Evaluation of an oscillometric blood pressure monitor for use in anesthetized sheep


  • Cynthia M Trim,

    Corresponding author
    1. Department of Large Animal Medicine, College of Veterinary Medicine, University of Georgia, Athens, GA, USA
    • Correspondence: Cynthia M Trim, Department of Large Animal Medicine, College of Veterinary Medicine, University of Georgia, Athens, GA 30602, USA. E-mail: ctrim@uga.edu

    Search for more papers by this author
  • Erik H Hofmeister,

    1. Department of Small Animal Medicine and Surgery, College of Veterinary Medicine, University of Georgia, Athens, GA, USA
    Search for more papers by this author
  • John F Peroni,

    1. Department of Large Animal Medicine, College of Veterinary Medicine, University of Georgia, Athens, GA, USA
    Search for more papers by this author
  • Merrilee Thoresen

    1. Department of Large Animal Medicine, College of Veterinary Medicine, University of Georgia, Athens, GA, USA
    Search for more papers by this author



To determine the accuracy of an oscillometric blood pressure monitor in anesthetized sheep.

Study design

Prospective study.


Twenty healthy adult sheep, 11 males and nine females, weighing 63.6 ± 8.6 kg.


After premedication with buprenorphine or transdermal fentanyl, anesthesia was induced with ketamine-midazolam and maintained with isoflurane and ketamine, 1.2 mg kg−1 hour−1, ± lidocaine, 3 mg kg−1 hour−1. Invasive blood pressure measurements were obtained from an auricular arterial catheter and noninvasive measurements were from a cuff on the metatarsus or antebrachium. Simultaneous invasive and noninvasive measurements were recorded over a range (55–111 mmHg) of mean arterial pressures (MAP). Isoflurane concentration was increased to decrease MAP and decreasing the isoflurane concentration and infusing dobutamine achieved higher pressures. Invasive and noninvasive measurements were compared.


Correlation (R2) was good between the two methods of measurement (average of three consecutive readings) for systolic (SAP) (0.87), diastolic (DAP) (0.86), and mean (0.90) arterial pressures (p < 0.001). Bias ± SD between noninvasive and invasive measurements for SAP was 3 ± 8 mmHg, for DAP was −10 ± 7 mmHg, and MAP was −7 ± 6 mmHg. There was no significant difference between the average of three measurements and use of the first measurement. Correlations using the first measurement were SAP (0.82), DAP (0.84), and MAP (0.89). Bias ± SD for SAP was 3 ±10 mmHg, for DAP was −11 ± 7 mmHg, and MAP was −7 ± 6 mmHg. The oscillometric monitor slightly overestimated SAP and underestimated DAP and MAP for both average values and the first reading.

Conclusions and clinical relevance

This oscillometric model provided MAP measurements that were acceptable by ACVIM standards. MAP measurements with this monitor were lower than those found with the invasive technique so a clinical diagnosis of hypotension may be made in sheep that are not hypotensive.


Measurement of arterial blood pressure is an important part of monitoring the anesthetized animal and is strongly recommended, particularly during inhalation anesthesia, by the American College of Veterinary Anesthesiologists (ACVA 1995) and the Association of Veterinary Anaesthetists (www.ava.eu.com). Low blood pressure may progress to cardiac arrest during anesthesia or contribute to organ malfunction following anesthesia. Measurement of blood pressure via a catheter in a peripheral artery is optimal but not always justified for short duration anesthesia or feasible due to lack of appropriate equipment, difficulty in catheterization or financial concerns.

