Measurement of tidal volume using Respiratory Ultrasonic Plethysmography in anaesthetized, mechanically ventilated horses

Abstract

Objective  To compare tidal volume estimations obtained from Respiratory Ultrasonic Plethysmography (RUP) with simultaneous spirometric measurements in anaesthetized, mechanically ventilated horses.

Study design  Prospective randomized experimental study.

Animals  Five experimental horses.

Methods  Five horses were anaesthetized twice (1 week apart) in random order in lateral and in dorsal recumbency. Nine ventilation modes (treatments) were scheduled in random order (each lasting 4 minutes) applying combinations of different tidal volumes (8, 10, 12 mL kg⁻¹) and positive end-expiratory pressures (PEEP) (0, 10, 20 cm H₂O). Baseline ventilation mode (tidal volume = 15 mL kg⁻¹, PEEP = 0 cm H₂O) was applied for 4 minutes between all treatments. Spirometry and RUP data were downloaded to personal computers. Linear regression analyses (RUP versus spirometric tidal volume) were performed using different subsets of data. Additonally RUP was calibrated against spirometry using a regression equation for all RUP signal values (thoracic, abdominal and combined) with all data collectively and also by an individually determined best regression equation (highest R²) for each experiment (horse versus recumbency) separately. Agreement between methods was assessed with Bland-Altman analyses.

Results  The highest correlation of RUP and spirometric tidal volume (R² = 0.81) was found with the combined RUP signal in horses in lateral recumbency and ventilated without PEEP. The bias ± 2 SD was 0 ± 2.66 L when RUP was calibrated for collective data, but decreased to 0 ± 0.87 L when RUP was calibrated with individual data.

Conclusions and clinical relevance  A possible use of RUP for tidal volume measurement during IPPV needs individual calibration to obtain limits of agreement within ± 20%.

Introduction

Hypoventilation is a common side effect of general anaesthesia in horses. Recumbency and anaesthetic drugs often cause a decrease in tidal volume and respiratory rate (Robinson 2009). Mechanical ventilation with or without positive end-expiratory pressure (PEEP) may be necessary during anaesthesia to correct hypoventilation and improve gas exchange. The choice of appropriate settings for mechanical ventilation is, amongst other factors, facilitated by the measurement of tidal volume and respiratory rate. Pitot tube based spirometry at the level of the endotracheal tube is a practical method to perform this measurement during routine clinical anaesthesia in horses (Moens et al. 2009). An alternative completely non-invasive approach for estimating tidal volume and respiratory rate uses the measurement of thoracic and abdominal excursions following volume changes of the lungs during spontaneous and mechanical ventilation. Respiratory Inductive Plethysmography (RIP) is based on this principle and is an established method to monitor respiration in infants (Warren & Alderson 1986) but it is also used in standing horses (Amory et al. 1994; Miller et al. 2000; Hoffman et al. 2001, 2007). The RIP technique is based on measurement of changes of a cross-sectional area. Electrical coils embedded in belts are placed around the thorax and abdomen and changes in circumference are translated into changes of electric inductance, which are then recorded. If simultaneously, respiratory volumes are measured by another method, this signal can be calibrated for respiratory volumes (Cohn et al. 1982; Tobin et al. 1983; Brown et al. 1998). Respiratory ultrasonic plethysmography (RUP) is a new technology which allows measurement of the changes of thoracic and abdominal circumferences based on the measurement of sound velocity through fluid filled tubes (Schramel 2008; Schramel et al. 2012).

The aims of this study were to evaluate the relationship of RUP output data and spirometric tidal volume measurements in horses under general anaesthesia undergoing intermittent positive pressure ventilation (IPPV).

Materials and methods

The study was discussed and approved by the institutional ethical committee and the Austrian Federal Ministry of Science and Research (TVG Nr 68205/171-II/106/2008). Eight horses were enrolled in the study, being (mean ± SD) 9.8 ± 11.7 years old and weighing 517 ± 68.5 kg. The animals were considered healthy based on clinical examination, routine haematological and biochemical blood testing and lung function testing prior to the study. The data of the first three horses were excluded from analysis because a modification of the characteristics of the reference (baseline) ventilatory modus was necessary to avoid interference from occasional spontaneous respiratory efforts during IPPV. The remaining five horses were 13 ± 6 years old and weighed 531 ± 77 kg. Each horse was anaesthetized twice, once in dorsal and once in left lateral recumbency in random order and with a 1 week interval between anaesthetics.

