Respiratory disease has long been recognized as of considerable economic importance in newborn foals (1), and as an important cause of morbidity and death in neonates presented for veterinary care (2, 3). Optimal respiratory support is highly desirable to optimize survival and preserve respiratory function in animals bred largely for their athletic potential.

The use of non-invasive ventilation (NIV) is now widely regarded as the most effective approach for respiratory support of human neonates (4, 5), with continuous positive airway pressure (CPAP) shown to reduce the number of preterm infants requiring admission to neonatal intensive care (6), and to decrease the risk of bronchopulmonary dysplasia or death in neonates requiring respiratory support (7). The technique involves the delivery of a constant positive (greater than atmospheric) pressure to the airway and preserves spontaneous respiration. The physiological effects are complex and likely to vary depending on the underlying pathology (8), but benefits have been attributed to increased functional residual capacity, decreased work of breathing and reduced airway resistance (4). Previous studies have demonstrated that CPAP is associated with improved respiratory function in a number of veterinary species (9–12). CPAP has recently been shown to improve gas exchange in healthy foals with pharmacologically induced respiratory suppression (13), however hypercapnia was observed in treated foals in this study, and has been observed previously in anesthetized horses during CPAP (10, 11, 14).

Bi-level positive airway pressure (BiPAP) is also recognized for the management of respiratory insufficiency in human neonates, and has demonstrated improved treatment outcomes in preterm human neonates in comparison to CPAP (15, 16). By using lower expiratory pressures, BiPAP promises improved expiratory function and is recommended for management of conditions associated with hypercapnia, such as chronic obstructive airway disease or asthma (17–19). In human patients with obstructive airway conditions, expiratory airflow limitations may cause increased PaCO₂ due to overdistension of alveoli and consequent increased alveolar dead space (20–23), an effect which has been termed dynamic hyperinflation (24).

Whereas, human neonates exhale passively (25), both inspiration and expiration are active processes, requiring muscular effort, in foals (26). Foals might therefore be predisposed to expiratory flow limitations and retention of CO₂ if active breathing strategies are unable to overcome expiratory pressures during CPAP, and the technique might be thus be associated increased intrinsic positive end-expiratory pressure (PEEPi) and alveolar overdistension, and hence predispose to hypercapnia. Lower expiratory pressures during BiPAP might be expected to facilitate expiration and ameliorate this effect. The current study was undertaken to determine whether a commercially available bi-level respiratory device, available for the home care of people with respiratory disease, might represent a ventilatory support strategy in healthy foals with pharmacologically induced respiratory insufficiency. We hypothesized that BiPAP would be associated with improved respiratory function, and with less CO₂ retention, than observed previously during CPAP (13). Intentionally, the study was designed to evaluate low cost intervention and monitoring strategies that might be safely implemented in equine practice or on farm.

No effects attributable to replicate were observed for any parameter. Effects associated with sequence (BiPAP or supplementary O₂ at T2) are shown in Supplementary Materials. Heart and respiratory rates were highest in unsedated foals, and temperature decreased significantly throughout the study period (from 38.5 to 37.8°C). Mean arterial pressure did not change associated with time or treatment (Supplementary Figure S2).

Oxygenation (PaO₂) was greater in unsedated foals (T-1) than observed in sedated foals at T0 (P = 0.002) or T1 (P = 0.004, Figure 1). The administration of supplementary oxygen alone or with BiPAP was associated with significantly increased PaO₂ in comparison to results at all other sampling times (all P < 0.001), and results following BiPAP were significantly greater than after O₂ administration (P = 0.021). Arterial CO₂ (PaCO₂) results were not normally distributed and were resistant to transformation. Values were lowest in unsedated foals at T-1, and differences at this time and at T0 were significant when compared to results following administration of supplementary O₂ and following BiPAP (all P < 0.001, Figure 1); differences at other times were not significant. Results following supplementary O₂ administration were not significantly different to those obtained after BiPAP (P = 1.000). Hypercapnia (PaCO₂ >60 mHg) was observed for two foals (F8 and F9) following O₂ administration (60 and 66 mmHg, respectively), and following BiPAP on both occasions for F9 (60 and 68 mmHg). Changes to blood pH mirrored changes to PaCO₂ and effects on lactate and blood glucose treatment attributable to treatment were not observed (Supplementary Figure S3). Blood glucose concentrations increased across all sampling times (likely due to administration of xylazine), with results at T3 (P = 0.013) and T5 (P = 0.008) significantly higher than at T0, as were results during O₂ (P = 0.008) and BiPAP (P = 0.016, data not shown).