Noninvasive automated sphygmomanometry is commonly used in human and veterinary medicine. An oscillometric monitor measures and records the amplitude of pressure changes in the cuff caused by pulses in underlying arteries. As the pressure in an inflated cuff decreases, faint oscillations are detected at the systolic arterial pressure (SAP). The amplitude of the pulsatile oscillations increases to a maximum value corresponding with mean arterial pressure (MAP) and then decreases to diastolic arterial pressure (DAP) when the oscillations level off. The monitor uses fixed or variable parameter identification points based on the MAP value to electronically calculate systolic and diastolic pressures. The algorithm determining systolic and diastolic pressures varies between monitors and thus measurement results are not the same for all monitors. Factors such as changes in pulse pressure and shape of the arterial pulse introduce errors in measurement. The accuracy of oscillometric monitors varies not only according to the manufacturer and model of the monitor but also with species of animal, operating conditions such as conscious or anesthetized, and other factors causing abnormal or changes in pressures. Investigations of noninvasive blood pressure (NIBP) monitors in dogs have revealed conflicting results concerning the ideal site for cuff placement, for example, in one study measurement on the thoracic limb achieved greater accuracy than the pelvic limb (McMurphy et al. 2006), no difference was obtained between measurements on the thoracic or pelvic limb (Sawyer et al. 1991), or best correlation between invasive and noninvasive blood pressure measurements was obtained at the metatarsal or coccygeal sites (Haberman et al. 2006). Several different monitors have overestimated low blood pressure in conscious (Bosiack et al. 2010) and anesthetized dogs (Sawyer et al. 1991; Shih et al. 2010), provided acceptable measurement of MAP (Meurs et al. 1996; Deflandre & Hellebrekers 2007; MacFarlane et al. 2010), overestimated SAP and underestimated DAP and MAP (Sawyer et al. 2004), or underestimated blood pressures (Haberman et al. 2006; Wernick et al. 2010). The range of the limits of agreement between the NIBP measurements and invasive values is wide for some monitors, decreasing the reliability of the noninvasive measurements for decisions in clinical management (MacFarlane et al. 2010). Both satisfactory accuracy (Caulkett et al. 1998; Pedersen et al. 2002) and unsatisfactory results (Branson et al. 1997) have been identified with different models in cats. An evaluation of two NIBP monitors in anesthetized foals of <7 days of age using cuff sizes recommended by the manufacturers revealed that both monitors had similar performance (Giguère et al. 2005). In this study accuracy of MAP was acceptable with the cuffs on the tail or metatarsus for conditions of normotension and hypotension, although the Cardell 9402 was most accurate over the coccygeal artery and the Dinamap Pro 100 over the metatarsal artery.

The varied results obtained with different monitors in different species indicate that an NIBP monitor must be validated for a specific species and set of circumstances. The acceptable accuracy of blood pressure monitors has been addressed by various organizations primarily interested in the diagnosis of hypertension in conscious subjects. The American College of Veterinary Internal Medicine (ACVIM) has presented a Consensus Statement that provides a protocol and limits for validating devices for dogs or cats with measurements compared against intra-arterial pressure measurements, with the proviso that each device must be separately validated for a different set of circumstances (Brown et al. 2007). Validation of NIBP devices for use on human subjects follows guidelines of the Association for the Advancement of Medical Instrumentation (AAMI), the British Hypertension Society (BHS), and the European Society of Hypertension (ESH), and original protocols have undergone revision in recent years with the result that accuracy criteria have been tightened to parallel improvements in technology (O'Brien et al. 2010).

The rationale for this study was that blood pressure monitoring is particularly important during inhalation anesthesia because a surgical plane of anesthesia can be associated with hypotension. Consequently, the objective of this study was to evaluate the bias and precision of an oscillometric method of blood pressure measurement when compared with auricular arterial pressure in healthy sheep anesthetized with isoflurane. The hypothesis was that the oscillometric device would meet established criteria for accuracy in blood pressure monitoring by comparison with an invasive blood pressure measurement technique during normotension and hypotension (MAP ≤ 60 mmHg).

Materials and methods

Blood pressure measurements were obtained from 20 anesthetized sheep. They were 11 Suffolk and nine Western Range sheep, 11 wethers and nine females, weighing 63.6 ± 8.6 kg (range, 50–80 kg). Twelve sheep were between 1 and 2 years of age and eight sheep were 4–5 years. Anesthesia was administered as part of a research project involving surgical creation of metacarpal, metatarsal, and tibial bone defects to study bone healing. Approval for the project was obtained from the University Institutional Animal Care and Use Committee.