Anaesthesia and instrumentation

Prior to anaesthesia, the animals were starved for 12 hours, but allowed free access to water. A 12 gauge intravenous (IV) catheter was placed in the jugular vein and the horses were sedated with 0.02 mg kg⁻¹ acepromazine IV. About 20 minutes later, the animals received 0.75 mg kg⁻¹ xylazine and 0.025 mg kg⁻¹ butorphanol IV, followed 5 minutes later by induction of anaesthesia with 2.2 mg kg⁻¹ ketamine and 0.1 mg kg⁻¹ diazepam. After the horses became recumbent, the trachea was intubated (inner diameter 30 mm) and the animal was hoisted and positioned on the table. The endotracheal tube was connected to a large animal circle breathing system and mechanical ventilation was started immediately using 15 mL kg⁻¹ tidal volume and 0 cm H₂O PEEP (L.A. 95 – Smith ventilator; HSH Anaesthetic Equipment, Denmark). Anaesthesia was maintained with isoflurane in oxygen. Additionally, a constant rate infusion of ketamine (6 mg kg⁻¹ hour⁻¹), xylazine (0.3 mg kg⁻¹ hour⁻¹) and midazolam (18 μg kg⁻¹ hour⁻¹) was administered throughout the anaesthetic period. The animals received 0.9% NaCl solution at 7–10 mL kg⁻¹ hour⁻¹ IV. Physiological parameters were monitored using ECG, pulse oximetry, capnography and a direct arterial blood pressure monitoring system (Datex-Ohmeda S/5 Monitor; Planar Systems Inc., USA). Spirometry was performed with a mainstream sensor (Horse-Lite; Morpheus Engineering, Netherlands) attached between the endotracheal tube and the circle breathing system and connected to the Datex-Ohmeda S/5 Monitor. The RUP sensors, two ethanol filled rubber tubes with ultrasonic transducers at one end, were placed around the thorax (11th intercostal space) and around the abdomen (caudal of the last rib) and connected to a personal computer. Arterial blood was taken every 15 minutes, and analysed for pH and blood gases. Animals recovered in a routine manner and without assistance.

Mechanical ventilation

Nine different ventilation-modes were used in random order as treatments during each anaesthetic event. These consisted of combinations of different tidal volumes (8, 10, 12 mL kg⁻¹) and PEEP (0, 10, 20 cm H₂O). Each treatment ventilation mode was maintained for 4 minutes. Between treatments, the baseline-ventilation mode was applied for another 4 minutes using a tidal volume of 15 mL kg⁻¹ and PEEP of 0 cm H₂O.

Data acquisition and analysis

Data from the RUP system was transmitted via Bluetooth to a personal computer at a sampling rate of 10 Hz. Spirometry waveform data (volume, flow, pressure) were continuously downloaded to another computer at 25 Hz sampling frequency (S/5 Collect software; Datex-Ohmeda, Inc., GE Healthcare, Finland). Downloaded spirometry waveform data were resampled to 10 Hz (Matlab 7; The MathWorks, Inc., USA), exported to spreadsheets (Sigmaplot 11; Systat Software, Inc., USA) and aligned appropriately with the RUP data on the same time scale. Baseline ventilation modes were not included in the analysis because those were overrepresented in number compared to the treatments. Five artefact free consecutive breaths were selected for analysis from each ventilation mode. The highest and lowest values of tidal volume waveforms taken from spirometry and of thoracic and abdominal RUP waveforms from each respiratory cycle were determined and manually collected from the respective plots. The differences between the highest and lowest values were calculated resulting in datasets of spirometric tidal volumes, Δ thoracic and Δ abdominal RUP respectively. A dataset Δ combined RUP was created by addition of Δ thoracic and Δ abdominal RUP values.

