INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Careful selection of inspiratory flow is a major factor in the achievement of satisfactory respiratory muscle rest with mechanical ventilation. If the delivered flow is not sufficient to meet a patient's flow demands, inspiratory work increases significantly (1, 2). In addition, for a given tidal volume (VT) and total respiratory cycle time, an increase in inspiratory flow results in a shortening of the ventilator's inspiratory time (TI,_vent) and an increase in expiratory timea frequent goal in patients experiencing patient-ventilator dyssynchrony owing to gas trapping (3). However, an increase in delivered inspiratory flow is accompanied with an increase in respiratory frequency (f) in healthy subjects (4) and in intubated patients receiving mechanical ventilation (5). This flow-associated alteration in f has been shown to be independent of breathing route, inspired gas temperature, and delivered volume, and it persists after anesthesia of the airway (6). A change in inspired VT has also been shown to be associated with a change in f during wakefulness and sleep, under both isocapnic and hypocapnic conditions (7, 8).

In the above-cited studies, the investigators did not control for TI,_vent (6, 7); when VT was increased and inspiratory flow kept constant, TI,_vent increased and when inspiratory flow was increased and VT kept constant, TI,_vent decreased. Thus, the possible role of the concurrent changes in VT and TI,_vent in mediating the flow-associated changes in f has not been carefully delineated. Accordingly, we undertook a study designed to define the separate influences of delivered flow, TI,_vent, and VT on f during mechanical ventilation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fifteen healthy men, age 23 to 46 yr (mean 30.5), volunteered for the study; all but one were naïve to the purpose of this investigation. The study was approved by the Human Studies Subcommittees of Hines VA Hospital and Loyola University Medical Center, and informed consent was obtained from all subjects.

The study was performed in a quiet room with the subject resting on a bed in the supine position. Subjects were ventilated through a nasal continuous positive airway pressure (CPAP) mask (Med Systems, San Diego, CA) connected to a Puritan-Bennett 7200 ventilator (Puritan-Bennett, Carlsbad, CA). Airway pressure and flow were recorded continuously via a pneumotachograph placed between Y-piece of the ventilator circuit and the CPAP mask. Initially, the subject breathed spontaneously through the ventilator in the CPAP mode with positive end-expiratory pressure (PEEP) of 0 cm H₂O. The ventilator was then set in the assist-control mode; the back-up f was set at 1 breath/min, ensuring that all breaths were triggered by the subject. Inspiratory flow was set at 60 L/min and delivered with a square-wave profile. Neither supplemental oxygen nor PEEP was used. Delivered VT was initially set at the level that the subject inspired during spontaneous breathing in the CPAP mode, and then increased by 100 ml every minute until the subject felt comfortable. At this point, the subject acclimated to these "baseline" ventilator settings over 7 to 10 min before the experimental protocols were performed.

Experimental Protocols

Targeted flow protocol. The aim of this protocol, performed in eight subjects, was to determine the effect of a change in delivered inspiratory flow on f, while VT was kept constant, i.e., mean VT of 0.9 ± 0.2 (SD) L. By study design, TI,_vent was allowed to vary. Once the subject had acclimated to the baseline ventilator settings, VT was held constant while delivered flows of 30, 60, and 90 L/min were applied. Each setting was maintained for a minimum of four to five breaths, and a minimum of six transitions of flow were recorded in each subject. To minimize any confounding influence from CO₂ flux on breathing pattern, analysis was confined to 3 breaths before and 3 breaths after a transition in each protocol.

Targeted VT protocol. The aim of this protocol, performed in seven subjects, was to determine the effect of a change in delivered VT on f, while inspiratory flow was kept constant. By study design, TI,_vent was allowed to vary. Once the subject had acclimated to the baseline settings, the inspiratory flow was held constant at 60 L/min while VT settings of 0.5, 1.0, and 1.5 L were applied. At each VT setting, a minimum of eight transitions of VT were recorded in each subject.

Targeted flow-VT balance protocol. The aim of this protocol, performed in 10 subjects, was to determine the effect of changes in delivered VT and inspiratory flow on f, while TI,_vent was kept constant. To achieve a constant TI,_vent of 1 s, three selected inspiratory flows, 30, 60, and 90 L/min, were balanced by VT settings of 0.5, 1.0, and 1.5 L, respectively. Each setting was maintained for at least five breaths and at least eight transitions were obtained in each subject.

