Comparison of Thames Medical CAT+ Doppler and AutoCAT+ Automatic Blood Pressure Monitor devices for non-invasive blood pressure measurement in cats

Abstract

Objectives

A comparative assessment of systolic blood pressure (SBP) measurement was carried out for anaesthetised and conscious cats by using two non-invasive approaches based on the Doppler method and a newly developed oscillometric instrument.

Methods

SBP was recorded on 131 occasions in 26 cats entering a shelter environment. Six of these cats were monitored while under a general anaesthetic for elective procedures and the rest were conscious during routine health assessment. A paired approach was followed using the Doppler method followed immediately by the oscillometric approach. Mean values and coefficient of variations were calculated. A normal distribution was confirmed before a standard Bland–Altman analysis was completed.

Results

The mean SBP (±SD) for the 131 paired readings was 113.3 ± 23.9 mmHg and 116.3 ± 26.7 mmHg for the Doppler and oscillometric methods, respectively. The small difference in means was not significant. Anaesthetised cats had significantly lower SBP values than those that were conscious. The data set for 16 cats with three replicated paired measurements and a subset of 12 with five such replicated measurements also provided similar normally distributed mean values and a high correlation coefficient. The Bland–Altman plot suggested a positive bias of the oscillometric approach of +3.07 ± 12.3 mmHg (limits of agreement of −21.0 to 27.1 mmHg) and +4.93 ± 9.38 mmHg (limits of agreement of −13.5 to 23.3 mmHg) for the 16- and 12-cat subgroups, respectively. There was 100% agreement between the two methods in allocation to a hypertension class.

Conclusions and relevance

The results establish that the new AutoCAT+ instrument met some of the guidelines for assessing such instruments for veterinary use with cats with a normal range of blood pressure. Further work is needed with a larger data set spanning hypotension to hypertension for complete validation.

Introduction

The measurement of systolic blood pressure (SBP) is a vital parameter that should be routinely assessed in all cats. The International Society of Feline Medicine recommends a blood pressure assessment at least every 12 months in cats aged 7 years and over. ¹ Systemic hypertension and target organ damage are common clinical findings in feline medicine. They require accurate and regular blood pressure measurements for diagnosis and monitoring. ¹ Blood pressure should also be routinely measured in cats undergoing general anaesthesia. ²

The most accurate and precise measurement of SBP requires direct measurement with a pressure transducer reading via a catheter inserted into an artery.^{3 –6} This makes it an inappropriate approach for routine clinical investigations on grounds of cost, the need for anaesthesia and appropriate expertise. Consequently, an indirect approach prevails for routine measurement using an inflatable/deflatable cuff attached to the limb or tail of a cat, with the procedures well defined.^1,7 A commonly used approach uses a Doppler probe emitting high frequency sound waves coupled to a detector that provides an audible signal when the decreasing cuff inflation allows arterial blood flow to restart. The advantages of this non-invasive method (portable, inexpensive, conscious patient, less technical skill to perform, rapid result) are somewhat offset by concerns that readings more closely relate to mean blood pressure rather than the higher peak SBP.^{3 –6} This might lead to a false-negative interpretation if the clinician refers to SBP levels.

There is a longstanding interest in developing alternatives to the Doppler approach also using an inflatable/deflatable cuff. Oscillometric methods measure the difference between the cuff and arterial pressure continuously as the cuff deflates. An average between the systolic and diastolic blood pressure is obtained, and machine-specific algorithms are used to estimate SBP. ⁸ Such equipment has been calibrated against direct SBP measurements^{3 –6} and the Doppler method.^{9 –11}

The Doppler method of measuring blood pressure has been found to be more repeatable and easier to obtain than oscillometric readings. ¹² Common problems reported are detecting a signal and issues associated with the cuff. Many veterinary professionals report greater confidence in Doppler machines. ¹³ However, some veterinary professionals report that they find oscillometric machines easier and quicker to use. ¹³ The main concerns with oscillometric machines reported were cuff placement and the frequency of failure to gain a measurement.^{12 –14} Many veterinary professionals report that cats often appear more tolerant of oscillometric machines.

It is good practice to calibrate all oscillometric equipment for the measurement of SBP. The purpose of this study was to use standard procedures to compare the readings of a new oscillometric machine for use with cats with that of the accepted standard Doppler machine.

Materials and methods

Animals

This study took place at a rescue cattery (RSPCA Taylor’s Rehoming Centre, Dorchester) during routine health visits to the cats there and at Castle Veterinary Clinic during general anaesthesia for elective procedures. A total of 97 paired measurements were made on conscious cats and 34 paired measurements for those under general anaesthesia. Conscious cats were examined in their normal environment (RSPCA Taylor’s Rehoming Centre). Anaesthetised cats had the measurements taken as part of their routine anaesthetic monitoring at Castle Veterinary Clinic.

Machines

A single Doppler (CAT+ Doppler; Thames Medical) and a single oscillometric machine (AutoCAT+ Automatic Blood Pressure Monitor; Thames Medical) were used throughout the study.