Linear regression analyses were used to evaluate the relations between spirometric tidal volume data and each of the three Δ RUP datasets. The resulting coefficients of determination (R²) were used to identify factors influencing the correlation between spirometric tidal volume and Δ RUP measurements. The following subsets of data were generated and evaluated: ‘recumbency’, ‘PEEP’, ‘recumbency and PEEP’ and ‘individual horse and recumbency’ (Table 1). Next, Δ RUP circumference measurements were calibrated against spirometric tidal volume using regression equations (RUP = spirometry volume × slope + intercept). These equations were reorganised (calibrated RUP = (RUP−intercept)/slope) and applied to each Δ RUP dataset to calculate tidal volumes measured by RUP. Thereafter all gained data was calibrated. For collective calibration, the regression equation was calculated using either Δ thoracic, Δ abdominal or Δ combined RUP data from all horses. For individual calibration, regression analyses were performed separately for each horse and each anaesthetic event with each RUP type (thoracic, abdominal or combined) and those equations which provided the best fit (i.e. left lateral recumbency, thoracic RUP) were selected for calibration. The calculated tidal volumes of RUP and the spirometric volumes were used to generate difference and mean plots (RUP versus spirometry) and the 2 SDs of the differences were used to assess agreement between the devices (Bland & Altman 1986). Regression analyses were considered significant when p < 0.05.

Table 1. Coefficients of determination (R²) resulting from linear regression analyses of tidal volume (measured by spirometry) versus changes in thoracic, abdominal or combined (thoracic + abdominal) Respiratory Ultrasonic Plethysmography (RUP) measurements in mechanically ventilated horses under anaesthesia

Row	Horses	Recumbency	PEEP (cm H2O)	Sample-size	Thoracic RUP (R^2)	Abdominal RUP (R^2)	CombinedRUP (R^2)
Different subsets of data (each row) are analyzed separately. These subsets were generated by splitting the whole dataset using the following grouping factors: recumbency (rows 2–3), PEEP (rows 4–6), recumbency and PEEP (rows 7–12), horses’ names and recumbency (rows 13–22). Coefficients of determination (R^2) indicate the predictive value of a model therefore the higher the coefficient is, the more suitable is RUP for measuring tidal volume in a given subset of data. All, whole dataset of the group indicated on the top of the column; DR, dorsal recumbency; LR, left lateral recumbency; PEEP, positive end-expiratory pressure; R^2, coefficient of determination.
1	All	DR + LR	All	90	0.29	0.34	0.43
2	All	DR	All	45	0.22	0.35	0.36
3	All	LR	All	45	0.50	0.38	0.58
4	All	DR + LR	0	30	0.52	0.43	0.63
5	All	DR + LR	10	30	0.28	0.34	0.44
6	All	DR + LR	20	30	0.16	0.38	0.37
7	All	DR	0	15	0.57	0.42	0.58
8	All	LR	0	15	0.81	0.45	0.79
9	All	DR	10	15	0.26	0.39	0.35
10	All	LR	10	15	0.58	0.32	0.69
11	All	DR	20	15	0.10	0.43	0.36
12	All	LR	20	15	0.29	0.39	0.42
13	Benno	DR	All	9	0.01	0.80	0.17
14	Benno	LR	All	9	0.96	0.95	0.97
15	Hansi	DR	All	9	0.90	0.81	0.90
16	Hansi	LR	All	9	0.62	0.87	0.77
17	Incampo	DR	All	9	0.72	0.74	0.86
18	Incampo	LR	All	9	0.31	0.60	0.63
19	Indigena	DR	All	9	0.67	0.73	0.71
20	Indigena	LR	All	9	0.74	0.81	0.78
21	Zoltaire	DR	All	9	0.64	0.31	0.47
22	Zoltaire	LR	All	9	0.90	0.82	0.96

