Advertisement

Investigation of biomarkers for impending fluid overload in a feline acute haemorrhage-resuscitation model

  • Gareth E. Zeiler
    Correspondence
    Correspondence: Gareth E Zeiler, Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria, Private Bag X04, Onderstepoort, 0110, South Africa.
    Affiliations
    Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria, Onderstepoort, South Africa

    Section of Anaesthesia and Critical Care, Valley Farm Animal Hospital, Pretoria, South Africa

    Brain Function Research Group, School of Physiology, University of the Witwatersrand, Johannesburg, South Africa
    Search for articles by this author
  • Brighton T. Dzikiti
    Affiliations
    Clinical Sciences Department, Ross University School of Veterinary Medicine, Basseterre, St. Kitts and Nevis
    Search for articles by this author
  • Peter Kamerman
    Affiliations
    Brain Function Research Group, School of Physiology, University of the Witwatersrand, Johannesburg, South Africa
    Search for articles by this author
  • Friederike Pohlin
    Affiliations
    Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria, Onderstepoort, South Africa

    Section of Anaesthesia and Critical Care, Valley Farm Animal Hospital, Pretoria, South Africa
    Search for articles by this author
  • Roxanne K. Buck
    Affiliations
    Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria, Onderstepoort, South Africa
    Search for articles by this author
  • Andrea Fuller
    Affiliations
    Brain Function Research Group, School of Physiology, University of the Witwatersrand, Johannesburg, South Africa
    Search for articles by this author
Published:August 20, 2021DOI:https://doi.org/10.1016/j.vaa.2021.04.011

      Abstract

      Objective

      To determine biomarkers for impending fluid overload during intravenous fluid administration in a feline haemorrhage-resuscitation model.

      Study design

      Randomized crossover study.

      Animals

      A group of six domestic cats (mean age and weight: 21 months; 4.9 kg, respectively).

      Methods

      The cats underwent three treatments, 2 months apart. They were anaesthetized and instrumented to measure a range of physiological, blood gas, haematological and biochemical variables over time. Samples were taken during a health check, before haemorrhage, after haemorrhage and then at 30 minute intervals during fluid resuscitation and 24 hours later. The three treatments were: 1) control, sham haemorrhage and resuscitation; 2) lactated Ringer’s solution (LRS); and 3) 6% tetrastarch 130/0.4 (Vol) where the cats underwent a controlled haemorrhage then resuscitation by administering LRS and Vol at 60 and 20 mL kg–1 hour–1, respectively, for 120 minutes. Fluid overload was identified by nasal discharge and radiographic evidence. Biomarkers were variables that exceeded the reference interval for cats during treatment. Potential biomarkers were analysed using receiver operating characteristic curves (p < 0.05).

      Results

      Mean ± standard deviation total blood loss was 10.2 ± 2.3, 29.3 ± 9.0 and 29.1 ± 6.3 mL kg–1 for control, LRS and Vol, respectively. The total volume of LRS and Vol administered was 120 and 40 mL kg–1, respectively. Haematocrit, albumin, magnesium, chloride-to-sodium ratio and sodium-chloride difference were identified as potential biomarkers. These variables exceeded the reference intervals from 30 minutes of resuscitation onwards. A chloride-to-sodium ratio > 0.84 was the most sensitive (90%) and specific (75%) of all potential biomarkers.

      Conclusions and clinical relevance

      Changes in physiological variables, haematocrit and albumin were poor biomarkers of impending fluid overload compared with electrolytes. Finding the ideal biomarker to identify impending fluid overload of commonly used intravenous fluids should improve the safety of their administration in cats.

      Keywords

      Introduction

      Fluid overload occurs in veterinary medicine; however, most of the evidence is anecdotal or based on retrospective studies (
      • Cavanagh A.A.
      • Sullivan L.A.
      • Hansen B.D.
      Retrospective evaluation of fluid overload and relationship to outcome in critically ill dogs.
      ;
      • Thomovsky E.
      • Brooks A.
      • Johnson P.
      Fluid overload in small animal patients.
      ;
      • Ostroski C.J.
      • Drobatz K.J.
      • Reineke E.L.
      Retrospective evaluation of and risk factor analysis for presumed fluid overload in cats with urethral obstruction: 11 cases (2002-2012).
      ). There is no clear definition of fluid overload in human or veterinary medicine, but it occurs when the amount of fluid entering the body exceeds the amount of fluid lost from the body (
      • Thomovsky E.
      • Brooks A.
      • Johnson P.
      Fluid overload in small animal patients.
      ). An increase in body weight of more than 10% from baseline because of fluid administration could result in the clinical manifestation of fluid overload (
      • Thomovsky E.
      • Brooks A.
      • Johnson P.
      Fluid overload in small animal patients.
      ). Fluid overload is associated with organ oedema and effusions, which generally culminate in a poor outcome because of poor microvascular tissue perfusion and oxygenation (
      • Mobley A.
      • Sullivan L.
      Retrospective determination of fluid overload in critically ill dogs.
      ;
      • Malbrain M.L.N.G.
      • van Regenmortel N.
      • Saugel B.
      • et al.
      Principles of fluid management and stewardship in septic shock: it is time to consider the four Ds and the four phases of fluid therapy.
      ). The clinical manifestation of fluid overload occurs after fluid administration; therefore, many fluid therapy plans have been proposed in an attempt to minimize the occurrence of fluid overload.
      In veterinary anaesthesia, static and dynamic biomarkers have been the subject of recent investigation to determine fluid responsiveness in dogs (
      • Duffy A.L.
      • Butler A.L.
      • Radecki S.V.
      • Campbell V.L.
      Comparison of continuous arterial pressure waveform analysis with the lithium dilution technique to monitor cardiac output in conscious dogs with systemic inflammatory response syndrome.
      ;
      • Drozdzynska M.J.
      • Chang Y.
      • Stanzani G.
      • Pelligand L.
      Evaluation of the dynamic predictors of fluid responsiveness in dogs receiving goal-directed fluid therapy.
      ;
      • Celeita-Rodriguez N.
      • Teixeira-Neto F.J.
      • Garofalo N.A.
      • et al.
      Comparison of the diagnostic accuracy of dynamic and static preload indexes to predict fluid responsiveness in mechanically ventilated, isoflurane anesthetized dogs.
      ). If there is a positive response, then fluids can be administered until a goal-directed target is reached (
      • Boyd C.
      • Smart L.
      Hypovolemic shock.
      ;
      • Linklater A.
      Management of hemorrhagic shock.
      ). However, many of these dynamic biomarkers require the animal to be mechanically ventilated which decreases their usefulness in general practice. Furthermore, there are no clear guidelines detailing how much fluid should be administered before a patient is defined as non-fluid responsive (
      • Prittie J.
      Optimal endpoints of resuscitation and early goal-directed therapy.
      ). In human medicine, evaluating these static and dynamic biomarkers as a method of accurately predicting fluid responsiveness has being questioned (
      • Michard F.
      • Teboul J.
      Predicting fluid responsiveness in ICU patients. A critical review of the evidence.
      ;
      • Michard F.
      Volume management using dynamic parameters. The good, the bad, the ugly.
      ;
      • Ueyama H.
      • Kiyonaka S.
      Predicting the need for fluid therapy – does fluid responsiveness work?.
      ). We speculate that fluid non-responders are prone to developing fluid overload. Early detection of impending fluid overload during the perioperative period is difficult because of the similarity in physiological responses during anaesthesia, disease or trauma that mask its subtle manifestation. The overt clinical signs of fluid overload include peripheral oedema, cavity effusions and fluid drainage from surgical sites and respiratory airways (
      • Thomovsky E.
      • Brooks A.
      • Johnson P.
      Fluid overload in small animal patients.
      ).
      To determine if a novel biomarker could be identified, we created iatrogenic fluid overload during resuscitation of severe acute haemorrhage-induced hypovolaemia using a volume ratio fluid therapy plan. A wide range of physiological, blood gas, haematological and biochemical variables were collected before and after haemorrhage and during fluid resuscitation using an isotonic crystalloid and a synthetic colloid solution. We considered a biomarker to be a variable that exceeded the reference interval for cats during resuscitation, which was only evident in animals with overt clinical signs of fluid overload compared with those that did not develop fluid overload. We hypothesized that at least one biomarker could be identified that could indicate impending fluid overload during crystalloid and colloid fluid therapy.

