Introduction

Blood work is routinely performed in human and veterinary medicine, often several times daily. Blood may be obtained by direct venipuncture, via a peripheral arterial or venous catheter, or via a central venous indwelling catheter (CVC). Three methods of blood sampling exist when an indwelling catheter is used: discard, reinfusion, and push/pull.¹ The discard technique involves aspirating a predetermined volume of blood from the indwelling catheter in order to clear the dwell volume and various contaminants, then disposing of that blood before obtaining the sample from the catheter for analysis. Discarding, rather than reinfusing the blood, is the most widely practiced method of sampling in human medicine because of concerns for blood contamination, blood clots forming in the syringe, and hemolysis of the aspirated volume.^1,2

The discard or presample volume (PSV) during blood sampling varies between hospitals and is not standardized, with studies reporting conflicting results. Some studies advocate withdrawing a certain percentage of the catheter dead space, whereas others recommend withdrawing a specified volume of blood. The matter is further complicated by a wide array of available indwelling catheters that are placed differently, are manufactured from different materials, and can vary substantially in size.

Typically, reinfusion is practiced in patients of the smallest size, such as pediatric and neonatal patients, in whom discarding any volume of blood might result in significant blood loss. In these patients, withdrawing the smallest amount of blood is recommended to lessen the chance of blood clotting or contamination before reinfusion, as well as to minimize the patient’s blood loss. The practice of reinfusion is standard in small animal medicine because veterinary patients are frequently small and are likely to rapidly become anemic if the withdrawn volume before sampling is discarded. The withdrawal volume that best approximates values obtained via direct venipuncture has not been studied in small animal patients, regardless of whether it is discarded or reinfused. At the authors’ practice, as well as a number of veterinary institutions, it is customary to withdraw a set amount of volume, either 2.5 or 5 mL, depending on the size and clinical status of the patient and what is perceived as safe. This would typically constitute anywhere from 3–18 times (300–1800%) of the catheter dead space volume, depending on the size of the catheter and the utilized port.

A few veterinary studies compared obtaining various blood tests from indwelling catheters to direct venipuncture. One study reported no significant difference in complete blood cell count and serum biochemistry results when comparing samples drawn from a CVC to jugular venipuncture in horses.³ This study used 3 times the dead space of the catheter as a discard volume. Another study found no significant differences looking at thromboelastographic values collected via various venipuncture techniques with blood collected from the standard size CVC using a 5 mL PSV that was reinfused.⁴ A study by Maeckelbergh and Acierno used a 3 mL PSV when comparing standard coagulation tests and fibrinogen level collected via a CVC to direct venipuncture in sick dogs and reported no clinically significant differences between the collection methods.⁵ No information is available in veterinary medicine regarding the accuracy of venous blood gas (VBG) values obtained via a CVC. VBGs are one of the most commonly obtained tests in veterinary emergent patients in which precise results are critical. In addition, there is enough conflicting information about obtaining samples for a coagulation panel from a CVC to warrant further investigation.^5–9

The objective of the current study was to evaluate the effect of different catheter PSV on laboratory measurements in small animal clinical patients and compare them with direct venipuncture. The study aimed to establish guidelines regarding the minimum PSV necessary for accurate measurement of coagulation tests and VBG values. The null hypothesis was that no difference would be detected among values collected using different PSV when compared with direct venipuncture.

Materials and Methods

This was a prospective observational study performed at the Matthew J Ryan veterinary hospital at the University of Pennsylvania after Institutional Animal Care and Use Committee approval was obtained. Dogs were included when treatment for their medical condition required the presence of an indwelling CVC and owner consent was obtained. Dogs of any age, sex, type of disease, or severity of illness were included. Patients were excluded if they weighed <10 kg, had a packed red blood cell volume (PCV) of <20%, or could not be reliably sampled from the CVC or if the blood samples could not be drawn via direct venipuncture. Additionally, dogs with cardiovascular instability receiving rapid volume resuscitation or blood product transfusions were excluded.