Significant time-sequence interactions were observed for spirometry variables including respiratory rate during spirometry (RRs), tidal volume (Vt), inspiratory time (Ti), expiratory time (Te), and peak inspiratory flow (PIF) (Supplementary Figure S4); BiPAP was associated with significantly lower RR (P = 0.014) and significantly longer inspiratory (P < 0.001) and expiratory (P = 0.020) times at T2 than observed following O₂ administration at this time. Differences at other time points were not significant, so data have been combined for analysis of treatment effects. Sedation with diazepam at T0 was associated with a significant decrease in RRs relative to each other sampling times (Figure 2), except following BiPAP (all P < 0.050). Respiratory rate (RRs) during BiPAP was lower than was observed at T1 (P = 0.041), T3 (P = 0.012) or following O₂ administration (P < 0.001). Tidal volume (Vt) was greatest in standing foals following diazepam sedation (T0) and values observed at this time and in unsedated foals at T-1 were significantly greater than observed in recumbent foals with the exception of during BiPAP (all P < 0.05). The administration of BiPAP was associated with significantly greater Vt than was observed at T1 (P = 0.001), T3 (P < 0.001), T5 (P = 0.011) and during O₂ administration (P < 0.001), and effects were more pronounced at T2 than at T4 (Supplementary Figure S4). Recumbency was associated with a significant reduction in minute ventilation relative to results from standing, unsedated foals (T-1), but this effect was not observed during BiPAP (Figure 2).

Significant effects were observed for both inspiratory and expiratory time, reflective of changes observed in RRs (Figure 2), but there was no effect on I:E ratio (data not shown). Peak inspiratory flow was greatest during BiPAP (−1.52 L/s), and significant effects were observed compared to values obtained at T1 (P = 0.005) and T3 (P = 0.037). Expiratory flows were greatest in unsedated foals (T-1), and significant differences were observed at T1 (P = 0.002), T3 (P = 0.016), and during BiPAP (P = 0.003).

Time and sequence effects were observed for data derived from analysis of inspired/expired gas composition due to differences following O₂ administration or BiPAP at T2 (Supplementary Figure S5). Differences at other time points were not significant, so data have been combined for analysis of treatment effects. As expected, the administration of supplementary O₂ was associated with an increased FiO₂ during both mask supplementation and BiPAP, compared to FiO₂ when breathing room air (all P < 0.001), and values were significantly greater during BiPAP than during administration of O₂ only (P = 0.004). Oxygen concentrations in expired air were also greater following the administration of supplementary O₂, but differences were not observed between mask O₂ administration and BiPAP (P = 0.172). Oxygen extraction was much greater during BiPAP than at all other time points (all P < 0.005), including during O₂ administration (P < 0.001, Figure 2). Oxygen extraction was also greater during mask O₂ administration than at T1 (P < 0.001) or T5 (P = 0.004), and in standing foals at T0 relative to T3 (P = 0.046) and T5 (P < 0.001). Maximum concentrations of CO₂ (CO₂max) were observed at T0 in standing foals following administration of diazepam and observed differences were significant in comparison with results at T-1 (P = 0.016), T1 (P = 0.0003), T3 (P = 0.026), T5 (P = 0.0002), and following O₂ administration (P = 0.031). Results after BiPAP were significantly greater than following O₂ administration (P = 0.047). Minimum concentrations of CO₂ were observed during BiPAP, and differences were significant in comparison with T3 (P = 0.021) and T5 (P = 0.013). The elimination of CO₂ was greatest at T0 in diazepam sedated foals, with significant differences observed between results at T0 and all other sampling points, with the exception of BiPAP (all P < 0.030, Figure 2). Results observed following BiPAP were significantly greater than observed at T1 (P = 0.047) or following O₂ administration (P = 0.004).