      Materials and methods

       Animals

      A group of six neutered domestic cats [three males and three females; aged 21 ± 1 months; 4.9 ± 1.2 kg (mean ± standard deviation)] were used in this study. This study was part of a larger fluid resuscitation project and only data relevant to the present study are reported. The sample size was based on availability of cats from the research colony and the study budget, as previously described (
      • Zeiler G.E.
      • Fuller A.
      • Rioja E.
      • et al.
      Development of a severity scoring system for acute haemorrhage in anaesthetized domestic cats: the CABSS score.
      ). The cats were housed in a communal indoor–outdoor cattery. The animal ethics committees of the Universities of Pretoria (v006-15) and Witwatersrand (2017-10-68-C-AREC) in South Africa approved the study.
      The cats underwent a health check 1 week before each treatment that included a physical examination and venous blood analyses. Food, but not water, was withheld 8 hours before treatment.

       Study procedures

      Cats were randomly assigned (balanced single block design; www.randomization.com) to receive three treatments each at 2 month intervals. On the morning of treatment, the cat underwent a clinical examination, was weighed and premedicated with buprenorphine hydrochloride (0.02 mg kg–1, Temgesic; Reckitt Benckiser Healthcare, RSA) intramuscularly and left undisturbed for 45 minutes. An indwelling catheter (22 gauge, Jelco; Smith Medical International, UK) was aseptically inserted into one of the cephalic veins and secured. Alfaxalone (Alfaxan-CD RTU; Afrivet, RSA) was administered intravenously and titrated to induce general anaesthesia. The trachea was intubated with a cuffed endotracheal tube (4.0–4.5 mm internal diameter; Teleflex Incorporated, RSA) and connected to a paediatric circle breathing system (Compact paediatric breathing system; Intersurgical, RSA). General anaesthesia was maintained with isoflurane (Isofor; Safeline Pharmaceuticals, RSA), initially set to 2%, in oxygen at a fixed flow rate of 80 mL kg–1 minute–1. A standardized target end-tidal isoflurane concentration (Fe′Iso) for the study was set at 1.7% (
      • Shaughnessy M.R.
      • Hofmeister E.H.
      A systematic review of sevoflurane and isoflurane minimum alveolar concentration in domestic cats.
      ). The ventral neck region was clipped and surgically scrubbed before the cat was transferred to the procedure table. An isotonic crystalloid fluid (Lactated Ringer’s solution, LRS; Fresenius Kabi, RSA) was infused at a maintenance fluid rate of 5 mL kg–1 hour–1, by an electronic device (Infusomat; BBraun, Germany) via a y-ported administration set (Infusomat Space Set; BBraun) connected to the cephalic catheter. The thorax was radiographed by taking two orthogonal views (dorsoventral and right lateral). The cat was then placed in dorsal recumbency, and probes and leads were placed and connected to a multiparameter machine (Datex-Ohmeda Cardiocap 5; GE Healthcare, Finland) to monitor physiological variables as follows: heart rate by electrocardiogram, peripheral oxygen haemoglobin saturation by pulse oximetry, end-tidal carbon dioxide (Pe′CO2) and respiratory rate by capnometry, Fe′Iso by gas analysis, and oesophageal temperature with a thermistor probe. A catheter (22 gauge, 50 mm, Arrow arterial catheterization set; Arrow International, PA, USA) was inserted into the right pre-superficialized (superficialized 15 months earlier during gonadectomy) carotid artery by cut-down technique to allow for invasive systolic (SAP), diastolic (DAP) and mean (MAP) arterial blood pressure monitoring and intermittent blood sampling for gas tension analyses. A catheter (22 gauge, 50 mm, Arrow arterial catheterization set; Arrow International) was inserted percutaneously into the left jugular vein for intermittent blood sampling and facilitation of controlled haemorrhage later. Both catheters were inserted using the Seldinger technique (
      • Seldinger S.I.
      Catheter replacement of the needle in percutaneous arteriography: a new technique.
      ). One of the three randomly allocated treatments then commenced. The treatments were divided into two phases, the haemorrhage phase and resuscitation phase, as follows:
      Control: a waiting period of 15 minutes to simulate controlled haemorrhage followed by sham resuscitation of 120 minutes duration.
      LRS: controlled haemorrhage followed by LRS administration at 60 mL kg–1 hour–1 for 120 minutes during resuscitation.
      6% Tetrastarch 130/0.4 (Vol): controlled haemorrhage followed by Vol (Voluven 130/0.4; Fresenius Kabi) administration at 20 mL kg–1 hour–1 for 120 minutes during resuscitation.
      During the controlled haemorrhage phase, blood was drawn manually into a semi-closed system using 20 mL syringes primed with citrate-phosphate-dextrose (4 mL; JMS blood bag, 450 mL; JMS Singapore, Singapore) via the jugular catheter at a targeted rate of 2 mL kg–1 minute–1 until one of two end points: 1) maximum of 30 mL kg–1 blood drawn; or 2) MAP of < 48 mmHg that persisted for at least 3 minutes.
      During the resuscitation phase, the randomized resuscitation fluid was infused via a second electronic device, where its administration set was connected to the y-port of the maintenance fluid administration set. Both fluids were administered simultaneously.
      Blood samples were obtained, and physiological variables recorded at fixed time points during the haemorrhage and resuscitation phase. After the 120 minute resuscitation phase, blood was sampled; the monitoring equipment and catheters, except the jugular catheter, were removed, and the thorax was radiographed. The carotid artery catheter was removed, and digital pressure was applied until a stable clot was formed. If a stable clot did not form within 15 minutes, then a haemostatic absorbable mesh (BloodStop iX; Life Science Plus, CA, USA) was applied using digital pressure for 10 minutes, or until haemostasis was achieved. Then the cut-down incision site was sutured using a two-layer closure technique with absorbable suture material (MonoPlus 5/0; BBraun). The cat was given a single subcutaneous injection of meloxicam (0.2 mg kg–1, Petcam; CiplaVet, RSA) and intravenous injection of buprenorphine (0.03 mg kg–1). The cat was then transferred to the intensive care unit for recovery and overnight observation without the need for any further interventions or treatments. The next day, a blood sample was taken, and the thorax radiographed before the cat was transferred back to the cattery. All cats were rehomed though an adoption process 1 month after concluding the study.

