INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

PRESSURE SUPPORT VENTILATION (PSV) has been one of the most frequently applied modes for partial ventilatory support. Because patients under PSV have control of the ventilatory rate and the inspiratory assist time, they feel more comfortable with PSV than with other partial ventilatory support modes (e.g., synchronized intermittent mandatory ventilation) (11). Nevertheless, recent clinical studies have revealed that patients under PSV may frequently encounter patient-ventilator asynchrony. Ventilators may not be in synchrony with the onset of the patient inspiratory effort, which causes inspiratory asynchrony (or trigger asynchrony). Studies on inspiratory asynchrony have indicated that inspiratory asynchrony is related to a high patient work in breathing and difficulty in weaning patients from the mechanical ventilation (7, 8). In addition, patient-ventilator asynchrony may also be present during the onset of exhalation, i.e., expiratory asynchrony (9, 10, 14, 16). In this situation, the termination of the ventilator flow occurs either before or after patients stop their inspiratory efforts. Expiratory asynchrony not only causes discomfort to patients but costs patients unnecessary inspiratory and expiratory work as well (12). When the termination of the ventilator flow falls behind the end of the patient inspiratory effort (i.e., delayed termination), the patient recruits his expiratory muscles to "fight" against the ventilator flow, which increases expiratory workload (10). When the termination of the ventilator flow occurs before the end of the patient inspiratory effort (i.e., premature termination), inspiratory muscle work continues into or even throughout the ventilator's expiratory phase, thus resulting in inefficient inspiratory muscle work (16). Furthermore, a high lung volume caused by the previous breath with delayed termination may result in trigger failure of the subsequent inspiratory effort in patients with chronic obstructive pulmonary disease (COPD) (14). Premature termination in PSV, on the other hand, sometimes causes retriggering of inspiration and a stuttering pattern of ventilator assistance (16). Although expiratory asynchrony has been of clinical concern for years, there are very few studies exploring the mechanisms of expiratory asynchrony (20). Younes (20) used computer simulation to evaluate the effects of selected levels of respiratory mechanics and patient effort on the duration of ventilator assistance time during PSV. Because his presentation was limited to relatively few, selected levels of resistance, compliance, and patient effort, more general relationships governing expiratory asynchrony cannot be deduced.

Flow cycling is the primary method for intensive care ventilators to terminate their inspiratory flow delivery during PSV. The ventilator is cycled off when the inspiratory flow has decayed to a certain level (i.e., termination criterion). Most current ventilators use an arbitrary termination criterion to terminate the inspiratory flow delivery during PSV (e.g., 5% of the peak flow in the Siemens Servo 300, 25% of the peak flow in the Siemens Servo 900 and Bird 8400ST, 5 l/min in the Nellcor Puritan Bennett 7200ae). These arbitrary criteria have been shown to cause expiratory asynchrony in certain patient categories. Van de Graaff and co-workers (16) found that patients with prolonged and slow inspiratory efforts sometimes suffered premature termination under PSV with the Siemens Servo 900C ventilator. With this same ventilator, Jubran et al. (10) revealed delayed termination of ventilator flow in patients with a long time constant and a high support pressure level. In a mechanical simulation study evaluating expiratory synchrony during PSV in different ventilators (17), we identified a delayed ventilator flow termination with weak inspiratory effort and long time constant using the Siemens Servo300. Surprisingly, ventilator flow was never cycled off by the flow criterion with the Nellcor Puritan Bennett 7200ae in the tested conditions.

With the understanding that a single level of the flow criterion in a specific ventilator probably would not satisfy all patient categories and the chances of premature termination and delayed termination are always likely, some ventilator manufacturers have introduced user-selectable termination criteria into their newest model ventilators in an effort to improve expiratory synchrony (e.g., Nellcor Puritan Bennett 840, Hamilton Galileo). Clinicians can select a termination flow criterion (expiratory trigger sensitivity) to optimize the expiratory synchrony. This function, although it provides flexibility to clinicians, breaks the simplicity of the application of PSV and has been shown to be difficult even with visual observation of the bedside airway pressure waveform (4, 6). Part of the difficulty in using this function is attributed to lack of clarification of the mechanisms of expiratory asynchrony in different patient mechanic conditions during PSV.