Paired results for pulse oximetry (SpO₂) and hemoglobin saturation determined by blood gas analysis (sO₂) were available for 47 data sets from the current study. SpO₂ results correlated significantly with sO₂ (r = 0.61, 95% CI 0.34–0.78, P < 0.001), but there was poor agreement between these two methods of assessing hemoglobin saturation (Figure 3). Although bias was minimal (1.4%, standard deviation 5.4%), the observed limits of agreement were large (−11.95 to 9.2%), and increased divergence was observed for results obtained from the most hypoxic foal. Paired results for CO₂max and PaCO₂ were available for 136 data sets (Figure 3). There was poor but significant correlation between results (r = 0.25, 95% CI 0.09–0.40, P = 0.003), and agreement was poor (bias 14.0 ± 8.9%) with broad limits of agreement (−3.5 to 31.5%).

Improved blood oxygenation was observed following BiPAP relative to mask administration of supplementary O₂. BiPAP was also associated with decreased RR, increased Vt and increased minute ventilation, suggesting that respiratory mechanics were improved by respiratory support. Increased Vt represents a more efficient ventilation strategy than increased RR, as there is increased alveolar ventilation relative to ventilation of airway dead space, and decreased RR is likely to be associated with decreased work of breathing. Increased inspiratory pressure during BiPAP was associated with increased PIF, and the adverse effects on PEF observed during CPAP in previous studies (13) were not observed in the current study, presumably due to the lower expiratory pressures during BiPAP, relative to CPAP. As expected, increased FiO₂ was observed during mask administration of supplementary O₂ and during BiPAP, so observed increases in arterial oxygenation likely reflect the steeper diffusion gradient resulting from these changes. Surprisingly, FiO₂ was higher during BiPAP than during O₂ administration to foals. This observation was not expected during positive pressure respiratory support, where increased flow is associated with decreased partial pressure of O₂ (27).

Increased PaCO₂ was observed in foals following both forms of respiratory support, including hypercapnia (PaCO₂ >60 mmHg) following O₂ administration (two foals) or BiPAP (one foal). Hypercapnia has been reported in response to O₂ supplementation in human neonates (28) and foals (13, 29), and may be due to reduced respiratory drive, increased metabolic rate, hypoventilation due to sedation or effects of equipment dead space. As the minimum inspired CO₂ was greater during the administration of supplementary O₂ than during BiPAP, it is likely that mask administration of supplementary O₂ in the current study was associated with CO₂ retention, a problem that might be avoided by the use of a non-rebreathing valve during BiPAP, or by nasal insufflation of oxygen. This effect was not observed during BiPAP. Despite decreased RR, minute ventilation during BiPAP was the same as observed in standing, unsedated foals in the current study suggesting that BiPAP prevented reduced ventilation associated with sedation and recumbency. However, our hypothesis, that lower expiratory pressures associated with BiPAP and improved expiratory function would ameliorate hypercapnia was not demonstrated. As was observed during CPAP (13), the observed increases in PaCO₂ and pH in the current study were modest, and consistent with current ventilation strategies that accept increased arterial CO₂ tension and hypercapnic acidosis (“permissive hypercapnia”) as acceptable consequences without adverse effects on outcome (30, 31), and with possible therapeutic effects (32). Sedation, and the supraphysiological PaO₂ values observed in the current study, might have contributed to the observed hypercapnia.