       Data collection and analysis

      The total blood loss at each time point was cumulative and included the controlled haemorrhage volume (not applicable in control) and all blood samples for profiling. The blood loss in control was defined as mild haemorrhage (< 10 mL kg–1). A severe haemorrhage was defined when more than 22 mL kg–1 (> 40% assumed circulating volume) of blood was lost. The total volumes of fluids that were administered during resuscitation were used for reporting because all cats received maintenance fluids during all treatments and that volume (10 mL kg–1 over 120 minutes) was considered negligible. Data were collected in the form of thoracic radiographs, blood sampling (venous and arterial) and physiological measurements that were taken at various time points throughout the study procedures and various ratios and gradients that were calculated, where applicable (Table 1).
      Table 1Data collection time points, ratios and equations used in this study in which six isoflurane-anaesthetized cats underwent a severe haemorrhage (approximately 29 mL kg−1) event and fluid resuscitation using either lactated Ringer’s solution (120 mL kg−1) or 6% tetrastarch 130/0.4 (40 mL kg−1) to restore the intravascular volume or a sham control treatment. The cats were premedicated with buprenorphine (0.02 mg kg−1) and anaesthesia was induced with alfaxalone (2–4 mg kg−1) administered intravenously.
      Data collection time points
      Time periodPhysiologicalHaematologySerum biochemistryElectrolyteBlood gasRadiographs
      1 week prior to treatment (T –7d)
      Heath checkTPRXXX
      Day of treatment (time period in minutes; T0 is end of haemorrhage and beginning of resuscitation)
      –15 (T –15)FullXXXXX
      –10Full
      –5Full
      –1Full
      0 (T0)Full∗XXXX
      30 (T30)Full∗XXXX
      60 (T60)Full∗XXXX
      90 (T90)Full∗XXXX
      120 (T120)Full∗XXXXX
      Next day after treatment (T24h)
      24 hours laterTPRXXXX
      TPR, temperature-pulse-respiration; Full, complete data set of heart rate, respiratory rate, direct arterial blood pressure, end-tidal carbon dioxide measurement, oesophageal temperature and pulse oximetry; T–15, baseline measurements; –10, –5, –1, time intervals during the haemorrhage phase of data collection; T0, time point after haemorrhage and immediately before starting the resuscitation phase; T30, T60, T90, T120, time intervals during the resuscitation phase; Full∗, physiological parameters were captured at 10 minute intervals from time T0 to T120; X, when data were collected.
      Ratios and equations
      Ratio or equationComments
      VgainVgain = EBV((H0 – Hf)/H0); Gross equation (
      • Gross J.B.
      Estimating allowable blood loss: corrected for dilution.
      )

      Where Vgain is the volume gained; EBV is the estimated initial blood volume prior to the resuscitation phase (55 ml kg–1 minus the cumulative blood loss volume;
      • Groom A.C.
      • Rowlands S.
      • Thomas H.W.
      Some circulatory responses to haemorrhage in the cat: a critical level of blood volume for the onset of hypotension.
      ;
      • Mott J.C.
      The effect of haemorrhage on haemoglobin concentration, blood volume and arterial pressure in kittens and cats.
      ); H0 is the initial haematocrit; and Hf is the final haematocrit. The volume gained was divided by the total volume infused to estimate the percentage of administered fluid remaining within the intravascular compartment which was then plotted on a graph over time.
      Ht/AlbHaematocrit (L L–1) to serum albumin concentration (g L–1) ratio
      Na:KSerum sodium (mmol L–1) to serum potassium (mmol L–1) ratio
      Na–ClSerum sodium (mmol L–1) to serum chloride (mmol L–1) difference
      Cl:NaSerum chloride (mmol L–1) to serum sodium (mmol L–1) ratio
      PaCO2-Pe′CO2Arterial to end-tidal carbon dioxide gradient. Calculated by subtracting the end-tidal carbon dioxide value from the arterial partial pressure of carbon dioxide value, both taken at the same time point.
      PaO2-PaO2Aa gradient, or alveolar to arterial partial pressure of oxygen gradient.