Food was withheld for 20–24 hours and water for 8–12 hours before anesthesia. On the day of anesthesia, all sheep received flunixin meglumine (1.1 mg kg−1, Banamine; Shering-Plough Animal Health, NJ, USA) SC and ceftiofur sodium (2.2 mg kg−1, Naxcel; Pfizer Animal Health, NY, USA) SC or IM. Preanesthetic medication in 16 sheep was buprenorphine (0.01 mg kg−1, Buprenorphine HCl Injectable; Hospira Inc., IL, USA) IM 45 minutes before induction of anesthesia. The remaining four sheep were premedicated with fentanyl patches (2 μg kg−1 hour−1; 25 μg hour−1, Apotex, FL, USA and 50 μg hour−1, Actavis South Atlantic LLC, FL, USA) applied to the lateral surface of an antebrachium 20–24 hours before anesthesia. Anesthesia was induced in all animals with ketamine (5 mg kg−1, Ketaset; Fort Dodge Animal Health, IA, USA) and midazolam (0.25 mg kg−1, Midazolam HCl Injectable; Abraxis, IL, USA) mixed in the same syringe and administered through a 16-gauge 83 mm (3.25 inch) catheter (BD Angiocath; Becton Dickinson Infusion Therapy Systems Inc., UT, USA) in a jugular vein. After tracheal intubation, the animal was placed in right or left lateral recumbency and anesthesia maintained with isoflurane (IsoFlo; Abbott Laboratories, IL, USA) in oxygen via a small animal circle circuit (Quantiflex; Matrx Medical Inc., NY). Ventilation was controlled at 8–9 breaths minute−1 (SurgiVet SAV 2500 V72500; Smiths Medical PM Inc., WI, USA) and tidal volume was adjusted to maintain end-tidal CO2 between 4 and 5.45 kPa (30 and 41 mmHg). The sheep received a continuous infusion of ketamine, 1.2 mg kg−1 hour−1 (diluted in saline to deliver 10 mL hour−1, Medfusion 2001; Medex Inc., CA, USA) IV. The sheep with the fentanyl patches also received an injection of lidocaine over several minutes (1 mg kg−1, Lidocaine HCl Injection; Hospira Inc.) followed by a continuous infusion of lidocaine (3 mg kg−1 hour−1; Medfusion 2001). Lidocaine (1 mg kg−1) and bupivacaine (0.5 mg kg−1, Sensorcaine; APP Pharmaceuticals LC, IL) were infiltrated subcutaneously and circumferentially just distal to the carpus on each thoracic limb designated for metacarpal surgery as part of a concurrent research project. Lactated Ringer's solution (Hospira Inc.) was administered IV at 10 mL kg−1 hour−1. Sheep that had received buprenorphine for premedication were injected with an additional dose 4 hours after the first administration and again at 12 hours. Flunixin meglumine was repeated the following morning.

Invasive and noninvasive blood pressures, electrocardiogram (ECG), end-tidal CO2, and temperature were measured using a multi-parameter monitor (SurgiVet Advisor Vital Signs monitor V9203; Smiths Medical PM Inc., WI, USA). SAP, DAP and MAP were measured invasively using a 22-gauge 0.6 ID × 25 mm catheter (Surflo; Terumo Medical Corporation, NJ, USA) in an auricular artery and displayed on the monitor (arterial extension 120 cm monitoring line; Arrow International Inc., PA, USA; TruWave Pressure Transducer; Edwards Lifesciences LLC, CA, USA). The accuracy of electronic calibration over 0–180 mmHg was verified before anesthesia using a mercury manometer. The damping coefficient was not determined. Pulse rates were obtained from the arterial pressure waveform. Standard ECG leads were attached and monitored on Lead II. Noninvasive oscillometric blood pressure was measured by application of the ‘purple’ cuffs supplied for use with the monitor (Technicuff, Smiths Medical PM Inc,) around the metatarsus of the upper pelvic limb or above the carpus on the upper thoracic limb. The wool was clipped from the pelvic limb but not the thoracic limb prior to application of the cuff. The circumference of each limb was measured. A medium size cuff recommended for limb circumferences 5–15 cm was attached to the pelvic limb and a large size cuff recommended for limb circumferences 9–25 cm was attached to the thoracic limb. The bladder inside the cuff extends the entire length and the cuff completely encircles the limb, thus there is no requirement to align part of the cuff with the artery. The position of the leg was adjusted so that the horizontal midpoint of the cuff was level with the manubrium. Values for NIBP were digitally displayed. The CO2 analyzer was calibrated before anesthesia (CO2 Calibration Gas, Smiths Medical PM Inc.). Inspired and end-tidal isoflurane concentrations (E'Iso) were measured (POET IQ, Criticare Systems Inc., WI, USA) and monitor accuracy was verified (Anesthesia Calibration Gas; Airgas Specialty Gases, KS, USA).

