Overcoming the Adverse Effect of Humidity in Aerosol Delivery via Pressurized Metered-Dose Inhalers during Mechanical Ventilation

Overcoming the Adverse Effect of Humidity in Aerosol Delivery via Pressurized Metered-Dose Inhalers during Mechanical Ventilation CARLOS F. LANGE and WARREN H. FINLAY Aerosol Research Laboratory, Department of Mechanical Engineering, University of Alberta, Edmonton, Alberta, Canada The well-known problem of reduced drug delivery that occurs when heated, humid air is used with pressurized metered-dose inhalers (pmdis) and spacers in intubated settings is carefully studied with Airomir using an in vitro model under controlled conditions of temperature and humidity. A better understanding of the physical processes leading to the aforementioned drop in performance is obtained, and a method is devised to circumvent the problem without having to reduce the temperature or humidity of the ventilator circuit. The present study shows that the mole fraction of water vapor in the ventilation air (and not the temperature) is the major factor behind the sharp drop in the amount of drug delivered to the lung. However, the presence of water vapor does not affect performance because of hygroscopic growth. Instead, it influences the initial atomization process and the early stages of aerosol generation. Removal of these negative effects can be achieved by using a larger spacer that allows longer times for the aerosol to evaporate, as is demonstrated in the present study. The use of pressurized metered-dose inhalers (pmdis) for delivery of aerosolized drugs during mechanical ventilation has been shown to be useful for intubated adults (1, 2), and children (3). However, the efficient use of pmdis depends on a series of factors, which if not carefully considered may reduce the effectiveness of drug administration (4). A controversy that appears to have been settled in the last few years is the importance of the use of a spacer or in-line chamber as opposed to in-line pmdi adaptors (5 7). Several other factors, such as ventilator settings, type of spacer, and spacer position in the ventilator circuit, have been investigated with the aim of maximizing drug delivery to the lung. However, one of the most adverse of all these factors is the humidity and heat in the ventilator circuit. In particular, air delivered to mechanically ventilated patients is heated and humidified to prevent damage of the airway mucosa. The conditioning of the air, however, is responsible for dramatic reductions in drug delivery to ventilated patients. Reductions of 40 to 50% in lung deposition have been reported with in vitro adult models (5, 8, 9), whereas reductions of up to 85% were found in pediatric models (7, 10, 11), when normal ventilator conditions (35 to 37 C and 98% relative humidity []) were compared with dry ambient conditions (25 to 27 C and 10% ). The aim of the present study was to investigate in vitro the causes for the reduction of lung deposition of pmdi aerosols in heated and humidified ventilator circuits, as well as to develop a potential solution to this problem without altering the ventilator settings or circuit. Because the largest reductions were reported for pediatric ventilator circuits, the settings for (Received in original form September 8, 1999 and in revised form November 22, 1999) Correspondence and requests for reprints should be addressed to Warren H. Finlay, Aerosol Research Laboratory, Department of Mechanical Engineering, University of Alberta, Edmonton, AB, T6G 2G8 Canada. E-mail: warren.finlay@ualberta.ca Am J Respir Crit Care Med Vol 161. pp 1614 1618, 2000 Internet address: www.atsjournals.org the in vitro model were chosen to correspond to the ventilation of an infant. METHODS Experimental Setup A model of a ventilator circuit was constructed according to the scheme shown in Figure 1. A large climatized chamber (0.67 m 3 ) was employed to control temperature and humidity of the circuit air in the model. The importance of keeping the cascade impactor in thermal and hygroscopical equilibrium with the conditioned air used in the ventilator circuit was demonstrated by Finlay and Stapleton in a previous study (12). The inspiratory air was drawn from the chamber by insulated tubing connected through a one-way valve to the distal end of the in-line spacer (Figure 1, ➆). Two different spacers were investigated. The bulk of the