      The partial pressure of alveolar oxygen was calculated using the following formula:

      PAO2 = FiO2(Pbar-PH2O) – (PaCO2/RQ)

      Where PaO2 is the partial pressure of alveolar oxygen; FiO2 is the fraction of inspired oxygen (% value as a decimal; set at 0.96); Pbar is the barometric air pressure (mean barometric pressure 661 mmHg; altitude 1250 meters); PH2O is the saturated vapour pressure of water at 37 °C which was set to 47 mmHg; PaCO2 is the arterial partial pressure of carbon dioxide (mmHg); and RQ is the respiratory quotient set at 0.8.
      CaO2CaO2 = (Hüfner’s constant × Hb × SaO2) + (0.003 × PaO2)

      Where CaO2 is the arterial oxygen content (mL dL–1); Hufner’s constant set at 1.39 mL dL–1 for cats (
      • Herrmann K.
      • Haskins S.
      Determination of P50 for feline haemoglobin.
      ); Hb is the haemoglobin concentration (g dL–1); SaO2 is the arterial haemoglobin oxygen saturation of (% value as a decimal); PaO2 is the arterial partial pressure of oxygen (mmHg).
      CvO2CvO2 = (Hüfner’s constant × Hb × SvO2) + (0.003 × PvO2)

      Where CvO2 is the venous oxygen content (mL dL–1); Hüfner’s constant set at 1.39 mL dL–1 for cats (
      • Herrmann K.
      • Haskins S.
      Determination of P50 for feline haemoglobin.
      ); Hb is the haemoglobin concentration (g dL–1); SvO2 is the venous haemoglobin saturation of haemoglobin (% value as a decimal); PvO2 is the venous partial pressure of oxygen (mmHg).
      OEROER = (CaO2 – CvO2)/CaO2 × 100%

      Where OER is the oxygen extraction ratio; CaO2 is the arterial oxygen content (mL dL–1); CvO2 is the venous oxygen content (mL dL–1).
      PF ratioArterial partial pressure of oxygen (mmHg) is divided by the fraction of inspired oxygen (% value as a decimal) to calculate the PF ratio.
      Thoracic radiographs were taken using a mobile unit (Poskom diagnostic x-ray unit model PXP-40HF; Tube model D-124 Toshiba, RSA) which was set to 75 kV and 2.0 mAs for all exposures. Images were captured on electronic cassettes (24 × 30 cm; Fujifilm IP cassette type CC, RSA) and evaluated by an experienced veterinarian to identify pulmonary oedema or pleural effusion using diagnostic imaging software (easyImage; IMV imaging, RSA).
      Prior to collection of arterial or venous blood samples from the catheters for laboratory analyses, an initial 1.5 mL volume was drawn and discarded. Venous blood samples were drawn from the jugular catheter into a syringe and immediately decanted into various vacuum storage tubes (BD Vacutainer tube; Becton Dickinson and Company, UK) in the following order: citrate (observations reported in a separate study), serum and then ethylenediaminetetraacetic acid (EDTA). The stored blood was submitted immediately to an onsite laboratory for analyses as follows: 1) haematology (EDTA tube); 2) serum biochemistry including albumin, blood urea nitrogen, creatinine, glucose, lactate (serum tube); and 3) electrolytes including sodium, chloride, potassium, phosphorus, total magnesium, ionized calcium (serum tube) using daily-calibrated machines run by experienced veterinary clinical pathologists. For blood gas analyses, arterial and venous blood samples were drawn into heparinized syringes (BD A-line; Becton Dickinson and Company) and sent to the onsite laboratory for analysis within 5 minutes of collection, using a daily calibrated machine (mean barometric pressure 661 mmHg; altitude 1250 metres; interpreted at a fixed temperature of 37 °C).
      The volume of resuscitation fluid gained within the intravascular compartment over time during the resuscitation phase was estimated using the Gross equation (Table 1;
      • Gross J.B.
      Estimating allowable blood loss: corrected for dilution.
      ).
      Data were assessed for equal variance and distribution by examining histograms and evaluating descriptive statistics. The duration of the haemorrhage phase was compared using a one-way analysis of variance. End points of the haemorrhage phase were described using count data. Physiological, haematological, serum biochemical, electrolyte and blood gas data were compared among treatments over time using a general linear mixed model (main effects of treatment and time, main outcome was the interaction treatment × time) and significant findings underwent post-hoc Tukey pairwise comparisons. Model fits were assessed by visually inspecting residual plots. Data of all variables were tabulated among treatments over time alongside the cumulative blood loss and resuscitation fluid volumes (both reported in mL kg–1). Cats were identified as either fluid overloaded (yes) by the presence of any of the following: 1) radiographic lung patterns and pleural effusions, 2) facial oedema, 3) thoracic or pelvic limb oedema, and 4) serous or serosanguinous fluid discharge from the respiratory tract (
      • Thomovsky E.
      • Brooks A.
      • Johnson P.
      Fluid overload in small animal patients.
      ); or not fluid overloaded (no) in the absence of these findings, determined at time point 120 minutes. Significant variables (based on outcomes of described statistical tests) were further evaluated to determine potential biomarkers. The biomarkers were defined as a change of a variable value outside an expected clinical reference interval for a domestic cat with overt signs of fluid overload. Variables that fulfilled the biomarker definition were plotted on receiver operating characteristic curves, area under the curve and the Youden index using the dichotomous outcomes (yes: fluid overloaded; no: not fluid overloaded) to determine optimal cut-off points to guide volume resuscitation. Data were analysed using commercially available software (MiniTab 18.1; Minitab Inc., PA, USA; and MedCalc 19.0.3; MedCalc Software, Belgium) and significance interpreted at p < 0.05. Data are reported as mean ± standard deviation and only p values for the interaction treatment × time are reported.