Measuring blood pressure

After being allowed to acclimate to the handling area with a familiar handler, each cat had the right forelimb clipped distal to the carpal pad on the palmar aspect and ultrasound gel was applied. Each cat then had their limb measured with a tape measure, and blood pressure cuff size was determined using the range printed on the cuff (CAT+ Doppler; Thames Medical). This ensured that cuff width was 30–40% that of limb circumference. The cat was acclimated to the cuff by inflating and deflating it. The sensor of the Doppler blood pressure machine (CAT+ Doppler; Thames Medical) was applied to the palmar aspect of the distal metacarpals, and blood pressure was taken using the common digital branch of the radial artery. Immediately after taking the blood pressure measurement, the cuff was disconnected from the sphygmanometer and attached to the oscillometric machine (AutoCAT+; Thames Medical) to measure SBP. Thus, the paired readings were taken in the same position as close as possible temporally. Repeat paired measurements were made when cats remained calm and allowed repeated measurements. This resulted in a varying number of paired readings per cat for analysis. All readings for conscious cats were taken by the author. Readings obtained under general anaesthetic were taken by a registered veterinary nurse who had received training in the use of both methods of blood pressure measurement.

Statistical analysis

Most analyses used a standard statistical package (SPSS v30; IBM) with figures produced using Excel (Microsoft). Methodology followed a standard text. ¹⁵ Means (± SD) were obtained and given throughout. The value sets were compared using paired sample t-tests and Pearson’s correlation coefficient (r) was calculated. Linn’s concordance correlation coefficient (Lin’s CCC) was determined for the data pairs, and Fisher’s r to Z transformation was carried out to derive its 95% confidence limits using an application in Excel. ¹⁶ Linear regression was completed for individual paired measurements and subsequently for a subset of 16 and 12 cats for which a minimum of three or five paired replicated measurements were gained, respectively (SPSS, Regression, Curve estimation). The effect of anaesthesia on those cats receiving that treatment was analysed using Univariate ANOVA in SPSS for both the Doppler and AutoCAT+ measurements. Bland–Altman plots were obtained in SPSS (Analyze, Descriptive statistics) and plotted in Excel with the limit of agreement set for the bias of 1.96 ± SD. A power analysis was carried out after other analyses to determine an appropriate sample size for further work using the online R Statistics Package (version 4.5.1) with the package pwr.

Results

The SBP of a total of 26 cats was measured during a routine health examination or general anaesthetic by the Doppler and then the AutoCAT+ methods with a minimal delay between switching instruments. They had known ages ranging from 6 months to 15 years.

A total of 131 paired readings were obtained, but the number of paired readings achieved for each cat depended on it remaining calm. These are plotted in Figure 1 and include some high and low values. No values were set aside. The mean for these 131 values analysed per cat using the Doppler method was 113.3 ± 23.9 mmHg. The coefficient of variation (CV) was 21.1%. The corresponding mean for the AutoCAT+ method was 116.3 ± 26.7 mmHg (CV = 23.0%). The small difference in the two means is not significant (P = 0.51).

Figure 1

Relationship between the two methods for (a) individual 1 31 paired measurements of systolic blood pressure (SBP) of 26 cats, (b) the means for 16 cats for which there were a minimum of three replicated paired measurements and (c) the means for 12 cats with a minimum of five replicated paired readings. Cats were either under a general anaesthetic (closed symbols) or conscious (open symbols)

A fuller comparison of means and their variance was based on 16 cats, for which a minimum of three paired measurements were achieved, and the subset of 12 animals, for which at least five paired measurements were made. The occasional aberrant readings observed in Figure 1a do not affect the resultant means (Figure 1b,c). The means with three paired replicated readings were 106 ± 24.4 mmHg (CV = 22.9%) and 110 ± 24.1 mmHg (CV = 21.9%) for the Doppler and AutoCAT+ approaches, respectively. These means were similar using the paired t-test (P = 0.245). For the 12 cats from which at least five paired readings were obtained, the means were 98.6 ± 22.6 mmHg (CV = 22.8%) and 105 ± 23.5 mmHg (CV = 24.5%). These means were also similar using the paired t-test (P = 0.165). Pearson’s r and Lin’s CCC for the group of 12 cats was 0.826 (range 0.480–0.950) and 0.934 (range 0.809–0.978), respectively. Values in brackets are 95% confidence limits. Lin’s CCC was determined using the fifth replicate reading for each cat. The effect of anaesthesia had a significant effect on SBP for both the Doppler and AutoCAT+ methods (P <0.001, univariate ANOVA). That approach also established that there was a small but significant effect of age for the group of 16 cats with both methods (P <0.025 in both cases) but not for cuff size (P = 0.52 in both cases). The means for sex were significantly different for AutoCAT+ only (P = 0.009).

Linear regression of the AutoCAT+ and Doppler methods (Figure 1a) established a significant correlation (P <0.001) for the 131 paired readings. The slope with the intercept set at zero was 1.018 ± 0.021, which does not differ from 1 (t = 0.866; P >0.5). As expected, that high level of relationship was achieved for the paired readings with either three or five replicated measurements, with R² indicating that 98.7% and 99.3% of the variance were described by the regression lines for three and five replicated measurements, respectively (Figure 1b,c).