Results

In the current study RUP was easy to use and none of the animals had to be excluded due to technical problems. Scatter plots with the original measurements are shown in Fig. 1 and the coefficients of determination (R²) are presented in Table 1. In all grouped data the agreements were weak, as indicated by low R² values for the three RUP variables (row 1 in Table 1). There were tendencies for lower R² values in dorsal compared to lateral recumbency (rows 2–3 in Table 1) and when PEEP was applied (rows 4–6 in Table 1). As a consequence, the highest R² values were seen for horses in lateral recumbency without PEEP, especially for the thoracic Δ RUP data (row 8 in Table 1) and the lowest ones in dorsal recumbency with 20 cm H₂O PEEP, again with Δ thoracic RUP data (row 11 in Table 1). There was a general tendency for higher R² values with the Δ combined RUP data (row 1–12 in Table 1). When the individual experimental settings (each horse in each recumbency) were examined, the R² values were generally much higher compared to the collective analysis, but none of the RUP datasets proved to be uniformly better than the others (rows 13–22 in Table 1).

Figure 1.  Scatter plots of tidal volume (measured by spirometry) versus changes in circumferences of thorax and abdomen or the combination of both (delta thoracic (a), abdominal (b) or combined (c) measured by Respiratory Ultrasonic Plethysmography (RUP)) in five mechanically ventilated horses under anaesthesia (n = 90). There were significant correlations between variables in each plot (regression lines are indicated). The coefficients of determination are indicated in Table 1.

The results based on the clustering of all individual data and subsequent calibration showed wide limits of agreement with the spirometric volumes for each of the RUP variables in the Bland-Altman analysis. The biases ± 2 SD were 0 ± 3.62 L, 0 ± 3.21 L and 0 ± 2.66 L for Δ thoracic, Δ abdominal and Δ combined RUP respectively (Fig. 2a; only Δ combined RUP data are shown). Furthermore, systematic deviations could be found between differences and means of all three RUP variables, suggesting a tendency for RUP to underestimate tidal volume at lower and an overestimate it at higher volumes (Fig. 2a). Calibration for each experiment separately improved the agreement markedly (0 ± 0.87 L); systematic relationships between differences and means were not found (Fig. 2b).

Figure 2.  Difference versus mean plots of tidal volumes measured by spirometry and following calibration by Respiratory Ultrasonic Plethysmography in five mechanically ventilated horses under anaesthesia (n = 90). The results of collective and individual calibrations are shown on plot (a) (collective) and (b) (individual), respectively. Mean ± 2 SD of differences are indicated by dashed lines. Significant correlation exists between variables on plot (a) but not on plot (b) (regression line is indicated).

Discussion

The main finding of this study was that RUP – derived tidal volume measurements, need individual calibration to achieve limits of agreement within ± 20% of spirometric measurement.

In principle there is a strict linear relationship between inspired volume and the lung volume displaced and recorded by RUP during spontaneous ventilation. However during mechanical ventilation this relation becomes nonlinear due to gas compression effects and possible changes in thoracic compliance during the course of anaesthesia. To the best of our knowledge this relationship has never been investigated in mechanically ventilated horses. In the present study using mechanical ventilation, the overall correlation between spirometric tidal volume and tidal volume estimated by RUP was found to be low. The best correlation (R² = 0.43) was obtained using the combined RUP signal. This is much less than measured with RIP in awake resting lambs (R² = 0.99), but of similar magnitude as in lambs during a phase of active sleep (R² = 0.46) (Warren et al. 1988). In the study presented here, data from the thoracic and abdominal RUP band as well as the combination of both were included in the analysis. This is also used for studies with RIP and refers to the fact, that both, thorax and abdomen play an important role in generating the tidal volume (Konno & Mead 1967; Brown et al. 1998; Aliverti et al. 2011).

Many factors can cause discrepancies between delivered volume as measured by spirometry and the volume displaced and indirectly measured by RUP. Small differences are due to blood shifts in and out of the chest, the net result of gas exchange for carbon dioxide and oxygen and the expanding effect of humidification and warming of the respiratory gas. Other reasons for poor correlations can be asynchrony of breathing and changes in body positioning of the patient. Asynchrony of breathing, characterised by phase differences between thoracic and abdominal compartment movements, has been described in children and horses (Sorenson & Robinson 1980; Warren & Alderson 1985, 1986). However asynchrony per se is not present during mechanical ventilation as both thoracic and abdominal compartments should always move in phase. There may be independent changes of the compliance in the thoracic or abdominal compartment present due to individual relaxation of corresponding muscles during general anaesthesia. The horses in this study were not paralysed, and possibly clinically undetectable changes in respiratory and abdominal muscle relaxation may have caused compliance changes in the two compartments. This would have contributed to the discrepancy in volume measurements although they should be minimised by using the combined RUP signal.