      Results

      During the haemorrhage phase, the simulated haemorrhage time was 15.2 ± 0.8 minutes and did not differ from the controlled haemorrhage times for LRS (17.5 ± 2.3 minutes) and Vol (18.6 ± 4.4 minutes) (p = 0.137). The total blood loss was 10.2 ± 2.3, 29.3 ± 9.0 and 29.1 ± 6.3 mL kg–1 for control, LRS and Vol, respectively. The total volume of LRS and Vol administered was 120 and 40 mL kg–1, respectively. The maximum blood draw volume end point was applied four times (in two LRS and two Vol treatments), while the MAP end point was applied for the remainder of times (eight times in four LRS and four Vol treatments). Fluid overload was not detected in control but was present in LRS (n = 5) and Vol (n = 3).
      During LRS and Vol treatments, and only during the haemorrhage phase, heart rate increased (p = 0.014) while SAP decreased (p = 0.014). These variables returned to baseline values within 10 minutes of resuscitation (Fig. 1). The other physiological variables (except the DAP and MAP) remained unchanged during all treatments (Table S1).
      Figure 1
      Figure 1Mean values and 95% confidence interval plots of the heart rate (a) and systolic arterial blood pressure (b) recorded over time in six isoflurane-anaesthetized cats undergoing three treatments. The treatments were: 1) control using sham haemorrhage and sham fluid resuscitation (Control), or controlled haemorrhage followed by 2) lactated Ringer’s solution (LRS), or 3) 6% tetrastarch 130/0.4 (Vol) infusions for resuscitation. Measurements were taken before haemorrhage (T –15), after haemorrhage (T0) and then at 10 minute intervals (T10, T20, T30) during fluid resuscitation.
      During the resuscitation phase, the haematocrit decreased (p < 0.001) from 0.17 to 0.15 L L–1 between 30 and 120 minutes during LRS, and from 0.19 to 0.11 L L–1 during Vol administration, but did not change during the control sham resuscitation. Red cell count decreased continuously (p < 0.001) from 30 to 120 minutes during both treatments. The mean corpuscular volume, mean corpuscular haemoglobin concentration, leukogram and thrombogram remained unchanged during all treatments (Table S2). The serum albumin decreased during LRS and Vol in a similar manner from 30 to 120 minutes but did not change in the control (p < 0.001). Serum albumin shifted from 16 to 14 g L–1 for LRS and 17 to 11 g L–1 for Vol from 30 to 120 minutes (Fig. 2; Table S3). Lactate increased from 30 to 120 minutes during LRS treatment only (p = 0.004). The other biochemical variables and the haematocrit-to-serum albumin ratio values remained unchanged during all treatments.
      Figure 2
      Figure 2Mean values and 95% confidence interval plots of the haematocrit (a) and serum albumin concentration (b) measured over time in six isoflurane-anaesthetized cats undergoing three treatments. The treatments were: 1) control using sham haemorrhage and sham fluid resuscitation (Control), or controlled haemorrhage followed by 2) lactated Ringer’s solution (LRS), or 3) 6% tetrastarch 130/0.4 (Vol) infusions for resuscitation. Measurements were taken 1 week before treatments during health checks (T –7d), at baseline (T –15) approximately 15 minutes before the end of haemorrhage, after haemorrhage immediately before fluid resuscitation (T0), during resuscitation at 30 minute intervals (T30, T60, T90, T120) and 24 hours later (T24h). The dashed horizontal lines indicate the lowest recommended haematocrit and albumin concentration considered to indicate severe anaemia and severe hypoproteinaemia, respectively.
      The total magnesium decreased significantly from 30 to 120 minutes for LRS and Vol but not for the control (p < 0.001). Magnesium decreased from 0.58 to 0.45 mmol L–1 for LRS and from 0.69 to 0.59 mmol L–1 for Vol (Table S4). The other electrolytes remained unchanged in all treatments. However, the chloride-to-sodium (Cl:Na) ratio (p < 0.001) increased and sodium-chloride (Na–Cl) difference (p = 0.003) decreased significantly in LRS and Vol. The sodium-to-potassium ratio remained unchanged in all treatments. The electrolytes were the only variables that demonstrated potential for use as impending fluid overload biomarkers, and potentially cut-off values for fluid resuscitation (Table 2; Fig. 3).
      Table 2Receiver operating characteristic (ROC) curves and the Youden index was used to determine optimal cut-off points to guide volume resuscitation of six isoflurane-anaesthetized cats undergoing severe haemorrhage (approximately 29 mL kg–1) and fluid resuscitation using either lactated Ringer’s solution (120 mL kg–1) or 6% tetrastarch 130/0.4 (40 mL kg–1) to restore the intravascular volume. Cats were premedicated with buprenorphine (0.02 mg kg–1) and anaesthesia was induced with alfaxalone (2–4 mg kg–1) administered intravenously.
      VariableROC AUCYouden indexAssociated criterionSensitivitySpecificity+LR–LRPPVNPP
      Cl:Na0.8310.650> 0.8490753.60.138286
      Na–Cl0.7780.556< 21100562.30.069100
      Mg0.9750.889< 0.57891004.00.1110090
      Cl:Na, Chlorine-to-sodium ratio; +LR, positive likelihood ratio value; –LR, negative likelihood ratio value; Mg, magnesium concentration; Na–Cl, sodium-to-chlorine difference; NPP, Negative predictive value; PPV, positive predictive value; ROC AUC, area under the ROC curve.
      Figure 3
      Figure 3Mean values and 95% confidence interval plots of the Cl:Na ratio (a), Na–Cl difference (b) and serum magnesium concentration (c) recorded over time in six isoflurane-anaesthetized cats undergoing three treatments. The treatments were: 1) control using sham haemorrhage and sham fluid resuscitation (Control), or controlled haemorrhage followed by 2) lactated Ringer’s solution (LRS), or 3) 6% tetrastarch 130/0.4 (Vol) infusions for resuscitation. Measurements were taken 1 week before treatments during health checks (T –7d), at baseline (T –15) approximately 15 minutes before the end of haemorrhage, after haemorrhage immediately before fluid resuscitation (T0), during resuscitation at 30 minute intervals (T30, T60, T90, T120) and 24 hours later (T24h). The dashed horizontal lines indicate the reference interval for variable plotted on the graph for a healthy domestic cat.
      Blood gas values, gradients (PaCO2-Pe′CO2; PaO2-PaO2) and ratios (OER; PF ratio) remained unchanged in all treatments (Table S5). Arterial (p < 0.001) and venous (p < 0.001) oxygen content decreased from 30 to 120 minutes in LRS and Vol but not in control. Arterial oxygen content decreased from 9 to 8 mL dL–1 for LRS and from 9 to 5 mL dL–1 for Vol from 30 to 120 minutes.
      The resuscitation fluid volume gained at 120 minutes was approximately 13 mL kg–1 or 11% of the total volume infused for LRS and 16 mL kg–1 or 40% for Vol, respectively (Fig. 4). The total LRS administered was at a 4.1:1 ratio to the total blood loss, whereas Vol was administered at a ratio of 1.4:1.
      Figure 4
      Figure 4Mean values and 95% confidence interval plots of the estimated percentage of resuscitation fluid remaining within the intravascular compartment over time in six isoflurane-anaesthetized cats undergoing three treatments. The treatments were: 1) control using sham haemorrhage and sham fluid resuscitation (Control), or controlled haemorrhage followed by 2) lactated Ringer’s solution (LRS) or 3) 6% tetrastarch 130/0.4 (Vol) infusions for resuscitation. Calculations were made after haemorrhage immediately before fluid resuscitation (T0) and during resuscitation at 30 minute intervals (T30, T60, T90, T120). The dashed horizontal line indicates 0%.