To better understand the mechanisms of expiratory asynchrony caused by the flow termination criteria in PSV of mechanical ventilators, we developed a mathematical model. Solving this model revealed that expiratory synchrony is governed by two ratios: the ratio of the respiratory time constant () to patient neural inspiratory time (TI) and the ratio of the ventilator set pressure support level (Pps) to patient inspiratory muscle pressure (Pmus). The use of this model for automation of expiratory trigger sensitivity in the development of future mechanical ventilators is also possible.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

The Model

(1)

On the basis of this differential equation, the inspiratory flow waveform during PSV can be solved analytically by assuming the form of the input pressure signals Paw(t) and Pmus(t).

Paw(t) is simulated by assuming that the ventilator is triggered as soon as inspiratory effort succeeds in either generating inspiratory flow or reducing Paw below the baseline pressure level (here assumed to be zero). Once the ventilator is triggered, Paw is assumed to exponentially increase to the Pps level with a ventilator time constant (_v) and then maintain that level until the termination of inspiration (Fig. 1). Thus

(2)

where t  0. The time course of Pmus can be approximated by the following second-order polynomial function (13)
where d is a constant; TI is defined as the time between the onset of the increase in inspiratory Pmus and the start of its decline. It is assumed that Pmus reaches its maximum and becomes flattened at the end of the neural inspiratory effort (t = TI); thus the maximal Pmus (Pmus max) can be expressed as Pmus max = d · TI². Substituting this into the above equation yields the expression

(3)

where 0  t  TI. Substituting Eqs. 2 and 3 into Eq. 1 and solving the resultant differential equation for the initial condition of (t = 0) yields

(4)

where  is the time constant of the patient respiratory system ( = R/E).

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Model-derived time courses of airway pressure (Paw; top) and inspiratory muscle pressure (Pmus; bottom). When the ventilator is triggered, Paw is assumed to exponentially increase to the set pressure support level (Pps) with a time constant of v . Pmus is approximated by a second-order polynomial function during neural inspiration time (TI). After the end of inspiration, Pmus decays exponentially during the entire expiration with a nearly linear decline (at a slope of ) during the first 0.1 s. Pmus max, maximal Pmus; t, time.

Rearranging the above equation to express (t) in terms of Pps/Pmus max and /TI yields the equation

(5)

This is a fundamental equation describing the inspiratory flow profile during PSV, which implies that the flow profile as the function of normalized time (t/TI) is determined by Pps/Pmus max, /TI, and _v/TI. Furthermore, this equation also implies that, during PSV at a given level of _v/TI, the ratio of the flow at the end of patient neural inspiration to peak inspiratory flow (TI/_peak) can be uniquely expressed and graphically plotted on the plane of Pps/Pmus max vs. /TI. This provides the framework for the following analyses.

Computation of TI/_peak. To simplify the calculations, _V/TI is assumed to be 0.06. This _V/TI value represents cases in which, for example, patient neural TI is 1.0 s and ventilator _V is 0.06 s [_V of 0.06 s corresponds to the flow acceleration percent setting of 90% in the Nellcor Puritan Bennett 840 ventilator (15)]. The inspiratory flow rates at t during PSV over the entire TI were then computed with Eq. 5 by stepwise changing Pps/Pmus max and /TI. _peak and TI were determined in these calculations. In this study, Pps/Pmus max was changed from 0.1 to 3.0 (at 0.1 intervals) and /TI from 0.1 to 2.0 (at 0.1 intervals). In this way, we obtained 30 × 20 matrices for _peak and TI, which yielded TI/_peak on the Pps/Pmus max-/TI plane. To evaluate the separate effects of /TI and Pps/Pmus max on TI/_peak, TI/_peak values as a function of /TI (ranging from 0.1 to 2.0 at 0.1 intervals) at five different levels of Pps/Pmus max (i.e., 1/3, 2/3, 1, 2, 3) were also plotted.

