arterial hyperlactatemia was ≥5.0 mmol/L. A similar discrepancy was also observed in convenience samples of human emergency patients [33]. Discrepancies can reflect differences in sampling site or differences in local tissue perfusion. Arterial blood is considered to contain a representative concentration of the lactate present in the entire body, whereas venous lactate concentrations can vary more depending on the local tissue perfusion from where the sample is obtained [3]. Venous blood obtained from the jugular vein, as was done in the current study, contains blood of the cranial portion of the body. The fact that most dogs in this study presented with acute abdominal disease might explain, in part, the discrepancies noted between arterial and venous lactate concentrations.

A systematic review regarding lactate measurements in adults in the emergency department reported a sensitivity of venous lactate concentrations of 94%–100%, meaning that normal venous lactate concentrations accurately rule out arterial hyperlactatemia in people [24]. In our study cohort, the presence of normal venous lactate concentrations was likely to indicate normal arterial lactate concentrations (positive predictive value of 91.8%). Although multiple studies performed in people with sepsis suggest using venous samples as surrogates for arterial samples to assess lactate concentration [18, 20, 21, 22, 23], other studies advise taking an additional arterial blood sample whenever venous lactate is increased [24, 34]. One study in people presenting to an emergency service showed that the presence of venous hyperlactatemia more than doubled the odds of having arterial hyperlactatemia [33]. In the current study, it was unlikely that venous hyperlactatemia was found in dogs with arterial normolactatemia (negative predictive value was 75.6%). Furthermore, despite the discordant results between arterial and venous lactate concentrations on the Bland–Altman plot when arterial hyperlactatemia was ≥5.0 mmol/L, hyperlactatemia (present vs. absent) would be diagnosed on both arterial and venous blood samples in all those cases, which is a clinically important finding. Based on the results of the current study, venous lactate concentrations can be used in critically ill dogs as a substitute for arterial concentrations. However, it is important for lactate concentrations to be interpreted individually. Venous hyperlactatemia can indicate the presence of local tissue hypoperfusion, even in cases of arterial normolactatemia, and may need to be addressed.

In this study, the severity of the arterial hyperlactatemia seemed to be underestimated in dogs with severe venous hyperlactatemia. Conversely, some dogs with mild venous hyperlactatemia showed arterial normolactatemia. In an experimental study in Beagles in which hepatic blood flow was occluded and lactate concentrations measured over time, venous (jugular) and arterial (femoral) hyperlactatemia developed in parallel; yet, the mean venous lactate concentration tended to be consistently higher than the arterial [35]. Unfortunately, no individual measurements were provided, nor were correlations reported in that study; therefore, we cannot conclude that our results are truly discordant.

Previous studies in healthy and postanesthetic dogs reported that arterial blood glucose concentrations were marginally but significantly higher than venous concentrations [27, 36]. The same observation was made in the dogs presenting for emergency surgery in the current study. A very strong correlation was found between arterial and venous glucose concentrations in arterial hyperglycemia, normoglycemia, and hypoglycemia. Nevertheless, disagreement existed between arterial and venous glucose concentrations in some dogs with normoglycemia and hyperglycemia.

Similar to lactate concentrations, glucose concentrations were studied as prognostic indicators in critically ill human and veterinary patients [2, 10, 11, 12, 28, 37]. In a large prospective study, nondiabetic critically ill dogs that did not survive hospitalization had higher glucose concentrations than survivors [12]. The odds ratio for mortality in nondiabetic people with suspected sepsis was higher for those with hyperlactatemia associated with hyperglycemia, compared with those with normoglycemia [11]. In a retrospective study in nondiabetic critically ill dogs, those with hyperlactatemia had higher glucose concentrations compared to dogs with normolactatemia [28]. In the cohort of anesthetized, critically ill dogs in the current study, however, no correlation was found between hyperlactatemia and hyperglycemia.