experiments were performed with a commercially available device (Aerochamber MV; Trudell Medical International, London, ON, Canada) with an approximate length of 11 cm and a diameter of 4.5 cm. Two experiments at the highest temperature (37 C) were performed with a larger in-house built prototype with a length of 29 cm and diameter of 10 cm. The proximal end of the spacer was connected to a Y-piece attached to a pediatric endotracheal tube (ETT) (length 20 cm and internal diameter 4 mm; Mallinckrodt Medical, St. Louis, MO). The Y-piece was positioned directly at a small opening in the wall of the climatized chamber and it was verified that insulation of the spacer and the Y-piece was not necessary, as it had no measurable influence on the results. The other limb of the Y-piece was the exhalation line, consisting of a one-way valve and a bacterial/viral filter (Respirgard, Model 303; Marquest Medical Products, Englewood, CO) open to ambient air. The other end of the ETT was connected through a T-junction to a cascade impactor (Andersen 1 ACFM Non-Viable, 8 stages; Graseby Anderson, Smyrna, GA) for particle sizing. In order to maintain the constant flow rate, 28.3 L/min, required in the cascade impactor and the alternating flow rate, Q, required in the ventilator circuit (comprised of the spacer and ETT), an artificial lung machine was used in conjunction with a pump and a compressor, as shown in Figure 1. The vacuum pump ➈ ensured a constant flow rate of 28.3 L/min through the cascade impactor. During inhalation in the ventilator circuit, the artificial lung ➄ injected a constant complementary flow rate of 28.3 Q L/min. At the same time, the three-way valve ➁ blocked the line, sending the compressor flow back into the climatized chamber. As a result, the desired inhalation flow rate Q of conditioned air was drawn through the ventilator circuit. In the second half of the breathing cycle, the three-way valve ➁ was automatically opened, feeding the compressor flow of 56.6 L/min into the system, while the artificial lung was draining Q 28.3 L/min from the system. Because the cascade impactor continued to use 28.3 L/min, a flow rate of Q remained to be exhaled through the ventilator circuit over the exhalation filter. All flow rates were set and checked by means of a mass flowmeter (Model MF300; Omega, Stamford, CT). The system settings used throughout the present investigation corresponded to a flow rate of Q 4.8 L/min (square wave) and a tidal volume of VT 75 ml (32 breaths/min), simulating the ventilation of an infant of 6 to 8 mo age. The inhalation:exhalation ratio was 1:1, a usual choice in pediatric clinical practice and also recommended by Fink and coworkers (9). There was no end-inspiratory pause, which

Lange and Finlay: Humidity Effect on pmdi Aerosols 1615 repeated five times (n 5) and values are given in the form mean SD. A nonlinear least-squares fit of the cumulative distribution of the deposition in the stages of the cascade impactor was used to determine the mass median aerodynamic diameter (MMAD) and the geometric standard deviation (GSD) of the aerosol. The total mass deposited in the impactor was considered the inhaled mass. The exhaled mass corresponded to the amount present in the ETT during exhalation. Statistical comparisons were done using analysis of variance (ANOVA) with Tukey multiple means comparisons and multiple linear regressions (14) (SYSTAT, Evanston, IL). Figure 1. Scheme of the ventilator model. Inside the climatized chamber: ➀ compressor (56.6 L/min); ➁ three-way valve triggered by the artificial lung; ➂ cascade impactor; ➃ ETT. Outside the chamber; ➄ artificial lung (alternating 28.3 Q and Q 28.3 L/min); ➅ Y-piece connector; ➆ insulated inspiratory limb with air supplied by the climatized chamber plus one-way valve and in-line spacer (Q and Q L/min); ➇ expiratory limb with one-way valve and filter (Q L/min); ➈ vacuum pump (28.3 L/min). Mouloudi and coworkers (13) found to have little influence in the case of mechanically ventilated patients. and humidity of the air flowing through the ventilator circuit and of the supplementary air for the cascade impactor were controlled by the climatized chamber (Hotpack, Waterloo, ON, Canada). The chamber s large internal volume and heating/cooling capacity were able to keep the temperature of the system within 2 C of the nominal value during the entire experimental run. Experiments