A minimum of 30 minutes elapsed after induction of anesthesia before the first data collection. Measurements were not obtained from both sites in all animals as access was governed by the surgical procedures performed. Measurements were taken either before or after, but not during, the pelvic limb procedures. The sequence of low and high normal blood pressures was randomized. The vaporizer setting was initially increased to 3 or 3.5% and then decreased to a value that maintained a constant E'Iso associated with an MAP ≤ 65 mmHg. High normal blood pressure was achieved by decreasing the administration of isoflurane until the depth of anesthesia appeared to be light, as judged by return of the palpebral reflex and rotation of the eye, and by providing an IV infusion of dobutamine (diluted in 500 mL saline to 100 μg mL−1, Dobutamine; Hospira Inc.). A minimum of 15 minutes was allowed for a change in the blood pressure and a constant E'Iso was maintained for 5 minutes or more before recording pressures. The dobutamine infusion rate was kept constant during the recording period. At each time point, three simultaneous measurements of invasive and NIBP were recorded at approximately 30 second intervals with a 5 minute interval between time points. Data were recorded by hand and transferred to Excel spread sheets. Data are expressed as mean ± SD.

Statistical analysis

The relationship between invasive and oscillometric noninvasive MAP for measurements taken on the thoracic limb was compared with the pelvic limb by comparing linear regression slopes. There was no significant difference, so data from thoracic and pelvic limbs were pooled for all further analyses. All tests were performed on the data in two ways: by averaging the three measurements at a time point and by taking only the first measurement. Linear regression was used to compare invasive with oscillometric pressures. Agreement was calculated using a Bland-Altman plot of the differences in measurements (oscillometric-invasive) against the average of each pair of measurements. Cohens' kappa was used to calculate agreement for the ability to diagnose hypotension, defined as an MAP ≤ 60 mmHg, by oscillometric NIBP compared with the invasive method. The percentages of oscillometric readings that were within 5, 10, 15, and 20 mmHg of the invasive readings were calculated. Significance was set at p < 0.05. Analyses were performed using statistical software (Graphpad Prism v5.02; Graphpad Software Inc., CA, USA).


The sheep were anesthetized for 1.5 ± 0.2 hours and received 8.9 ± 2.1 mL kg−1 hour−1 of lactated Ringer's solution. Hypotension was induced with E'Iso 1.6 ± 0.2% and higher blood pressures were achieved at E'Iso 0.96 ± 0.25% and a dobutamine infusion of 0.46 ± 0.12 μg kg−1 minute−1. The ranges of pressures that were tested were SAP 74–133 mmHg, DAP 42–105 mmHg, and MAP 55–111 mmHg. Hypotension was defined in this study as an MAP ≤ 60 mmHg and the invasive pressure values that were associated with hypotension were SAP 80 ± 3 mmHg, DAP 46 ± 2 mmHg, MAP 58 ± 2 mmHg, and heart rates of 85 ± 17 beats minute−1. Corresponding NIBP values were SAP 82 ± 7 mmHg, DAP 37 ± 7 mmHg, and MAP 52 ± 6 mmHg. A single premature ventricular depolarization not associated with the paired measurements was observed in one sheep. Recovery from anesthesia was without complications.

The mean circumference of the metatarsus was 10.9 cm (range, 10.5–11.5 cm) and of the antebrachium was 15.8 cm (range, 14–17.5 cm). The medium cuff width was 6.4 cm and the large cuff width was 9.25 cm resulting in a range of cuff width:limb circumference of 56–61% for the pelvic limb and 53–66% for the thoracic limb.