Targeted inspiratory pause protocol. The aim of this protocol, performed in seven subjects, was to determine the effect of a change in TI,_vent on f, while inspiratory flow and VT were kept constanti.e., mean VT of 0.8 ± 0.2 L. Once the subject had acclimated to the ventilator settings, VT and flow were held constant while inspiratory pauses of 0.4, 0.8, 1.4, and 2 s were applied. Each setting was maintained for at least 5 breaths. At least 19 transitions were recorded in each subject.

Data Analysis

For each trial that involved transitions in inspiratory flow, VT, or TI,_vent, three breaths before the transition and three breaths after the transition were analyzed. Analysis included a breath-by-breath determination of f based on the flow signal. For each trial, the mean f for the three breaths before and after the transition were calculated; the values of f from similar trials were pooled and the mean calculated for each subject. Data were analyzed by analysis of variance (ANOVA) with repeated measurements. A p value of < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Targeted Flow Protocol

An increase in delivered flow from 30 L/min to 60 and 90 L/min caused increases in f from 12.9 ± 2.9 breaths/min to 15.5 ± 3.4 and 18.2 ± 3.5 breaths/min, respectively (p < 0.001), and decreases in TI,_vent from 1.94 ± 0.48 s to 1.00 ± 0.20 and 0.70 ± 0.14 s, respectively (p < 0.001). The response of f to the imposed alterations in TI,_vent is shown in Figure 1 (left panel), and the correlation between f and TI,_vent (r = 0.69, p < 0.001) is shown in Figure 2, left panel. The coefficients of variation of f and TI,_vent for each of the flow settings in this and subsequent experiments are shown in Table 1.

View larger version (18K): [in this window] [in a new window]   Figure 1.   (Left panel ) Respiratory frequency in eight healthy subjects receiving assist-control ventilation, in whom TI,vent was varied by alterations in inspiratory flow at a constant VT. (Right panel ) Respiratory frequency in seven healthy subjects in whom TI,vent was varied by alterations in VT at a constant inspiratory flow. In both experiments, the imposed decrease in TI,vent resulted in increases in f (p < 0.001).

View larger version (17K): [in this window] [in a new window]   Figure 2.   The relationship between TI,vent and f during mechanical ventilation with a targeted inspiratory flow and constant VT (eight subjects; r = 0.69, p < 0.001) (left panel ), with a targeted VT and constant inspiratory flow (seven subjects; r = 0.86, p < 0.001) (middle panel ), and with a targeted inspiratory pause and constant VT and constant inspiratory flow (seven subjects; r = 0.86, p < 0.01) (right panel ).

Targeted VT Protocol

At a constant inspiratory flow, an increase in delivered VT from 0.5 L to 1 and 1.5 L caused a decrease in f from 23.6 ± 4.3 breaths/min to 16.2 ± 1.7 and 12.4 ± 4.1 breaths/min, respectively (p < 0.001). The response of f to the imposed alteration in TI ,_vent is shown in Figure 1 (right panel), and the correlation between f and TI,_vent (r = 0.86, p < 0.001) is shown in Figure 2, middle panel.

Targeted Flow-VT Balance Protocol

When flow was increased from 30 to 60 L/min, but balanced by changes in VT to maintain a constant TI,_vent, f decreased from 17.9 ± 5.1 to 16.0 ± 3.6 breaths/min (p < 0.05); no further change in f was noted when flow was increased to 90 L/min (15.5 ± 4.7 breaths/min) (Figure 3, left panel).

View larger version (18K): [in this window] [in a new window]   Figure 3.   (Left panel ) Respiratory frequency in 10 healthy subjects receiving mechanical ventilation at a VT of 0.5, 1.0, and 1.5 L balanced by flows of 30, 60, and 90 L/min, respectively, to achieve a constant TI,vent of 1 s. When flow was increased from 30 to 60 L/min, f decreased from 17.9 ± 5.1 to 16.0 ± 3.6 breaths/ min (p < 0.05); no further change in f was noted when flow was increased to 90 L/min (15.5 ± 4.7 breaths/min). (Right panel ) Respiratory frequency in seven healthy subjects receiving mechanical ventilation at a constant VT and constant inspiratory flow, in whom TI,vent was varied by alterations in end-inspiratory pause. An increase in TI,vent produced a decrease in f (p < 0.001).