were performed at 25 C, 30 C, and 37 C. The minimum value of at each temperature was limited by the ambient humidity (15% at 25 C, or 8% at 37 C). The value stayed constant in the cases of low humidity (ambient humidity) or was kept within 5% of the saturation value ( 100%) in the humid cases. For the only case of intermediate (50%), at 37 C, larger oscillations of up to 10% occurred during the experiment. Besides the chamber s own sensors, the temperature and humidity were monitored with a portable thermohygrometer (Hanna Instruments, Limena, Italy) with accuracy of 2.0% in and 0.4 C in temperature. Preceding each new temperature and setting, the system was allowed to run for at least 30 min to achieve equilibrium. At the beginning of an experiment, the pmdi canister containing salbutamol sulfate (Airomir, 100 g/puff; 3M Canada, London, ON, Canada) was first primed by firing it into ambient air. After vigorous shaking of the canister, the pmdi was actuated once into the spacer, coinciding with the beginning of an inspiratory cycle. This procedure was repeated at 1-min intervals to a total of 10 actuations for each experiment. Deposited Mass Assessment and Particle Size Distribution After each experiment, the amounts of salbutamol sulfate deposited in the ventilator circuit parts and in the cascade impactor were assayed. The ETT, Y-piece, exhalation filter, and the plates of the cascade impactor were washed each with 5 ml of distilled water. The spacers together with the top of the pmdi canister were washed with 25 ml of distilled water. The solutions were submitted to ultraviolet (UV) measurements at 224 nm (absorbance). Salbutamol sulfate (Sigma Chemicals, St. Louis, MO) was used as a standard for calibration. Results of the assay are expressed in percentages of the total nominal mass delivered to the system (1 mg). Each experiment was RESULTS The deposition results with the Aerochamber spacer for the various combinations of temperature and are summarized in Table 1. The inhaled mass varied significantly (p 0.01) between 19.1% and 41.1% of the administered dose. The large SD occurring for some cases corresponds to instances where the largest oscillations occurred in the chamber s and temperature during the experiments. It is noteworthy that at ambient temperature (25 C) the inhaled mass did not vary significantly (p 0.1) between the relatively dry ambient humidity and the complete saturation of the ventilator air. On the other hand, a significant reduction (p 0.01) of more than 50% in the inhaled mass was observed at 37 C between dry and saturated conditions. When the temperature of the ventilator air was increased while keeping the absolute concentration of water vapor at ambient levels (i.e., mole fraction of water vapor stayed constant, but the value of decreased), a significant improvement (p 0.01) of 25% in the inhaled mass was observed. However, when water vapor was added in order to keep the air saturated while increasing its temperature to 37 C, the inhaled mass dropped significantly (p 0.01) by more than 33% of its saturated ambient temperature value. A clearer picture of this apparently contradictory behavior is obtained by relating the inhaled mass to the mole fraction of water vapor in air (p 0.01 in multiple linear regression with temperature), instead of the relative humidity. The values for the mole fraction were obtained from the Clausius-Clapeyron equation (15). A two-dimensional contour plot of lines representing constant values of the inhaled mass based on the values of Table 1 can be seen in Figure 2. This figure reveals that for constant values of water vapor mole fraction there is almost always an increase in the inhaled mass with temperature. Conversely, there is always a decrease in the inhaled mass with increasing mole fraction and this decrease becomes more pronounced at higher temperatures. TABLE 1 DRUG DEPOSITION IN THE VENTILATOR CIRCUIT MODEL WITH AEROCHAMBER SPACER Mole Fraction* Spacer Deposition (mean SD) ETT Inhaled Exhaled 25 15 0.005 54.8 1.9 7.7 1.1 32.7 1.6 5.4 0.8 25 100 0.030 56.3 2.5 8.3 1.1 28.8 1.4 6.3 0.9 30 11 0.005 52.1 6.2 4.4 1.4 38.7 2.2 4.6 1.2 30 100 0.040 58.6 7.6 14.9 7.0 22.3 2.5 4.6 1.0 37 8 0.005 49.8 1.8 4.6 0.9 41.1 1.2 5.9 1.8 37 50 0.030 58.7 7.0 6.0 1.9 30.3 5.4 3.7 1.4 37 100 0.058 66.8 3.2 7.9 3.1 19.1 2.1 6.9 2.0 * Mole fraction of water vapor in air. ETT Y-piece. Cascade impactor.