Although reversed within 10 min of cessation of BiPAP, the observed hypercapnia demonstrates the necessity for close monitoring during the implementation of respiratory support in equine neonates. Evaluation of NIV should consider effects on both oxygenation and carbon dioxide, but the current study demonstrated poor agreement between non-invasive methods of continuous monitoring of O₂ saturation and CO₂ accumulation. Pulse oximetry has been recommended as an appropriate alternative to invasive sampling for determination of hemoglobin saturation, and previous studies have suggested that placement of transmission or reflectance sensors on the lip or tongue ensures the most reliable assessment of SaO₂ in foals (33). However, bias and limits of agreement in that study (33) were similar to those observed in the current study and well-outside accepted standards of care (34). Probe placement on the lip or tongue was not possible during mask administration of respiratory support in the current study, and was not tolerated by unsedated foals. Direct measurement of PaO₂ and sO₂ by co-oximetry is more accurate than blood gas analysis for determination of oxygenation (35), but both techniques require arterial samples, and neither provides an immediate result or allows for continuous monitoring. Arterial samples can be difficult to obtain in hypovolaemic foals or animals with distal limb oedema, and risks associated with arterial puncture include pain, hemorrhage, arterial injury, aneurism formation, thrombosis, and distal ischaemia. End-tidal CO₂ (PETCO₂) is commonly monitored during anesthesia as an indirect measure of PaCO₂. Whilst PETCO₂ has been reported as an acceptable technique for monitoring of neonatal foals (34), studies during NIV in people have suggested that, as observed in the current study, the technique was not predictive of PaCO₂ or changes in PaCO₂ (36). The association between PETCO₂ and PaCO₂ assumes that the patient exhales fully, and that end-expiratory gases approximate gas composition in the alveoli. This assumption is not valid if, as we have hypothesized, foals are not exhaling completely. For this reason peak expired CO₂ concentrations have been termed FeCO₂max in the current study. Alternative techniques to assess ventilatory function, such as volume capnography and electrical impedance tomography (37), offer greater capacity to more accurately assess response to respiratory support.

A number of limitations were noted in the current study. Findings in healthy foals with pharmacologically-induced respiratory insufficiency may not be predictive of responses in neonates with spontaneous disease, and characterization of the technique in neonates with spontaneous respiratory disease is required. We were unable to document alveolar ventilation, physiological dead space or PEEPi in the current study, which was designed to evaluate a readily available intervention and monitoring strategies that might be readily implemented in a practice setting. The assessment of non-invasive monitoring was based on a small number of observations, and the data set did not include results from foals with severe hypoxaemia or hypercapnia.

Consistent with previous studies evaluating CPAP, BiPAP was an effective respiratory support strategy for healthy foals with pharmacologically induced respiratory insufficiency. BiPAP was associated with increased PaO₂, more efficient gas exchange and changes in respiratory mechanics including increased tidal volume, decreased respiratory rate, and increased peak inspiratory flow. The technique preserved minute ventilation in the face of reduced ventilation observed at other times associated with sedation and recumbency, but was associated with modest increase in PaCO₂. As in previous studies, the use of a commercially available ventilator intended for at-home care of adults with chronic obstructive respiratory conditions or sleep apnoea represents an available and potentially cost effective option for use in equine practice, although there is a need for careful and frequent monitoring of patient oxygenation and ventilation (carbon dioxide elimination) during NIV. Our results suggest that monitoring of alveolar ventilation, pressure-volume curves, and PEEPi might be important for effective NIV of foals and to better characterize the response of foals to respiratory support. The use of lower expiratory pressures in the current study did not prevent hypercapnia, and increases in PaCO₂ observed in the current study were similar to those observed previously during CPAP in healthy foals. Observed effects on PaCO₂ were rapidly reversed and predominantly within acceptable bounds for permissive hypercapnia. Although not a primary objective of the current study, our results suggested the non-invasive monitoring approaches used in this study were not reliable, and techniques are needed for more accurate, non-invasive assessment of respiratory function in foals during NIV.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Charles Sturt University Animal Care and Ethics Committee (ACEC A18044).

All authors contributed to experimental design and conduct of experimental procedures. SR and LB contributed to data analysis and early manuscript drafts, with editing and correction by all authors.

This study was funded by AgriFutures Australia (PRJ-011159). A contribution towards costs of publication was received from Charles Sturt University's Tri-Faculty Open Access Publication Scheme.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.