As it is known that the portable device used to determine glucose concentrations is affected by both hemodilution and hemoconcentration [29], corrected glucose concentrations were calculated in the current study. In people, it has been shown that hemoconcentration decreased lactate concentrations, especially in cases of hyperlactatemia [38]. Nevertheless, a study in dogs revealed that lactate concentrations measured by the portable device used were not different than a gold standard test for analyzing lactate concentrations; however, only 10 of 125 dogs had hemoconcentration in that study [39]. Unfortunately, hematology was performed before surgery in the current study and not simultaneously with measurement of lactate and glucose. As some dogs presented with hemoabdomen, their hematocrits would likely have changed by the time of lactate and glucose measurement, under anesthesia for corrective surgery. Increased insulin concentrations and relative insulin resistance can cause hyperglycemia in critically ill people [40]. In contrast to people, stress hyperglycemia is less commonly seen in dogs [12]. A large prospective study in nondiabetic critically ill dogs revealed that 11.4% of dogs were hyperglycemic at presentation, which increased to 15.5% during hospitalization [12]. In a study including 50 critically ill people, 28% had hyperglycemia at presentation, which increased to 64% after 1 week of hospitalization [41]. More studies are needed to determine the influence of hyperlactatemia and hyperglycemia in critically ill dogs and to better understand their molecular pathways.

Species‐specific reference intervals for hyperlactatemia are described in the literature and textbooks. In dogs, 2.5 mmol/L is generally considered the upper limit [25, 42]. Nevertheless, particularly in retrospective studies, it is often not mentioned which type of blood sample has been evaluated. This is remarkable because there is some evidence that, not only do samples derived from different blood collection sites result in different values, but the determination of lactate concentrations in different blood components (whole blood, plasma, serum) and by different instruments can also influence the results [39, 43]. In the current study, whole blood was used as a sample for both lactate and glucose, which in theory equals plasma concentrations in the absence of hemolysis due to difficult blood sampling.

This study has limitations. Resuscitation protocols and timing of sample collection after resuscitation were not standardized. All samples were taken after stabilization of the dogs while they were anesthetized, and likely both resuscitation and anesthesia will have influenced the lactate and glucose concentrations found. Nevertheless, as this study was designed to assess differences between arterial and venous concentrations, the actual values obtained were not used as prognostic indicators for the dogs. Also, although no alterations are expected within the 5‐min timeframe between measurements, simultaneous sampling might have provided even greater accuracy. This was not possible, however, due to the availability of only one device per parameter. In most cases, arterial blood was obtained just before the jugular blood sample. The arterial blood was collected from the dorsal pedal artery upon placement of a catheter for invasive blood pressure measurement before the line was attached. Venous blood was always collected via puncture of a jugular vein, as it was easily accessible and peripheral catheters were often already placed and used for stabilization, before anesthesia, which would have resulted in inadvertent dilution by IV fluids. Although both portable devices used are validated in dogs [39, 44, 45], neither of them is gold standard to determine lactate and glucose concentrations in canine blood. In addition, corrected glucose values may introduce variability. The hematocrit used to calculate corrected glucose concentrations was determined before anesthesia and, hence, might have differed from the dogs’ hematocrits at times of blood sampling. Finally, as blood pressure might have an influence on the agreement between arterial and venous lactate concentrations, it would have been valuable to take this into consideration. Nevertheless, as neither timing nor the method of blood pressure measurement was recorded at the time of blood sampling for the current study, this information was discarded to avoid introducing bias. The influence of blood pressure on arterial and venous lactate concentrations should be further examined, as it might potentially affect these levels differently.

In conclusion, this study demonstrated that venous blood sampling from a jugular vein may offer a practical and reliable alternative to arterial sampling from a dorsal pedal artery to evaluate lactate and glucose concentrations in critically ill dogs. However, in cases of severe hyperlactatemia (≥5.0 mmol/L), larger discrepancies between arterial and venous measurements were noted. Given the fact that, for prognostic relevance, serial lactate measurements are recommended, arterial and venous sample sites should not be used interchangeably; rather, the same site of blood collection should be selected to increase the accuracy of the evolution in time. The same applies to the glucose concentrations, which tended to be consistently lower in the jugular vein blood compared with the dorsal pedal artery blood. No relationship was found between lactate and glucose concentrations, indicating that hyperlactatemia and hyperglycemia may occur independently in critically ill dogs.

Author Contributions

Hilde de Rooster: conceptualization, writing – review and editing, supervision, project administration. Francesca Tirillini: writing – original draft, writing – review and editing, resources, visualization, software. Nausikaa Devriendt: conceptualization, investigation, writing – review and editing, supervision, data curation, formal analysis, software, validation, project administration. Ingeborgh Polis: methodology, supervision, validation, visualization.

Disclosure

The preliminary results of this study were presented at the Annual Scientific Meeting of the ECVS on July 6, 2019, Budapest, Hungary.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.