1616 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 161 2000 TABLE 3 PARTICLE SIZE DISTRIBUTION Distribution (mean SD) MMAD ( m) GSD Figure 2. Inhaled mass as a function of the temperature and the water vapor mole fraction in the air (derived from the ). 25 15 2.34 0.10 1.53 0.05 25 100 2.57 0.14 1.80 0.09 30 11 2.46 0.06 1.50 0.06 30 100 2.52 0.13 1.70 0.15 37 8 2.42 0.03 1.51 0.04 37 50 2.48 0.08 1.50 0.03 37 100 2.49 0.23 2.03 0.27 37* 8 2.49 0.08 1.48 0.04 37* 100 2.53 0.10 1.46 0.05 * Prototype spacer. Only the limit cases of heated dry or humid ventilator air were run with the prototype spacer. The corresponding results are shown in Table 2. The amount of inhaled mass was insensitive to the mole fraction of water vapor in the air and was similar to the value obtained for heated dry air with the Aerochamber (last case in Table 1). Table 3 shows particle size data for the different cases. No significant differences were found in MMAD or GSD (p 0.1) for any combination of cases. DISCUSSION Previous in vitro studies have not looked into the separate role of temperature and humidity in pmdi aerosol delivery during mechanical ventilation. In the particular case of pediatric ventilation, Garner and coworkers (11) measured a delivered amount of 5.3% using a 4-mm ETT and warm, humid air (35 C), as opposed to 20.3% with warm, dry air. The same group repeated the study later with somewhat different experimental setup (10), this time comparing warm, humid air conditions (34.3 C and 100% ) with cooler, dry ventilation air (27.1 C and 8.6% ). In accordance with the present results, Garner and coworkers (10) found a smaller difference in the delivered amount in their second study: 2.5% with warm, humid air and 7.3% in cooler, dry air. The lower inhaled mass estimates obtained in these two former studies compared with the present results is probably related to the differences in the experimental setup and breathing pattern. For instance, both of the above previous studies included a flow sensor and a 90 elbow adaptor in-line between the spacer and the ETT, and in the second study (10) an orifice-plate resistor was installed before the collection filter. Additionally, the spacers were installed between the Y-piece and the ETT, causing particles that remained suspended in the spacer to be exhaled. All of these differences would be expected to reduce the assessed inhaled TABLE 2 DRUG DEPOSITION IN THE VENTILATOR CIRCUIT MODEL WITH PROTOTYPE SPACER Mole Fraction* Spacer Deposition (mean SD) ETT Inhaled Exhaled 37 8 0.005 51.3 4.5 5.1 0.9 42.9 3.7 0.7 0.4 37 100 0.058 52.4 2.8 5.5 0.8 42.2 1.7 0.9 0.5 * Mole fraction of water vapor in air. ETT Y-piece. Cascade impactor. amount compared with the present setup. In the more recent study of Wildhaber and coworkers (7), the spacer was also installed between the Y-piece and the ETT. Among the various cases they investigated, the settings corresponding to the ventilation of a 4-kg child with dry air at ambient temperature came closest to the setup of the present work. The sum of inhaled (14.3%) and exhaled (23.0%) amounts obtained by Wildhaber and coworkers (7) agrees well with the corresponding sum in the present study (32.7% 5.4%), with the difference in partitioning between inhalation and exhalation largely due to the difference in position of the spacer in the circuit (i.e., distal to the Y-piece in our study, but proximal in Wildhaber and coworkers [7]). Because of concerns surrounding experiments with children and infants, there is little in vivo data on this field. Fok and coworkers (3) compared aerosol delivery from pmdis and nebulizers to ventilated and nonventilated infants and found the lung deposition in all cases to be extremely low, around 1%. These discrepancies between in vivo and in vitro experiments are caused mainly by limitations and simplifications of the artificial model. For example, the present model used a simplified breathing pattern and lacked real lung exhalation, because all particles that entered the cascade impactor were trapped. The exhalation filter collected only the particles that were suspended in the ETT at the onset of exhalation, which explains the relatively low exhalation values obtained in the present study. However, as Fink and coworkers (9) recently pointed out, there is still a correspondence between in vitro and in vivo results and tendencies, making the in vitro findings still useful for guiding clinical practice. Several explanations have been suggested for the drop in dose delivery in heated and humidified ventilator circuits. The most common explanation suggests that hygroscopic growth of the pmdi particles occurs in the humid air flow and results in increased deposition by impaction in the spacer and ETT (2, 7, 8, 11). However, this suggestion does not explain why even nonhygroscopic drugs, such as salbutamol base used in chlorofluorocarbon (CFC) pmdis, suffer from the same problem of reduced delivery (11). In addition, aqueous drugs administered through pmdis, such as in the present study, usually have hydrophobic surfactants that make them relatively nonhygroscopic (16). In a slightly different version of this hypothesis, some investigators (5, 9) consider that the sharp drop in temperature, experienced by the particles owing to propellant evaporation, would cause the particles to act as condensation nuclei. As a consequence, the particles mass would grow, as if they were hygroscopic. However, the extremely rapid evaporation of