In the present study the correlation between RUP and spirometry was significantly influenced by body position with much better R² in lateral recumbency. This effect was predominantly observed in the thoracic RUP band. It can be hypothesised that, in the dorsal position, the inter-individual variability is increased by a more pronounced pressure of the abdominal organs onto the diaphragm, resulting in changes of the thoracic shape and compliance. This effect would influence the correlation between tidal volume and thoracic RUP negatively.

Another factor playing a role in this study with horses undergoing IPPV was the fact that the relation between spirometric volume and the volume displaced (lung volume) differs fundamentally during mechanical ventilation compared to spontaneous respiration. The airway pressure applied induces a compression of gas in the lungs. This causes a non linear relationship between airway pressure and lung volume with the slope indicating compliance. Also the gas compression effect is not constant and generates an inherently non-linear and variable relation between spirometric volumes and lung volumes (Schramel et al. 2012). Differences between inspired and displaced lung volume have been described previously (Matamis et al. 1984; Aliverti et al. 2011). In the present study various tidal volume and PEEP combinations caused different levels of airway pressure and thus compression of gas. Increases in PEEP in the present study generated a decrease in correlation, probably by further deviation from the linear relationship used for calibration. This decrease was most pronounced in the thoracic RUP, which was expected as the thorax has been shown to contribute mostly to the generation of the tidal volume in the anaesthetized, mechanically ventilated human patient (Konno & Mead 1967; Lumb 2005). The degree of gas compression will also depend on respiratory system compliance (lungs plus chest wall). Alterations in the respiratory system compliance such as lung volume changes or lung disease will alter the difference between spirometric and RUP derived volume measurements. Especially in anaesthetized, recumbent horses, ventilated with different combinations of tidal volume and PEEP changes in compliance can be expected, for example, when atelectatic lung areas are recruited. The influence of individual body shape (shape of abdomen and thorax) on regional distribution of ventilation and on gas exchange has been described previously (Moens et al. 1995; Mansel & Clutton 2008). All aforementioned factors are thought to contribute to the poor inter-individual correlation of the calibration results. Indeed individual calibration per body position markedly decreased the limits of agreement. Individual calibration has also been described in human medicine with RIP measurements but usually require active patient cooperation (Tobin et al. 1983; Sackner et al. 1989). Other factors can possibly contribute to the poor correlation of spirometry and RUP. All horses underwent lung function tests but no clinically important individual differences were noted which could account for individual variability in the correlation of RUP and spirometry. It is also not clear if additionally the specific modalities during positioning of the horses on the table had an influence on the outcome of this study. Animals in lateral recumbency were placed on an air filled mattress, with the upper front and back legs supported by cushions, which allowed the uppermost part of thorax and abdomen to expand freely. Horses in dorsal recumbency however were placed between two air filled cushions and their front legs tied to the sides of the table. Whilst the abdomen could still move freely, the movements of the thorax were possibly limited. To detect and avoid interference with RUP measurements due to displacement of RUP belts, their position was regularly checked. The presence of a urinary catheter prevented an effect of an increased bladder size on the abdominal circumference.

RUP is able to detect small changes in circumferences of the thorax and abdomen and can be used to assess respiratory rate, pattern and thoraco-abdominal synchrony. It has the advantage of being non-invasive and with negligible resistance to breathing. However, in the clinical setting the location of the RUP belts might interfere with surgery such as the abdominal belt with colic surgery.

Despite its advantages described above, tidal volume measurement with RUP suffers from important bias even after individual calibration. Therefore it is not likely that RUP can replace spirometry for tidal volume measurements in horses undergoing IPPV. It is possible that RUP will perform better during spontaneous respiration when inspired volumes more closely equal lung volume changes.

Acknowledgements

The authors thank Roche for lending the Cobas b221System blood-gas analyser used in this study.