The distal port of the catheter was used for blood sampling. CVC size and manufacturer were recorded, and the dead space volume of the port was determined by consulting the packaging information. Each enrolled dog had four samples collected—direct venipuncture (sample 1), 300% of the catheter dead space (sample 2), 600% of the catheter dead space (sample 3), and 1200% of the catheter dead space (sample 4). The order of samples was randomized by an online randomization generator program.^a All samples were collected within 30 min of each other. Blood sample collection from the CVC was performed by two of the authors, and venipuncture was performed by the nurse caring for that patient. The sample for the direct venipuncture was collected from the jugular, cephalic, or saphenous vein using a 3 mm syringe and a 21 gauge needle. A total of 2.5 mL of blood was collected with each sample; 0.5 mL was transferred to 20 U/mL lithium heparin blood gas syringe for immediate VBG analysis,^b which included pH, venous carbon dioxide partial pressure, venous oxygen partial pressure, bicarbonate concentration, extracellular base excess, sodium (Na⁺), potassium (K⁺), chloride (Cl⁻), ionized calcium, lactate, and blood glucose measurements. An allotment of 1.8 mL of blood was transferred into a 3.2% sodium citrate tube and centrifuged at 1300g for 15 min before plasma was separated, frozen, and saved for batch testing of the coagulation panel;^c this included prothrombin time, activated partial thromboplastin time, and D-dimer concentration. Batch testing was performed twice during the course of the study, with each batch including 10 patients. The remainder of the blood was used to fill the microhematocrit tubes for PCV and total solids (TS) measurement via centrifugation at 14,110g for 3 min. The TS were determined on separated plasma by a refractometer.

The procedure for blood collection from the CVC was as follows: the injection cap of the distal sampling port was removed, and the predetermined PSV was pulled directly from the port into a syringe containing 0.5 mL heparinized saline with heparin concentration of 1 U/mL. After the PSV was collected, the sample was obtained by attaching a 3 mL syringe to the port. The withdrawn volume was then returned to the patient via the same port, and the port was flushed with 2 mL of 100 U/mL heparinized saline.

Most continuous variables were not normally distributed as determined visually and by the Skewness and Kurtosis tests for normality. Therefore, results are reported as median (and range), and the nonparametric Wilcoxon signed rank sum test was used for paired comparisons of continuous variables. The values obtained via direct venipuncture served as control values and were compared with the values obtained via different PSV for significant differences. In addition, different PSV were compared among each other as well. P < .05 was considered significant for all tests. All statistical evaluations were performed using a statistical software package.^d

Results

Twenty dogs were enrolled in the study. Median age was 4.5 yr (range, 1–15 yr). Five dogs were castrated males, seven were intact males, seven were spayed females, and one was an intact female. Breeds included Labrador retriever (5), mixed-breed dog (4), golden retriever (2), German shepherd dog (2), Bernese mountain dog (1), Great Dane (1), Siberian husky (1), English bulldog (1), collie (1), Pembroke Welsh corgi (1), and bull terrier (1). Median body weight was 31.6 kg (range, 10–55.8 kg). The reasons for placement of CVC included postoperative care of coil embolization for intrahepatic portosystemic shunt (7), gall bladder mucocele (4), septic peritonitis (2), cholangiohepatitis (1), colonic torsion (1), nephrectomy (1), biliary obstruction due to inflammatory duodenal mass (1), hit by car (1), gunshot wound (1), and uroabdomen (1).

The catheter manufacturer was Arrow^e in 18 dogs and Mila^f in 2 dogs. All dogs had 7 French catheters placed, and lengths ranged from 16 to 60 cm. Dead space of the distal port of the catheter ranged from 0.39 to 0.8 mL (Table 1).