Targeted Inspiratory Pause Protocol

The influence of imposed TI,_vent, achieved by variations in the inspiratory pause, on f was assessed at inspiratory flows of 60 (n = 4) and 90 L/min (n = 3). At the two flow settings, the resulting increase in TI,_vent caused a decrease in f (p < 0.001 for both settings). An increase in the applied inspiratory pause from 0 to 2 s caused f to decrease from 14.1 ± 1.8 to 10.1 ± 1.5 breaths/min (p < 0.001) at flow of 60 L/min and from 16.8 ± 0.6 to 11.7 ± 1.1 breaths/min (p < 0.001) at a flow of 90 L/min. Because the pause-induced decrease in f was equivalent for the two flow settings, the f values for the two flows were combined (Figure 3, right panel). The correlation between TI,_vent and f (r = 0.86, p < 0.001) at the combined flow settings is shown in Figure 2 (right panel).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Separate alterations in ventilator-delivered VT and inspiratory flow induced changes in f that were proportional to the associated alteration in TI,_vent (Figures 1 and 2). Alterations in delivered VT and inspiratory flow that were not accompanied by a change in TI,_vent did not produce unidirectional changes in f (Figure 3, left panel), and, contrary to current thinking, increases in inspiratory flow did not invariably induce tachypnea.

Puddy and Younes (4) have shown that an increase in inspiratory flow is associated with an increase in f, and Puddy and coworkers (7) and Tobert and coworkers (8) have shown that an increase in VT is associated with a decrease in f. These observations are in accord with the f response in our targeted-VT and targeted-flow protocols. However, the reported associations between an increase in VT and a decrease in frequency (7), and also between an increase in inspiratory flow and an increase in f (4), appear to conflict with some of our findings; we found that when VT settings of 1 and 1.5 L were balanced by flows of 60 and 90 L/min to achieve a constant TI,_vent, f was not altered (Figure 3, left panel). Reconciliation of these findings would require that, for flow rates at or above 60 L/min, the inhibitory action of an increase in VT on f be balanced exactly by the excitatory action of an increase in flow on f. It is even more difficult to accommodate the data in our targeted inspiratory pause protocol with the conclusions of Puddy and coworkers (7) (Figure 3, right panel). Reconciliation would require that the respiratory controller instantaneously integrates flow and T_I,_ventincluding the time during an inspiratory holdand, based on the computed mean inspiratory flow, the controller sets f. A more plausible explanation is that the respiratory controller is able to respond to the imposed TI,_vent during mechanical ventilation independently of the change in inspiratory flow and VT. When TI,_vent was decreased, as in the targeted-flow protocol, f increased (Figure 1, left panel); when TI,_vent was increased, as in both the targeted VT-protocol (Figure 1, right panel) and the targeted inspiratory pause protocol (Figure 3, right panel), f decreased; and when TI,_vent was kept constant although inspiratory flow was increased from 60 to 90 L/min, as in the targeted flow-VT balanced protocol, f did not change (Figure 3, left panel). The possibility that TI,_vent per se can modulate f in some experimental conditions is supported by the observation of Younes and coworkers (9) that a mechanical inspiratory time in excess of neural inspiratory time causes prolongation of expiratory time and, thus, delays the onset of the next inspiratory effort and slows f. The linkage of neural expiratory time to neural inspiratory time is well recognized (10).

The findings in our flow-VT balanced protocol are similar, in part, to those of Tobert and coworkers (8). When they increased inspiratory flow rates from 24 to 48 L/min, but maintained a constant TI,_vent through an increase in VT, f decreased although they did not state if the change in f reached statistical significance. We extended this observation to include flow rates commonly used in clinical practice, i.e., 60 and 90 L/min, and when we balanced them with VT settings to achieve a constant TI,_vent, f remained constant. Our inspiratory pause protocol indicates that it is not VT per se, but rather the time that VT is applied, i.e., TI,_vent, that determines a change in f in certain experimental conditions.