Paired invasive and NIBP measurements were taken from both thoracic limb and pelvic limb of eight sheep, from the pelvic limb only in nine sheep, and the thoracic limb only in three sheep. Ninety-one pairs of measurements were recorded, 56 from the pelvic limb and 35 from the thoracic limb. Each pair of measurements consisted of three measurements except that one time point in two sheep had only two readings. In one sheep this was because the oscillometric method would not record a number and timed out (device failure) and in the other sheep blood pressure was altered by the onset of surgical stimulus during measurement of the third pair of measurements. The average values for these two time points were generated from the two readings obtained. The NIBP monitor was slow to reach a pressure measurement in three sheep, one of which included the device failure just mentioned, and in each case this occurred with the cuff on the pelvic limb.

The correlation was good between invasive and NIBP using the average values of the three measurements at each time point, with R2 values of 0.87, 0.86, and 0.90 for SAP, DAP, and MAP, respectively (all p < 0.0001). The average bias for SAP was 3 ± 8 mmHg (range, −12–27 mmHg) [95% CI: 0.8–4], for DAP was −10 ± 7 mmHg (range, −25–10) [95% CI: −12 to −9], and for MAP was −7 ± 6 mmHg (range, −19–8) [95% CI: −8 to −6]. In general, the oscillometric method slightly overestimated SAP and underestimated DAP and MAP (Fig. 1). Cohen's kappa was 0.55 indicating only moderate agreement for the diagnosis of hypotension, with the majority (83%) of disagreements due to the oscillometric method indicating hypotension when the invasive method indicated none. The percentage of values within 5, 10, 15, and 20 mmHg are presented in Table 1.

Table 1. Percentage of time points where the averaged values obtained by three oscillometric measurements were within 5, 10, 15, or 20 mmHg of the invasive measurements
Variable5 mmHg, %10 mmHg, %15 mmHg, %20 mmHg, %
Systolic arterial pressure527992 97
Diastolic arterial pressure204877 92
Mean arterial pressure366993100
Figure 1.

Bland-Altman plot of difference (oscillometric-invasive) against the average of both invasive and oscillometric values. Solid lines indicate the mean difference and dashed lines indicate the limits of agreement (±2SD). Each dot represents one pair of direct and indirect readings (n = 91).

There was no statistically significant difference between using the first of three measurements compared with the average value of three successive measurements. Using the first values obtained, the correlation was good between oscillometric and invasive methods of measurement, with R2 values of 0.82, 0.84, and 0.89 for systolic, diastolic, and mean arterial pressures, respectively (all p < 0.0001). The average bias for SAP was 3 ± 10 mmHg (range, −25–30 mmHg) [95% CI: 0.7–5], for DAP was −11 ± 7 mmHg (range, −30–11) [95% CI: −13 to −10], and for MAP was −7 ± 6 mmHg (range, −21–6) [95% CI: −9 to −6]. In general, using the oscillometric method and a single measurement slightly overestimated SAP and underestimated DAP and MAP (Fig. 2). Cohen's kappa for hypotension was 0.54, with the majority (89%) of disagreements in the diagnosis of hypotension due to the oscillometric method indicating hypotension when invasive measurement indicated none.

Figure 2.

Bland-Altman plot of difference (oscillometric-invasive) against the first invasive and oscillometric values recorded. Solid lines indicate the mean difference and dashed lines indicate the limits of agreement (± 2SD). Each dot represents one pair of direct and indirect readings (n = 91).


The ACVIM protocol for validating NIBP monitors requires averaging of 3–8 successive measurements and this has been done in several investigations (Sawyer et al. 2004; Bosiack et al. 2010; Wernick et al. 2010). The ACVIM Consensus statement allows validation of an NIBP device when the mean difference of paired measurements for SAP and DAP treated separately is ±10 mmHg or less, with a standard deviation of 15 mmHg or less, that the correlation between paired measures is ≥0.9 across the range of measured values, and that 50% of all measurements for SAP and DAP treated separately lie within 10 mmHg, and 80% lie within 20 mmHg, of the reference method (Brown et al. 2007). No fewer than eight animals must be studied. Using these criteria the device tested in this report meets validation criteria for MAP when using the average of three consecutive measurements, and SAP meets all criteria except the correlation that was 0.87.