TI/_peak at the ranges of adult respiratory mechanics. Because expiratory asynchrony has been experienced primarily in adult applications of PSV (9, 10, 14, 16), we calculated TI/_peak values at wide ranges of adult respiratory mechanics. The respiratory mechanics used were lung compliance of 0.08, 0.04, and 0.02 l/cmH₂O; chest wall compliance of 0.2 l/cmH₂O; respiratory resistance of 5 and 20 cmH₂O · l¹ · s¹; Pps of 10, 20, and 30 cmH₂O; and Pmus max of 10 and 30 cmH₂O. The patient neural TI was assumed to be 1.0 s and _v/TI was the same as described above.

Expiratory synchrony with representative termination criteria. Two representative termination criteria were chosen in this study: 25 and 5% of peak inspiratory flow. This means that the ventilator is cycled off when the inspiratory flow decays to the level that is equal to the threshold level, i.e., 25 or 5% of peak inspiratory flow. With the use of these termination criteria, the ventilator will be in expiratory synchrony with the patient if TI/_peak is equal to the threshold level. It is in asynchrony with the patient when TI/_peak is higher than (delayed termination) or lower than (premature termination) the threshold level. When TI/_peak is higher than the threshold level at the end of patient neural inspiration, however, the decline in Pmus after neural inspiration augments the flow decay so as to decelerate flow to reach the threshold subsequently. Effect of the Pmus decline on the flow decay can be approximated as Pmus/R. Physiologically, the cessation of Pmus is not instantaneous; rather, the inspiratory muscle activity generally extends into the expiratory phase, resulting in residual inspiratory Pmus during neural expiration (13). In anesthetized humans and animals (2, 21), although the time course of the inspiratory Pmus during the entire expiration follows a curvilinear relationship, it is almost a linear decay during the first 0.1 s of the expiration. This is also supported by our data from the patients under mechanical ventilation with PSV (18). Taking this Pmus decay pattern into consideration, we designate the rate of the Pmus decline (in percent) during the first 0.1 s of expiration as  (Fig. 1), i.e., Pmus is assumed to decay by  × Pmus max. Therefore, the contribution of the Pmus decline to the flow decay after the end of TI would be modified to  × Pmus max/R. If we consider the delayed ventilator flow terminations of up to 0.1 s as synchronous terminations, the conditions of synchronous terminations can be written as
and
Rearranging

(6)

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Computation of TI/_peak

View larger version (18K): [in this window] [in a new window]   Fig. 2.   TI/peak plotted on the Pps/Pmus max-/TI plane. TI/peak, ratio of the flow at the end of the patient neural inspiration to the peak inspiratory flow (in percent); , time constant of the respiratory system.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   TI/peak as a function of /TI at different levels of Pps/Pmus max. Levels of Pps/Pmus max are 1/3 (A), 2/3 (B), 1 (C), 2 (D), and 3 (E).

TI/_peak at the Ranges of Adult Respiratory Mechanics

View this table: [in this window] [in a new window]   Table 1.   TI/peak at the ranges of adult respiratory mechanics

View this table: [in this window] [in a new window]   Table 2.   Correlation between TI/peak (y) and /TI (x) at the range of adult respiratory mechanics