The mechanisms responsible for the observed response in f can be reconciled, at least in part, with current knowledge of the Hering-Breuer reflex. First, for inspiratory flows ranging between 26 and 81 ml/s in cats, Baker and coworkers (11) suggest that graded inhibition of inspiration is not a function of inspiratory flow; instead, it is related to the delivered VT above functional residual capacity. In other words, the duration of neural TI is decreased in proportion to the time taken for a given VT to be delivered. More recent data in human subjects also suggest that inhibition of neural TI is related to the delivered VT rather than to an increase in inspiratory flowat least in the case of flows between ~ 48 and ~ 90 L/min (see Figure 8 in Reference 12). In addition, shortening of neural TI predisposes to a decrease in neural TE (10). Accordingly, volume-associated reduction in neural TI (11) probably accounts for the increase in f associated with increasing inspiratory flow in our flow-targeted protocol. Second, tonic stimulation of the vagus during neural TEequivalent to afferent discharge of the vagus throughout the time that lung inflation is maintained during neural TEcan lead to an increase in duration of neural TE (13), and prolongation is proportional to the intensity of the vagal stimulation (13). This mechanism for neural TE prolongation is likely responsible for the observed reduction in f accompanying the increase in VT in our targeted VT protocol. Third, the duration of neural TE has been shown to increase in proportion to the time that the vagus is stimulated at a constant level during exhalation (13). The latter mechanism could explain the decrease in f in our targeted inspiratory pause protocol. Fourth, the results of the protocol wherein targeted flow and VT were balanced to achieve a constant TI,_vent can be explained by the combined influences of inflation time and inflation volumeboth of which inhibit neural TI and prolong neural TE. For a given inflation time, a larger VT will obviously be delivered when flow is set at 60 versus 30 L/min. This increase in VT for a given inflation time can, in turn, effect a decrease in neural TI (11, 12). At the same time, an increase in VT from 0.5 to 1.0 L predisposes to the continuation of lung inflation into neural TE, which, in turn, can result in prolongation of neural TE (13). These changes in neural TI and neural TE have opposing effects on f; because switching the settings from a flow of 30 L/min and a VT of 0.5 L to a flow of 60 L/min and a VT of 1.0 L resulted in a decrease in f, it is likely that the prolongation of neural TE was dominant. When the settings were switched from a flow of 60 L/min and a VT of 1.0 L to a flow of 90 L/min and a VT of 1.5 L, the f responses were not uniform in our subjects. This nonuniformity may have resulted from a volume-induced reduction in neural TI combined with short-lasting lung distention during exhalation. The duration of overlap between imposed inflation and neural TE in some subjects may have been sufficient to prolong neural TE (and thus increase f), whereas in other subjects the overlap may not have been sufficient to effect a change in respiratory timing (see Figure 3).

The results of our study have important implications for patient management. During assist-control ventilation, physicians attempt to maintain a normal acid-base status by manipulation of delivered VT. However, as shown in Figure 1 (right panel ), an increase in VT without a change in inspiratory flow leads to an increase in TI,_vent, and a reciprocal decrease in the subject's f; the opposite scenario holds for a decrease in delivered VT. The change in f will negate or diminish the desired effect of the increase in VT, making it difficult to predict the change in minute ventilation that will result for a given change in VT. When VT was increased from 0.5 to 1.5 L in our subjects, minute ventilation increased from 11.7 ± 2.2 to 18.5 ± 2.5 L/min, which is less than the level of 35.0 ± 6.5 L/min expected for the same VT, should f have remained unchanged (p < 0.001).

Triggering of the ventilator is more difficult in patients with airflow limitation and dynamic hyperinflation, where the end-expiratory lung volume exceeds the relaxation volume of the respiratory system. We have previously shown (3) that breaths preceding nontriggering efforts had a shorter total respiratory cycle time, shorter expiratory time, and a higher intrinsic PEEP. In such a situation, inspiratory flow is commonly increased to achieve a decrease in TI,_vent and, thus, allow more time for exhalation. However, this practice may have the opposite effect; an increase in inspiratory flow from 30 to 90 L/min caused an increase in expiratory time in two subjects only, and in those two subjects the increase in expiratory time was marginal.

In summary, when the delivered VT or inspiratory flow were changed in such a manner that TI,_vent was allowed to increase, f decreased; alterations in delivered VT and inspiratory flow at a constant TI,_vent did not produce unidirectional changes in f. In conclusion, the imposed TI,_vent during assist-control ventilation can determine f independently of the change in inspiratory flow and VT.