TABLE 1 Lengths, Sizes, and Dead Space Volumes of the Catheters Used in the Study

View Fullscreen

Median values and ranges for all samples are listed in Table 2. The control values were significantly higher than those obtained from the indwelling catheter for PCV (300% PSV [P = .007], 600% PSV [P = .005], and 1200% PSV [P = .02]), TS (300% PSV [P = .006] and 600% PSV [P = .04]), and lactate (600% PSV [P = .04] and 1200% PSV [P = .01]). Control values were significantly lower than those obtained for pH (1200% PSV [P = .008]) and chloride (300% PSV [P = .04], 600% PSV [P = .003], and 1200% PSV [P = .03]). A significant increase was found in prothrombin time in samples drawn with 600% PSV compared with 1200% PSV (P = .008).

TABLE 2 Median Laboratory Values for Samples Obtained with Venipuncture and Various Presamples

View Fullscreen

Discussion

This study found significant differences in a number of blood values collected at various PSVs when compared with direct venipuncture. The PCV was substantially diluted by using all PSVs, and TS were significantly affected by the two smaller PSVs. This may have clinical implications for hospitalized patients, especially those patients who suffer from anemia, in whom a PCV drop of 2–3% may potentially be a trigger for further diagnostic investigation about the cause of the PCV drop or transfusion, depending on anemia severity. A study by Yucha and DeAngelo concluded that 3 times the catheter dead space is necessary to obtain a value within 1% of a true hematocrit from a peripheral catheter.¹⁰ Those findings were not supported by our study, in which a 1200% PSV best approximated the PCV and TS obtained by venipuncture.

When statistically significant VBG and coagulation parameters were evaluated, the differences did not appear clinically significant and would have no impact on clinical decision making in the evaluated patients. Similar findings were reported by other studies, in which statistically significant but clinically irrelevant differences were reported between the sample collection methods.^2,11 Therefore, 300% PSV may be adequate to evaluate the coagulation and VBG values. Several studies report the discard volume of 2–5 times the catheter dead space as an adequate amount to yield accurate results for electrolytes and oxygenation parameters from arterial lines in adults and children.^12,13 This is in line with the findings in the current study, and, although arterial oxygenation parameters were not evaluated, venous carbon dioxide partial pressure and venous oxygen partial pressure were minimally affected by the PSV. Among significantly different VBG values, chloride values were consistently higher in samples obtained with all of the PSV compared with venipuncture. One possible explanation is that 0.9% NaCl (normal saline) could have been administered through the CVC port used for sampling as part of fluid therapy or as a diluent for various medications. Normal saline contains a significantly higher chloride concentration compared with the physiologic canine values; it is feasible that none of the PSVs were sufficient to completely clear the chloride from the catheter lumens. It warrants mention, however, that even though the chloride differences are statistically significant, they are likely not clinically relevant and would not have resulted in additional diagnostic or therapeutic interventions.

In respect to the coagulation testing, this study differs from previous reports on obtaining coagulation samples from indwelling catheters in human patients. Six times the catheter dead space has been recommended to obtain accurate activated partial thromboplastin time from arterial lines.⁶ Several investigations indicated that no discard volume is large enough to consistently result in accurate coagulation parameters due to heparin contamination and recommended direct venipuncture for accurate coagulation results.^7–9 The difference between our results and these studies could be in the heparin concentration of the dwell volume within the catheters, as well as the inability to accurately quantify the catheter dead space in certain types of catheters that are tailored for individual patients during placement.¹⁴ One veterinary study corroborated our results, concluding that obtaining prothrombin time, activated partial thromboplastin time, and fibrinogen concentration via an indwelling catheter was a clinically acceptable alternative to direct venipuncture in sick dogs.⁵ Another study in healthy dogs did not find significant differences in coagulation values obtained via direct venipuncture compared with a jugular catheter for up to 48 hr after catheter placement.¹⁵ Finally, a study comparing platelet function and coagulation variables obtained via several different catheter types, including peripheral venous catheters and CVCs, did not find significant differences in coagulation variables between methods of collection.¹⁶ This study used 4–5mL discard volumes, and dead space volumes of the central lines were unavailable.