Lange and Finlay: Humidity Effect on pmdi Aerosols 1617 propellant (Stefan flow) very likely prevents condensed water from coating the droplets surface or keeps it at negligible amounts. Moreover, the fact that the inhaled mass did not change significantly with humidity at ambient temperature (Table 1), as well as the practically constant size distribution measured in the present study (Table 3) and by Kim and coworkers (17) make this hypothesis implausible. Another hypothesis raised by a group of researchers (10, 11) considers the increase in the air viscosity with temperature to be responsible for the decrease in inhaled mass, because of more deposition in the spacer. However, the viscosity increase would cause less impaction in the spacer, not more. Thus, instead this would partly explain the increase in inhaled mass at a higher temperature for constant water vapor mole fraction. Nonetheless, the dynamic viscosity of air increases by only 2% between 25 C and 37 C and so does not fully explain the effect of temperature on inhaled mass, and does not allow any explanation of the effect of water vapor mole fraction. One effect that probably contributes substantially to the 25% increase in inhaled mass between 25 C and 37 C with dry air is the faster evaporation of the propellant at elevated temperatures. In particular, at a higher temperature the air can deliver energy more rapidly to the droplets for the evaporation of the propellant. The less time it takes for the droplets to evaporate the propellant, the sooner they achieve their final size and avoid impacting in the spacer. This is the probable explanation for the increase in inhaled mass seen in Figure 2 at constant mole fractions but increasing temperature. This temperature effect is small, though, compared with the effect of changes in water vapor mole fraction. Indeed, as seen in Figure 2, increases in water vapor mole fraction can more than counter the benefits of a higher air temperature. On the basis of the present results two hypotheses can be considered that explain this effect. The first is that the presence of water molecules in the air acts on the droplets surface to reduce the vapor pressure of propellant and consequently the droplets evaporation rate. The other hypothesis is that the surface tension of the drug/propellant suspension may be increased by the presence of water molecules in the air, resulting in larger initial droplets in the early stages of atomization. The hypothesis of a decreased evaporation rate of propellant was also suggested by Dhand and Tobin (6) and by Fink and coworkers (8, 9) as a possible explanation for the larger losses in the ventilator line. Kim and coworkers (17) studied the size distribution of various drugs delivered by pmdis in dry and humid environments. They estimated that the initial size, immediately after atomization, of a pmdi droplet can range from 20 to 50 m. These initial droplets are composed of more than 99% highly volatile propellant (hydrofluoroalkane [HFA] in the case of Airomir), which will evaporate very rapidly. After complete evaporation of the propellant, Kim and colleagues measured distributions with MMAD between 2.4 and 5.5 m. The Stokes number, which is the nondimensional quantity that describes the ability of a particle to keep its own trajectory (and ultimately leads to impaction), is directly proportional to the droplet diameter squared. This means that the Stokes number (and, by consequence, the probability of impaction) decreases approximately by a factor of 80 from the initial to the final droplet state in a pmdi aerosol. If the droplets do not have a free, straight path to travel during the evaporation of the propellant, the large Stokes number prior to complete evaporation indicates that many particles will impact and be lost. However, if they are able to evaporate before encountering any walls or changes in flow direction they can become small enough to brake in the air and be carried toward the lung without impaction. This is a major reason why spacer devices should be used: to give the pmdi aerosol time, i.e. space, to evaporate without hitting any wall. But existing spacers were mainly designed for nonventilated patients (see, e.g., Corr and coworkers [18]) and later on found to be advantageous also to ventilated patients (5, 7), although always falling short of the efficiency obtained in the nonventilated cases. If the hypothesis of slower evaporation proves true, then the pmdi aerosol simply needs more time ( space) in a warm and humid ventilator circuit in order to deliver drugs as efficiently as in non-ventilated patients. The results from Farr and coworkers (19) provide a possible explanation for this hypothesis. Farr and coworkers investigated the formation of liposomes in the lung after the inhalation of aerosolized phospholipids. They found that the nonvolatile phospholipids tended to form a coherent barrier at the propellant/air interface, which retarded the evaporation of propellant by interfering with the acquisition of heat for evaporation from the surrounding air. In the present study, it is conceivable that the presence of water vapor in the air may cause interaction with surfactant molecules at the surface of the droplet, forming what Farr and coworkers called a coherent barrier. In agreement with the