Expiratory Synchrony With Representative Termination Criteria

View larger version (1514K): [in this window] [in a new window]   Fig. 4.   Divisions of the synchronous zone (A), premature termination zone (B), and delayed termination zone (C) with the termination criterion of 5% peak flow (top) and 25% peak flow (bottom). , Slope of the Pmus decline in the first 0.1 s of expiration. Left vertical curve represents the flow termination criterion line. The synchronous zone (A) is between this curve and middle curve ( 25% curve), assuming that  is 25%. The synchronous zone (A) would be extended to between the left and right curve ( 50% curve) if  is 50%, suggesting a faster decay of Pmus or that the active expiratory effort broadens the synchronous zone and narrows the delayed termination zone. Note that increasing the expiratory trigger sensitivity (i.e., termination flow threshold) from 5 to 25% shifts the synchronized zone (A) to the right.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Analysis of the mechanisms of expiratory asynchrony could be done with a mathematical model, computer simulation, mechanical model simulation, or even animal and/or human studies. Because it is difficult or unrealistic to manipulate the potentially involved parameters in animals and humans, animal and human studies do not fit well for this type of study. Mechanical model simulation and computer simulation have advantages because many parameters can be manipulated in a controlled manner and can be designed in a way that is mechanically closer to human physiology than mathematical models. However, the use of a mathematical model approach, with some assumptions and simplifications, allows easier elucidation of the basic rules behind expiratory asynchrony than the use of other approaches. In this study, we assumed the pressure-volume relationship to be linear in the range of tidal ventilation. We also assumed the pressure-flow relationship to be linear so that we could solve the mathematical model. This assumption, especially in intubated patients, is not true because the real pressure-flow relationship is more of nonlinearity. This nonlinearity will attenuate the peak flow and reduce TI/_peak, thus, in general, shifting the results presented here toward delayed termination. We could incorporate trigger delay time (T_delay) and trigger pressure (P_trigger) into the model by changing TI, Pmus, and Pps to TI  T_delay, Pmus max  P_trigger, and Pps + P_trigger, predisposing to delayed terminations; however, in the model, we assumed that the ventilator is triggered as soon as the patient starts inspiratory efforts for the simplicity of data presentation. The airway pressure waveform in the model was characterized by an exponential curve with a variable rate constant (1/_v), which incorporated changes in the slope of the pressurization. To simplify the data presentation, we fixed _v/TI at 0.06, which represents cases in which the patient TI is 1.0 s and the ventilator time constant is 0.06 s (_v of 0.06 s corresponds to the flow acceleration setting of 90% in the Nellcor Puritan Bennett 840 ventilator). On the basis of Eq. 5, an increase in _v/TI will reduce the peak flow and increase TI/_peak. It means that an increase in _v at a given TI will cause more delayed termination than what we presented in this study. This is consistent with the results from Bonmarchand et al. (3) who found that changing the pressurization time from 0.1 s to a maximum of 1.5 s greatly modified the ventilator TI, promoting delayed termination of inspiration. In the analyses of expiratory synchrony with the representative termination criteria (i.e., 25 and 5% of the peak flow), we arbitrarily defined a termination delay of up to 0.1 s as synchronized termination. These assumptions and simplifications can be easily criticized because they are not necessarily equivalent to those in reality; however, they are helpful and allow an in-depth mathematical analysis of the mechanisms of expiratory asynchrony in PSV.

With these assumptions and simplifications being kept in mind, this mathematical model study reveals the following results during PSV. 1) The ratio of the flow at the end of patient TI/_peak during PSV is determined by two ratios, /TI and Pps/Pmus max. 2) TI/_peak is affected more by /TI than by Pps/Pmus max. Within the data ranges presented in the Fig. 3, TI/_peak increases in a sigmoidal (s-shaped) relationship to /TI. An increase in Pps/Pmus max slightly shifts the TI/_peak-/TI curve to the right. 3) Under the selected adult respiratory mechanics, TI/_peak ranges from 1 to 85% and has an excellent linear correlation with /TI. 4) Single fixed levels of the flow termination criterion used in mechanical ventilators will always have chances of both synchronized termination and asynchronized (premature and delayed) termination, depending on the patient mechanics. An increase in /TI causes greater opportunity of delayed termination and less chance of premature termination. An increase in Pps/Pmus max narrows the synchronized zone, leaving the inspiratory termination predisposed to be in asynchrony (premature or delayed termination). Increasing the expiratory trigger sensitivity of a ventilator shifts the synchronized zone to the right, causing less delayed termination and more premature termination. An increase in  (indicating a faster Pmus decay or expiratory muscle activity) broadens the synchronous zone and narrows the delayed termination zone.