Because of significant dilutional effects, as well as statistically significant differences in a number of tested values, uniform blood sampling practices should be established at each institution, as well as in each patient, so that comparisons between values are the most representative of the patient status and are not affected by the testing method. Standardization of PSVs is also recommended when serial tests are performed in a given patient. In addition, if an injection cap is used on the catheter port, the dead space of the injection cap should be factored in, and uniform sampling should be performed with the cap on or off. If the injection cap is used, the gauge and length of the needle used to withdraw the sample should be standardized as well.

The sampling technique may influence the development of iatrogenic anemia in the intensive care unit. Balakrishnan, Drobatz, and Reineke documented the development of anemia in critically ill cats; cats with sampling lines were sampled more frequently with larger amounts of blood withdrawn.¹⁷ Although the sampling method was not described and the development of anemia is likely multifactorial, it is possible that the anemia is secondary to a large PSVs, reinfusion and subsequent hemolysis, discard of PSVs, or a combination of these effects. The push/pull or mixing method has been evaluated in veterinary medicine as an alternative to the discard or reinfusion method. It involves repeated aspiration and reinfusion of 300% of the catheter dead space volume without disconnection of the syringe from the catheter before obtaining the sample for analysis with a different syringe. This method may reduce the possibility of blood loss from discard volume, infection from catheter disconnection, and reinfusion of blood clots. The push/pull method was evaluated in dogs under anesthesia and was an acceptable method for collecting VBG values in dogs with no significant changes in PCV reported between direct venipuncture and blood collected from the catheter.² This method should be considered for reporting of PCV if direct venipuncture or collection of large PSVs are undesirable. The push/pull technique has not been evaluated for collection of hemostatic parameters, in which induced blood turbulence may be a concern. In addition, further studies are necessary to evaluate the effects of repeated sampling via the push/pull method on sample hemolysis.

This study has several limitations. The sample size is small, and, therefore, it is possible that the study is underpowered and may have failed to detect significant differences between the variables. Further studies will be necessary to elucidate if the larger PSV is superior in approximating PCV and TS compared with smaller PSV. In addition, this patient population did not have a large variation in VBG parameters and electrolytes, which makes it challenging to extrapolate these findings to patients with severe acid–base, electrolyte, blood glucose, and lactate derangements. Further studies, including a broader range of patients with specific blood work derangements as well as larger patient numbers, will be useful to elucidate these questions. Finally, withdrawing a percentage of the dead space of the catheter requires familiarity with the size of the placed catheter and customization of the PSV to that particular patient. Obtaining a fixed volume as a PSV may be more conducive to simplifying the procedure and minimizing error, although it may not be as precise. Disparate results are reported when aspirating a predetermined volume of blood to discard. Villalta-García et al. reported that a discard volume of 2 mL was sufficient to obtain an accurate sample from a CVC for a complete blood cell count, serum chemistry, and a coagulation profile, compared with direct venipuncture.¹⁸ Another study by Wyant and Crickman concluded that a 6 mL discard volume was necessary to avoid dilutional effects when sampling from nontunneled CVC.¹⁴ It is possible that the necessary PSV depends on the size of the catheter and the values measured and cannot be standardized. Fixed volumes might be advisable when a dwell volume of the catheter cannot be determined, as in vascular access ports that are cut as needed for the size of the animal. One study in cats compared hematologic and biochemical values obtained via a vascular access port and direct venipuncture.¹¹ This study used a 1.5 mL PSV that was subsequently reinfused, and this technique yielded comparable results to venipuncture for the evaluated parameters. Fixed PSVs were not evaluated in the current study.

Another limitation of this study is that there was no correction for multiple testing, and therefore, a type 1 statistical error is possible. However, correcting for multiple testing can increase the possibility that a true significant difference is not detected. In a small exploratory study such as this, it is important to avoid the risk of such false-negative results.¹⁹

Conclusion

In summary, 300% or 3 times the catheter dead space volume was adequate to obtain VBG and coagulation values from CVC similar to direct venipuncture in dogs in this study. A significant dilutional effect was found with all PSVs for the PCV, with 1200% PSV generating results closest to the PCV obtained from venipuncture. If the most accurate PCV is desired, direct sampling should be performed.