Bland & Altman (2007) recommend that averaging replicate pairs of measurements is appropriate only when the usual clinical measurement is the average of that number of observations. In clinical practice it is common to record only one set of values at each monitoring time point, unless an acute change in patient status has occurred. Therefore, evaluation of the oscillometric monitor by the first recorded reading is most appropriate. Using the first measurement, bias for MAP was unchanged but the correlation was decreased to 0.89. Average bias between SAP measured by the oscillometric NIBP device and SAP measured invasively in anesthetized sheep using the first measurement was 3 ± 10 mmHg. However, the NIBP measurements of DAP and MAP were consistently lower than invasive measurements, an inaccuracy that could lead to diagnosis of hypotension with subsequent treatment when the sheep was not hypotensive.

Differences between the guidelines of the AAMI, BHS, and ESH for NIBP monitors that might impact on validation of devices for use in veterinary medicine are that these protocols are directed at conscious human subjects with normotension and hypertension and require a blood pressure testing range of SAP ≤ 100 to ≥170 mmHg and DAP ≤ 50 to ≥120 mmHg with 22–44 measurements from a minimum of 10–12 subjects in each pressure range (O'Brien et al. 2010). The protocols require that the device be tested against reference pressures obtained using a mercury noninvasive sphygmomanometer and a stethoscope for auscultating Korotkov sounds, and measurements are no longer averaged. Thus the protocol requires comparison of two NIBP techniques at the same site of the body. AAMI accuracy criteria for comparison of auscultatory and oscillometric NIBP measurements are 5 ± 8 mmHg and an ESH-International Protocol (ESH-IP) Grade A pass requires 65% of measurements to be ≤5 mmHg, 81% to be ≤ 10 mmHg, and 93% to be ≤15 mmHg different from the standard measurement. The measurements obtained from this study did not meet these current validation requirements.

The problem with attempting to validate veterinary devices with the ESH-IP is that the reference standard for NIBP device validation in veterinary medicine is obtained from invasive pressure monitoring. Auscultatory NIBP measurements in humans differ by 1–7 mmHg for SAP and 8–18 mmHg for DAP when compared with invasive pressures (Lake 1990). Furthermore, not all invasive measurement sites in animals yield the same numbers (McMurphy et al. 2006). Peripheral arterial pressures differ from aortic pressures as the systolic pressure increases and diastolic pressure decreases as the distance from the heart increases (Guyton 1986). Many papers published comparing invasive and NIBP measurements have used a peripheral artery for comparison (Sawyer et al. 2004; McMurphy et al. 2006; Deflandre & Hellebrekers 2007; Bosiack et al. 2010; MacFarlane et al. 2010; Shih et al. 2010) while others have used femoral or aortic pressure (Pedersen et al. 2002; Haberman et al. 2006).

One possible explanation for the discrepancies between invasive and NIBP measurements could be that the invasive physiologic system was overdamped leading to measurement of lower values for SAP and higher values for DAP. Noncompliant tubing was used and the arterial catheter was frequently flushed, however, speculation would have been minimized if the damping coefficient of the system had been verified as optimal in accordance with the transducer manufacturer's instructions.

Systolic and pulse pressure variations ‘cycling’ may be observed on an arterial pressure waveform induced by changes in intrathoracic pressure resulting from controlled ventilation. The increase in thoracic pressure during inspiration causes a significant decrease in stroke volume in situations of decreased venous return to the heart and the systolic pressure variation is an indication of decreased cardiovascular function. A significant relationship has been documented between pulse pressure variation and plasma volume loss and measurement of pulse pressure variation has been used to predict an increase in cardiac output in a patient in response to volume loading (Durga et al. 2008; Pizov et al. 2010; Cannesson et al. 2011). These variations were observed on the arterial waveform of several sheep and may have been responsible for recorded values that were several mmHg different from pressures not influenced by controlled ventilation. Water was withheld from the sheep in this report for up to 12 hours before anesthesia and hypovolemia may have been present. Nonetheless, all sheep were administered the same dose rate of balanced electrolyte fluid to ensure consistent management.