Further analysis of the level of agreement of the two methods was based on the standard Bland–Altman analysis for the data set with three replicated measurements and its subset with five. This plots the residual SBP value for AutoCAT+ after subtracting the paired Doppler reading against the mean for the two readings (Figure 2). This analysis was conducted after confirming the difference between the two methods did not deviate significantly from a normal distribution. The limits of agreement (LOA) is set at ± 1.96 SD from the mean.

Figure 2

A Bland–Altman plot for (a) the means for 16 cats, for which there were a minimum of three replicated paired measurements, and (b) the means for a subset of 12 of those cats, with five replicated measurements. Each mean bias is given with its SD value and shown as a solid line relative to a zero bias, which is given as a dotted line. The upper and lower limits of agreement are set at ± 1.96 SD from the mean and shown as dashed lines. Cats were either under a general anaesthetic (closed symbols) or conscious (open symbols). SBP = systolic blood pressure

For the 16 cats with three pairs of replicated measurements, reading the mean bias was + 3.07 ± 12.3 mmHg, with the LOA shown as 1.96 × SD above and below that mean. For the subset of 12 cats with five replicated measurements, the mean bias was 4.93 ± 9.38 mmHg and a similar range for the LOA (Figure 2).

All cats but one present in both the three and five replicated data set were in the no-risk category for hypertension based on the four risk categories used previously. ¹¹ The remaining male castrated cat aged 11 years had a mean SBP of 148 ± 2.47 mmHg and 151 ± 2.41 mmHg using the Doppler and AutoCAT+ methods, respectively. This small difference was statistically significant (P = 0.19), but both methods placed it in the low-risk category (140–159 mmHg). ⁷

The paired t-test for the data set of 12 cats has a Cohen’s d of 0.429 and a Lin’s CCC of 0.934. Using these values, the estimated appropriate sample size for future work is 25 cats (R package pwr; P = 0.05).

Discussion

Two of the key requirements of a blood pressure measurement approach are accuracy and precision. The AutoCAT+ method showed similar mean values compared with the Doppler method for 131 paired readings of 26 cats, with means of 113.3 ± 23.9 mmHg and 116.3 ± 26.7 mmHg, respectively. The CV values of approximately 20% are appreciable but within the range listed in previous work. ⁹ Further analysis was based on the 16 cats for which three paired replicated measurements were made and the subset of 12 cats with five replicated paired measurements. In both cases, the values did not deviate significantly from a normal distribution. These preliminary tests ensured reliable application of the Bland–Altman analysis. The small difference in means is demonstrated in Figure 2. The mean bias values are 3.07 ± 12.3 mmHg and 4.93 ± 9.38 mmHg for three and five paired replicated measurements, respectively. This suggests that the accuracy of the AutoCAT+ method is very similar to that of the Doppler approach; however, that conclusion requires validation with a larger dataset. Previous work has established that oscillometric approaches frequently provide higher mean SBP than the Doppler method. In previous work on cats, 6/7 oscillometric means were greater than those obtained by the paired Doppler approach by 8–21 mmHg on five occasions and 53 mmHg for the sixth.^{9 –11} The Doppler method has also been reported to underestimate SBP cats relative to direct measurements.^{3 –6} The positive bias of the AutoCAT+ approach in this work was less than in several previous studies. In this preliminary dataset of 16 cats, classification agreement was complete in all but one case.

It is standard practice to use replicated measurements of blood pressure to assess variability. This is assessed by the mean bias and the LOA (± 1.96 SD of the differences) in Figure 2. The LOAs were ± 24.1 mmHg and ± 18.4 mm Hg for the three and five paired replicated measurements, respectively. This indicates an acceptable level of precision. The values are less than 50% of those in three previous comparisons of oscillometric and Doppler approaches.^9,11,17 This and the Lin’s CCC value of 0.934 indicate an encouraging level of precision. The bias in this work as in that of others is clinically wide. The resultant risk of a false-positive or false-negative diagnosis of hypertension should be mitigated. This can be addressed by multiple assessments over several weeks or in 7–14 days when severe hypertension is putatively detected. ⁷

Guidelines for assessing oscillometric monitors for veterinary use in the USA are available. ¹⁸ The AutoCAT+ instrument meets many of the latter’s guidelines. The 12 cats with five replicated measurements provided a Pearson’s correlation coefficient (r) of 0.826 and a Lin’s CCC of 0.934. Overinterpreting the strength of association that they indicate is unwise but they suggest a moderate fit.^19,20 AutoCAT+ shows promising potential based on several commonly used features of device validation, including some specified in previous work. ¹⁸ A power analysis suggests a subsequent study with 25 individuals is sufficient for further evaluation and an achievable number of cats. It should include a wide range of hypotension and hypertension values.

Conclusions

The results establish that the new AutoCAT+ instrument met guidelines for assessing such instruments for veterinary use with cats. Further work is needed with a larger data set spanning the hypertension range for complete validation.