      Discussion

      We detected signs of fluid overload with both the fluids administered to the cats and we identified several biomarkers that could be used as novel indicators of impending fluid overload. The use of biomarkers is a previously undescribed approach to guide fluid therapy and indicate when fluid volume resuscitation is sufficient. The biomarkers indicate that the administration of fluids at commonly recommended shock dose rates (LRS: 60 mL kg–1; Vol: 15 mL kg–1;
      • Rozanski E.
      • Rondeau M.
      Choosing fluids in traumatic hypovolemic shock: the role of crystalloids, colloids, and hypertonic saline.
      ) or ratio doses are excessive and inappropriate in anaesthetized cats with severe haemorrhage. In this study, the Cl:Na ratio and magnesium concentration were good biomarkers, but not heart rate, blood pressure, haematocrit or serum albumin concentration.
      Using our definition of impending fluid overload, in cats that have sustained an acute controlled intraoperative haemorrhage of approximately 24 mL kg–1 we recommend an initial resuscitation dose of 30 or 10 mL kg–1 for LRS or Vol, respectively, over 15–20 minutes. These fluid volumes should prevent iatrogenic fluid overload. These fluid doses are similar to those administered to ‘trauma cats’, which received an isotonic crystalloid (33.8 ± 19.3 mL kg–1) or colloid (9.5 ± 5.3 mL kg–1) during initial resuscitation efforts (
      • Wehausen C.E.
      • Kirby R.
      • Rudloff E.
      Evaluation of the effects of bovine hemaglobin glutamer-200 on systolic arterial blood pressure in hypotensive cats: 44 cases (1997-2008).
      ). Furthermore, the ratio of fluid administration to volume of blood lost was approximately 1.25:1 and 0.4:1 for LRS and Vol, respectively. These ratios are significantly lower than the 3:1 and 1:1 ratios for crystalloid and colloid replacement, respectively, that have been advocated for decades (
      • Moon-Massat P.F.
      Fluid therapy and blood transfusion.
      ;
      • Muir W.W.
      • Hubbell J.A.
      • Bednarski R.M.
      • et al.
      Handbook of veterinary anesthesia.
      ;
      • Clarke K.W.
      • Trim C.M.
      • Hall L.W.
      Veterinary Anaesthesia.
      ). Our interpretation of the biomarkers and recommended initial doses of fluids are similar to those reported by other investigators who have challenged these ratios (
      • Hahn R.G.
      Fluid therapy in uncontrolled haemorrhage – what experimental models have taught us.
      ;
      • Fodor G.H.
      • Habre W.
      • Balogh A.L.
      • et al.
      Optimal crystalloid volume ratio for blood replacement for maintaining hemodynamic stability and lung function: an experimental randomized control study.
      ). A recent rat-based model suggested that a 1:1 ratio should be used for isotonic crystalloid resuscitation; and they reported that using this ratio rats had adequate compensatory capacity with no pulmonary oedema (
      • Fodor G.H.
      • Habre W.
      • Balogh A.L.
      • et al.
      Optimal crystalloid volume ratio for blood replacement for maintaining hemodynamic stability and lung function: an experimental randomized control study.
      ). Also, if isotonic crystalloids are administered immediately after haemorrhage, then only a 1.6:1 ratio is required to restore the intravascular volume (
      • Hahn R.G.
      Fluid therapy in uncontrolled haemorrhage – what experimental models have taught us.
      ). These lower-than-expected dose ratios could be the origin of the sentiment that cats are not fluid tolerant, and that caution is advised during resuscitation. However,
      • Hahn R.G.
      Fluid therapy in uncontrolled haemorrhage – what experimental models have taught us.
      reviewed haemorrhage and resuscitation models using rats and pigs and found that the ratios for crystalloids and colloids are indeed lower than recommended. Therefore, cats are similar to other species in their response to fluid therapy. However, using the liberal fluid therapy that was advocated in the past, cats are erroneously considered to be species that is intolerant to fluid overload. Despite using a goal-directed fluid therapy plan, hypovolaemic cats with urethral obstruction were given a median range of 18.8 (8.4–30.9) mL kg–1 of an isotonic crystalloid fluid and developed fluid overload compared with 15.2 (5.5–42.6) mL kg–1 in those that did not develop fluid overload (
      • Ostroski C.J.
      • Drobatz K.J.
      • Reineke E.L.
      Retrospective evaluation of and risk factor analysis for presumed fluid overload in cats with urethral obstruction: 11 cases (2002-2012).
      ). Therefore, other biomarkers need to be identified that are not derived from guidelines and alternative volumes of fluids should be used during resuscitation.
      Fluid overload is not an easy clinical condition to treat; however, in the healthy cats of our study, by stopping fluid administration and allowing time in a low-stress environment, the overt fluid overload resolved within 24 hours. This finding suggests that healthy cats can tolerate severe haemorrhage and liberal fluid resuscitation without untoward effects. Additionally, there were no overt indications of oxygen debt, increased oxygen extraction or gas diffusion impairment (normal alveolar to arterial oxygen gradient), contrary to our expectations (
      • Thomovsky E.
      • Brooks A.
      • Johnson P.
      Fluid overload in small animal patients.
      ). This was despite the low haematocrit values (less than recommended transfusion triggers:
      • Barfield D.
      • Adamantos S.
      Feline blood transfusions a pinker shade of pale.
      ;
      • Kisielewicz C.
      • Self I.A.
      Canine and feline blood transfusions: controversies and recent advances in administration practices.