Although all factors, including Pps, Pmus, , and TI, influence TI/_peak, the weight of the influence of /TI is the most (Fig. 3, Table 1). TI/_peak is affected by /TI in a sigmoidal pattern across the full data ranges, as shown in Fig. 3. TI/_peak rises as  becomes longer at a given TI, which implies that the expiratory trigger sensitivity should be set higher in the conditions of longer time constant. As an example of the results from this study, TI/_peak turns out to be higher than 60% in the conditions of 20 cmH₂O · l¹ · s¹ resistance and 0.08 l/cmH₂O compliance at a TI of 1.0 s (Table 1). This is consistent with the findings of the delayed inspiratory termination shown both in patients with COPD (10, 14) and in a mechanical lung model study (17) in which the ventilators with peak flows of 5 or 25% as the termination criterion were used [the Siemens 900C (10), Taema Cesar or Hamilton Amadeus (14), and Siemens 300 (17)]. In another study on COPD patients, Jubran and co-workers (10) showed that 5 of their 12 studied patients displayed expiratory effort before the cessation of inspiratory flow (i.e., delayed termination). These five patients had an average time constant of 0.54 s compared with an average time constant of 0.38 s in the patients who displayed no expiratory effort during the inspiratory phase. The delayed termination at the long time constant can be explained because the inspiratory flow decay after the peak level is slower as a result of a longer time constant. The strong relationship between TI/_peak and  may also explain why expiratory synchrony can be achieved with the ventilator (Newport E200) in which expiratory trigger sensitivity is related to the elapsed inspiratory time (17).

The direct effects of the Pps level and the magnitude of the patient inspiratory effort (Pmus) on TI/_peak are not as important as those of /TI (Fig. 3, Table 1); however, the higher the Pps/Pmus max, the narrower the synchronized zone becomes with a certain flow criterion (Fig. 4). This means that the possibility of expiratory synchrony is reduced if a high level of pressure support is applied and/or the patient has a weak inspiratory effort (19). When the patient Pmus is predominant (compared with the ventilator support) in generating inspiratory flow, the ventilator is predisposed to be in synchrony with the patient in terms of the inspiratory termination and vice versa. These results are consistent with the findings in the above-mentioned study done by Jubran and co-workers (10) because the delayed termination existed at PSV of 20 cmH₂O but disappeared at PSV of 10 and 5 cmH₂O.

It is surprising to find that, even under the adult respiratory mechanics, TI/_peak can vary in a very wide range, from 1 to 85%. This finding explains why expiratory asynchrony occurs frequently when mechanical ventilators with a fixed level of flow termination criteria are used. Contrary to a fixed level of flow termination criteria, user-selectable expiratory trigger sensitivity in some of the most recently released ventilators (e.g., Hamilton Galileo and Nellcor Puritan Bennett 840) provides flexibility, allowing clinicians to manually select the flow termination criteria to achieve good patient-ventilator synchrony. However, because TI/_peak may change when any of the four parameters (Pps, Pmus, , and TI) change, clinicians may need to readjust the expiratory trigger sensitivity setting very frequently. This is obviously unrealistic in routine clinical practices. Because expiratory asynchrony is a clinical concern primarily in adult applications (9, 10, 14, 16) and TI/_peak has an excellent linear correlation with /TI in the range of adult mechanics (with correlation coefficient 0.96, Table 2), the adjustment of expiratory trigger synchrony for the purpose of patient-ventilator synchrony, in theory, could easily be done automatically by the current computer technologies in mechanical ventilator applications.

Because of the assumptions and simplifications that we used in this mathematical model study, the results presented here may not be directly extrapolated to clinical application. It is especially true in terms of the data of expiratory synchrony with the representative termination criteria (Fig. 4), since our model did not take into account the role of the pressure criteria in the inspiratory termination. The divisions of the premature, synchronized, and delayed termination zones in Fig. 4 were based on the assumption that the ventilator has only flow criteria to terminate the inspiratory flow. Many mechanical ventilators, in reality, are also equipped with pressure criteria as a backup to terminate the inspiration. By pressure criteria, the ventilator flow is terminated when Paw rises a certain amount (e.g., +1.5, 2.0, 3.0, and 20.0 cmH₂O in the Nellcor Puritan Bennett 7200ae, the Newport E200, the Siemens 900, and the Siemens 300 ventilators, respectively) above the Pps level. In fact, pressure criteria could become a primary method for terminating the ventilator flow in some ventilators (17) if they are strict (i.e., the ventilator is cycled off at a small supraplateau pressure). The addition of a strict pressure criterion may fundamentally narrow the delayed termination zone and widen the synchronized zone, especially in patients with active expiratory efforts (at the cost of the patient expiratory work). It should be kept in mind, however, that a strict pressure criterion may cause other problems, such as premature termination of inspiration in the case of pressure variation at the plateau level.

Contribution to Already Published Work

Potential Applications of This Study