The cuffs used for NIBP measurement in this study were ones recommended by the manufacturer for use with this monitor and one cuff size was used for each limb. The two breeds of sheep used in this study were of different conformation and one breed was thinner and leaner than the other. The cuffs used in this study achieved a ratio between cuff width and limb circumference of 53–66%. Early investigations of NIBP methods recommended cuff width:limb circumference ratio to be 40–60% for increased accuracy (Geddes et al. 1980; Sawyer et al. 1991). Availability of several different sizes of cuffs to reduce the ratio closer to 40% might have produced different results. The cuffs could have been placed above or below the hock or above or below the carpus, however, the locations of cuff placement were constrained by the surgical procedures performed. The wool on the pelvic limb was clipped only because a surgical procedure was planned for that site in some sheep after NIBP measurements were collected; therefore, all sheep were clipped to be consistent. Contact between cuff and skin would be improved after clipping and a thick wool coat under the cuff potentially could result in measurement error from nonuniform pressure distribution. Further investigation is needed to evaluate the effect of clipping versus no clipping, although there was no significant difference in correlation between measurements from the thoracic and pelvic limbs in this study suggesting that wool did not have an appreciable effect on the relationship between invasive and noninvasive measurements. Clipping of hair induced no significant difference in measurements in a study of an NIBP monitor in cats (Branson et al. 1997).

The range of pressures was not wide but consisted of values that might be expected during anesthesia. Hypotension was of greatest concern and that was the focus of the study; the project was not designed to evaluate hypertension. The method used in this study to decrease blood pressure was an increase in administration of isoflurane. The end-tidal concentration of isoflurane was not high compared to the minimum alveolar concentration (MAC value) of 1.4% isoflurane for sheep (Okutomi et al. 2009) but the infusion of ketamine or ketamine and lidocaine would be expected to significantly decrease the requirement for isoflurane (Doherty et al. 2007; Raske et al. 2010). The change in pulse pressure between light and deep anesthesia may have influenced measurements. Hypotension induced by isoflurane is primarily the result of vasodilation. Response of the monitor to hypotension associated with vasoconstriction, or decreased cardiac output such as following hemorrhage, may be different.

Paired NIBP and invasive blood pressure measurements were obtained during metacarpal surgery. There were no discernible responses to the thoracic limb procedures indicating that the nerve blocks were complete and surgical stimulus was not influencing the blood pressure on a beat-to-beat basis. This is supported by the fact that the correlation between NIBP and invasive measurements for thoracic and pelvic limbs was not significantly different.

The NIBP monitor evaluated here will be useful for monitoring anesthetized sheep. The data reported here indicate that an NIBP SAP measurement of ≤80 mmHg will identify hypotension associated with vasodilation. Use of the monitor without knowledge of its limitations may lead to diagnosis of hypotension from MAP when hypotension is not present. The clinical relevance of this inaccuracy depends on the action taken to treat the hypotension. Administration of a fluid challenge should be followed by urinary excretion without problems in healthy animals. However, retention of fluid during anesthesia despite only modest infusion of fluids has been reported to occur in human patients and in dogs (Boscan et al. 2010; Cagini et al. 2011). This failure to maintain fluid balance following anesthesia is multifactorial in origin and the significance is yet to be determined. The administration of a positive inotrope such as dobutamine to animals incorrectly diagnosed as hypotensive should not have serious consequences as the subsequent rise in blood pressure would be detected by the NIBP monitor and the dobutamine infusion would be adjusted.

Conclusions and clinical relevance

Our hypothesis was not proven for all parameters. The oscillometric NIBP model studied provided MAP measurements that were acceptable by ACVIM standards and SAP measurements that satisfied all criteria except that the correlation was 0.87 and not 0.9. The comparisons of first measurements obtained were not significantly different from averaging three successive measurements, except that the MAP correlation decreased to 0.89. However, the monitor underestimated MAP and indicated hypotension when hypotension was not present. The DAP measurements were not validated. No significant difference was found between pressures obtained from cuffs on the pelvic limb or thoracic limb.