      ), decreased oxygen carrying capacity and radiographic evidence of pulmonary oedema. These findings suggest that the oxygen demand did not exceed oxygen delivery and no oxygen debt accumulated, which has been described in haemodynamically stable anaemic human patients (
      • Napolitano L.M.
      • Kurek S.
      • Luchette F.A.
      • et al.
      Clinical practice guideline: red blood cell transfusion in adult trauma and critical care.
      ). In the treatments where controlled haemorrhage occurred, the haematocrit values and serum albumin concentration were very small during the fluid resuscitation but returned to approximately the same value as those immediately after the controlled haemorrhage by the 24 hour period. This observation suggests that the decrease in the haematocrit and serum albumin concentrations resulted from the fluids that were administered and that otherwise healthy cats could excrete the excess fluids within 24 hours. Furthermore, the increase in haematocrit immediately post haemorrhage was probably because blood entered the circulation by contraction of reservoir organs such as the spleen, liver and gastrointestinal tract (
      • Greenway C.V.
      • Lister G.E.
      Capacitance effects and blood reservoir function in the splanchnic vascular bed during non-hypotensive haemorrhage and blood volume expansion in anaesthetized cats.
      ). There was no evidence of erythrocyte regeneration (in the form of reticulocytosis). The decrease in serum albumin concentration immediately post haemorrhage was probably because of the loss of whole blood, with no movement of albumin from the interstitial space into the intravascular compartment, unlike the movement of reserve erythrocytes into the circulating from low flow organs (
      • Greenway C.V.
      • Lister G.E.
      Capacitance effects and blood reservoir function in the splanchnic vascular bed during non-hypotensive haemorrhage and blood volume expansion in anaesthetized cats.
      ).
      When fluids were administered posthaemorrhage, the Cl:Na ratio steadily increased as the volume of resuscitation fluids increased over time and the ratio shifted past the reference interval (0.76–0.83) for cats (
      • Goggs R.
      • Myers M.
      • De Rosa S.
      • et al.
      Chloride: sodium ratio may accurately predict corrected chloride disorders and the presence of unmeasured anions in dogs and cats.
      ). The Cl:Na ratio for LRS was 0.85 (112 mmol L–1:131 mmol L–1) and 1.0 (154 mmol L–1:154 mmol L–1) for Vol. We theorize that the fluids were replacing the lost plasma volume and therefore, over the 120 minute resuscitation period, the electrolyte concentrations were predominantly influenced by the administered fluid rather than homeostatic functions of volume balance. Therefore, we suggest that the Cl:Na ratio could be used as an impending fluid overload biomarker during fluid resuscitation. Most fluid used during resuscitation have a Cl:Na ratio > 0.85; however, few, like Normosol-R (Hospira; Lake Forest, IL, USA), have a ratio of 0.7 (98 mmol L–1:140 mmol L–1) and contain magnesium (3 mmol L–1). We speculate that if Normosol-R were used, then the Cl:Na ratio should decrease to < 0.76 (lower interval limit for cats). We propose that this biomarker be used to predict that sufficient Normosol-R has been administered, at which time fluid therapy should be discontinued to prevent iatrogenic fluid overload, but further research is required to confirm our speculation. The Na–Cl difference also demonstrated a consistent trend; however, the control group also demonstrated differences < 26 mmol L–1. Therefore, we suggest that the electrolyte shifts during general anaesthesia may contribute to this shift and not only the administration of resuscitation fluids. The magnesium concentration decreased over time when resuscitation fluids were administered and not with the control treatment; this effect was probably the result of simple dilution because the resuscitation fluids used in this study did not contain magnesium (
      • Whittaker J.D.
      • Downes F.
      • Becker H.
      • et al.
      Influence of perioperative serum magnesium for cardiac and noncardiac morbidity and mortality following emergency peripheral vascular surgery.
      ). This may be true for all resuscitation fluids that do not contain magnesium. Furthermore, the predictable shifts in the serum electrolyte concentrations, in particular the Na and Cl contractions, could be related to shifts in concentration due to the electrolyte composition of the fluids. Moreover, the shift in electrolyte concentration could be because of the body’s need to maintaining blood pH balance through maintaining electrical neutrality based in Stewart theory (
      • Muir W.
      Effects of intravenously administered crystalloid solutions on acid-base balance in domestic animals.
      ).
      This study had notable limitations. The sample size is small, so only had the power to detect large effects. However, our observations can be used to estimate samples sizes to validate our findings under various clinical conditions. The study was conducted in anaesthetized cats under controlled conditions. Therefore, extrapolation of our findings in conscious animals, those that have experienced trauma or those administered different drugs during anaesthesia will require further investigation. The Gross equation was used to estimate the volume gain from the fluids; however, this was a crude estimation because the equation is not a gold standard technique and is prone to errors when there is fluctuation between volume loss (haemorrhage) and volume gain (fluids). Despite estimated volume gains being similar to those published (
      • Rudloff E.
      • Kirby R.
      Colloid and crystalloid resuscitation.
      ), other more accurate techniques could have been used, such as using dye-labelled or radioactively-labelled erythrocytes or albumin (
      • Ertl A.C.
      • Diedrich A.
      • Raj S.R.
      Techniques used for the determination of blood volume.
      ).

      Conclusions

      Biomarkers for predicting impending fluid overload during haemorrhage-induced hypovolaemic resuscitation in anaesthetized cats were identified. Overall, physiological variables, haematocrit or serum albumin concentration were poor biomarkers compared with Cl:Na ratio and magnesium concentration. The Cl:Na ratio (> 0.84) might be a useful biomarker compared with magnesium concentration (< 0.57 mmol L–1) because they are more readily measured in clinical practice.

      Acknowledgements

      The authors would like to thank the people at the UPBRC who assisted in the PhD research project and cared for the cats during the study. Thank you to Eva Rioja for reviewing the manuscript before submission. We would also like to thank the following for their financial contributions: South African Veterinary Foundation , University of Pretoria Research Development Program , South African National Research Foundation .

      Supplementary data

      The following are the Supplementary data to this article:

      Authors’ contributions

      GEZ: study design, data collection and data analysis. BTD and AF: study design. RKB and FP: data collection. PK: data analysis. All authors contributed with manuscript editing.

      Conflict of interest statement

      The authors declare no conflict of interest.

      References

        • Barfield D.
        • Adamantos S.
        Feline blood transfusions a pinker shade of pale.
        J Feline Med Surg. 2011; 13: 11-23
        • Boyd C.
        • Smart L.
        Hypovolemic shock.
        in: Drobatz K.J. Hopper K. Rozanski E. Silverstein D.C. Textbook of small animal emergency medicine. Wiley Blackwell, USA2019: 986-999
        • Cavanagh A.A.
        • Sullivan L.A.
        • Hansen B.D.
        Retrospective evaluation of fluid overload and relationship to outcome in critically ill dogs.
        J Emerg and Crit Care. 2016; 26: 578-586
        • Celeita-Rodriguez N.
        • Teixeira-Neto F.J.
        • Garofalo N.A.
        • et al.
        Comparison of the diagnostic accuracy of dynamic and static preload indexes to predict fluid responsiveness in mechanically ventilated, isoflurane anesthetized dogs.
        Vet Anaes Analg. 2019; 46: 276-288
        • Clarke K.W.
        • Trim C.M.
        • Hall L.W.
        Veterinary Anaesthesia.
        11th edn. Saunders Elsevier, UK2014: 456
        • Drozdzynska M.J.
        • Chang Y.
        • Stanzani G.
        • Pelligand L.
        Evaluation of the dynamic predictors of fluid responsiveness in dogs receiving goal-directed fluid therapy.
        Vet Anaesth Analg. 2018; 45: 22-30
        • Duffy A.L.
        • Butler A.L.
        • Radecki S.V.
        • Campbell V.L.
        Comparison of continuous arterial pressure waveform analysis with the lithium dilution technique to monitor cardiac output in conscious dogs with systemic inflammatory response syndrome.
        Am J Vet Res. 2009; 70: 1365-1373
        • Ertl A.C.
        • Diedrich A.
        • Raj S.R.
        Techniques used for the determination of blood volume.
        Am J Med Sci. 2007; 334: 32-36
        • Fodor G.H.
        • Habre W.
        • Balogh A.L.
        • et al.
        Optimal crystalloid volume ratio for blood replacement for maintaining hemodynamic stability and lung function: an experimental randomized control study.
        BMC Anesth. 2019; 19: 1-9
        • Goggs R.
        • Myers M.
        • De Rosa S.
        • et al.
        Chloride: sodium ratio may accurately predict corrected chloride disorders and the presence of unmeasured anions in dogs and cats.
        Front Vet Sci. 2017; 4: 1-7
        • Greenway C.V.
        • Lister G.E.
        Capacitance effects and blood reservoir function in the splanchnic vascular bed during non-hypotensive haemorrhage and blood volume expansion in anaesthetized cats.
        J Physiol. 1974; 237: 279-294
        • Groom A.C.
        • Rowlands S.
        • Thomas H.W.
        Some circulatory responses to haemorrhage in the cat: a critical level of blood volume for the onset of hypotension.
        J Physiol. 1965; 172: 385-405
        • Gross J.B.
        Estimating allowable blood loss: corrected for dilution.
        Anaesth. 1983; 58: 277-280
        • Hahn R.G.
        Fluid therapy in uncontrolled haemorrhage – what experimental models have taught us.
        Acta Anaesth Scand. 2013; 57: 16-28
        • Herrmann K.
        • Haskins S.
        Determination of P50 for feline haemoglobin.
        J Vet Emerg Crit Care. 2005; 15: 26-31
        • Kisielewicz C.
        • Self I.A.
        Canine and feline blood transfusions: controversies and recent advances in administration practices.
        Vet Anaesth Analg. 2014; 41: 233-242
        • Linklater A.
        Management of hemorrhagic shock.
        in: Drobatz K.J. Hopper K. Rozanski E. Silverstein D.C. Textbook of small animal emergency medicine. Wiley Blackwell, USA2019: 1115-1125
        • Malbrain M.L.N.G.
        • van Regenmortel N.
        • Saugel B.
        • et al.
        Principles of fluid management and stewardship in septic shock: it is time to consider the four Ds and the four phases of fluid therapy.
        Ann Intensive Care. 2018; 8: 1-16
        • Michard F.
        Volume management using dynamic parameters. The good, the bad, the ugly.
        Chest. 2005; 128: 1902-1904
        • Michard F.
        • Teboul J.
        Predicting fluid responsiveness in ICU patients. A critical review of the evidence.
        Chest. 2002; 121: 2000-2008
        • Mobley A.
        • Sullivan L.
        Retrospective determination of fluid overload in critically ill dogs.
        J Vet Emerg Crit Care. 2012; 22: S8-S18
        • Moon-Massat P.F.
        Fluid therapy and blood transfusion.
        in: Seymore C. Duke-Novakovski T. BSAVA Manual of canine and feline anaesthesia and analgesia. 2nd edn. BSAVA, UK2007: 172
        • Mott J.C.
        The effect of haemorrhage on haemoglobin concentration, blood volume and arterial pressure in kittens and cats.
        J Physiol. 1968; 194: 659-667
        • Muir W.
        Effects of intravenously administered crystalloid solutions on acid-base balance in domestic animals.
        J Vet Intern Med. 2017; 31: 1371-1381
        • Muir W.W.
        • Hubbell J.A.
        • Bednarski R.M.
        • et al.
        Handbook of veterinary anesthesia.
        5th edn. Elsevier, USA2013: 313
        • Napolitano L.M.
        • Kurek S.
        • Luchette F.A.
        • et al.
        Clinical practice guideline: red blood cell transfusion in adult trauma and critical care.
        J Trauma. 2009; 67: 1439-1442
        • Ostroski C.J.
        • Drobatz K.J.
        • Reineke E.L.
        Retrospective evaluation of and risk factor analysis for presumed fluid overload in cats with urethral obstruction: 11 cases (2002-2012).
        J Vet Emerg Crit Care. 2017; 27: 561-568
        • Prittie J.
        Optimal endpoints of resuscitation and early goal-directed therapy.
        J Vet Emerg Crit Care. 2006; 16: 329-339
        • Rozanski E.
        • Rondeau M.
        Choosing fluids in traumatic hypovolemic shock: the role of crystalloids, colloids, and hypertonic saline.
        J Am Anim Hos Assoc. 2002; 38: 499-501
        • Rudloff E.
        • Kirby R.
        Colloid and crystalloid resuscitation.
        Vet Clin N Am: Small Anim Prac. 2001; 31: 1207-1229
        • Seldinger S.I.
        Catheter replacement of the needle in percutaneous arteriography: a new technique.
        Acta Radiol. 1953; 39: 368-376
        • Shaughnessy M.R.
        • Hofmeister E.H.
        A systematic review of sevoflurane and isoflurane minimum alveolar concentration in domestic cats.
        Vet Anaesth Analg. 2014; 41: 1-13
        • Thomovsky E.
        • Brooks A.
        • Johnson P.
        Fluid overload in small animal patients.
        Topics Compan Anim Med. 2016; 31: 94-99
        • Ueyama H.
        • Kiyonaka S.
        Predicting the need for fluid therapy – does fluid responsiveness work?.
        J Intensive Care. 2017; 5: 1-6
        • Wehausen C.E.
        • Kirby R.
        • Rudloff E.
        Evaluation of the effects of bovine hemaglobin glutamer-200 on systolic arterial blood pressure in hypotensive cats: 44 cases (1997-2008).
        J Am Vet Med Assoc. 2011; 238: 909-914
        • Whittaker J.D.
        • Downes F.
        • Becker H.
        • et al.
        Influence of perioperative serum magnesium for cardiac and noncardiac morbidity and mortality following emergency peripheral vascular surgery.
        J Cardiothor Vasc Anaesth. 2019; 33: 474-479
        • Zeiler G.E.
        • Fuller A.
        • Rioja E.
        • et al.
        Development of a severity scoring system for acute haemorrhage in anaesthetized domestic cats: the CABSS score.
        Vet Anaes Analg. 2020; 47: 499-508