present results, such an effect would be proportional to the number of water molecules in the air (mole fraction) rather than to the relative humidity of the air. The second hypothesis, namely an increase in the droplet s surface tension at higher mole fractions of water vapor in air, bears some similarity with the first hypothesis, because such an increase would cause the initial droplet sizes to be larger (i.e., the droplets would contain more propellant), and consequently need longer to evaporate, even at normal evaporation rates. The aerosol would again just need more time and space to evaporate completely and allow for efficient drug delivery. Because salbutamol sulfate is a suspension, the final size of the droplets is not correlated to the amount of propellant initially present. After complete evaporation, the size distribution of the aerosol will show no trace of the larger initial size. Unfortunately, particle sizing at short distances of the pmdi nozzle is very difficult, because the differing refractive indices of the air and the propellant distort measurements with optical methods, such as laser diffraction (20). Dunbar and coworkers (21) were able to obtain sizing data of CFC-propelled aerosols as close as 25 mm from the nozzle with phase-doppler particle analysis. But their measurements were performed at ambient conditions only and the sizes measured were already approximately 10 m, meaning that most of the propellant had already evaporated. An indirect way to determine the occurrence of larger initial droplet sizes would be the observation of a solution pmdi under dry and humid conditions. In a solution, the amount of drug in the droplet is in constant relation to the amount of propellant, so that a larger initial droplet, after complete evaporation, would result in a larger final particle. In their size comparison of several pmdi-generated aerosols under dry and humid conditions, Kim and coworkers (17) reported at least one drug of the solution type (Bronkometer; Breon Laboratories, New York, NY). They found a MMAD of 5.5 0.1 m in the case of dry air ( 1%) and of 5.7 0.1 m in the case of humid air ( 90%). This small difference of 4% in the MMAD practically discards this second hypothesis, at least in the case of the CFC-propellant used in the Bronkometer. Spacer Design Revisited Even being unable to confirm or discard definitely at this point either of the two considered hypotheses, the results of the present investigation make it possible to devise a solution

1618 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 161 2000 for this critical drawback in the pmdi aerosol delivery to ventilated patients. If either one of the hypotheses is true, or even if a combination of both applies, it is not necessary to avoid or bypass humidification of the ventilation air in order to obtain higher drug delivery efficiency. Instead, it suffices to design and optimize spacers for this specific application that give the aerosol droplets enough space to evaporate completely. Ideally, it would even be possible to surpass the delivery efficiency of the nonventilated case, because of the increased air viscosity and heat transfer rate at higher temperatures, as discussed earlier. To test this idea, two experiments using warm air (37 C) with low (8%) and high (100%) were repeated. This time, a large prototype was used as spacer device. The dimensions of the prototype were chosen based on a high-speed photographic analysis of the pmdi cloud (22). The results shown in Table 2 confirmed the prediction of an inhaled mass quantity similar to the amount obtained with heated and dry air with the Aerochamber (see Table 1, 37 C and 8%). There was initially concern about the large volume of the prototype spacer (approximately 2.3 L) compared with the small pediatric tidal volume simulated (75 ml). But the present results dismissed this concern. Apparently, the aerosol particles, upon evaporation of the propellant, remain suspended long enough to be carried past the ETT into the lung by the fast sequence of breathing cycles typical for pediatric ventilation. In the pursuit of an ideal spacer, one should bear in mind that the actual distance needed by droplets to evaporate will depend on several factors, such as type of drug and propellant and size of actuator orifice, among others (9, 23). Conclusions The effect of heat and humidity on a pmdi aerosol in a mechanical ventilation circuit was investigated. In particular, an explanation for the well-known drop in drug delivery efficiency in the heated and humid ventilator conditions was sought. By varying the conditions of temperature and humidity of the ventilation air, it was found in the present study that the decrease in inhaled drug mass was directly related to the mole fraction of water vapor in the ventilation air, rather than to the relative humidity or to the temperature. Based on the results of the present investigation, several hypotheses explaining the analyzed phenomenon were tested. Among others, it was shown that hygroscopic growth was not responsible for the excessive losses in the spacer and ETT. The most likely hypothesis is that of an interaction between the water molecules in the air and the surfactant present in the propellant/ drug suspension, causing a reduction in the evaporation rate of propellant. It was also shown that the adverse effect of the water vapor mole fraction can be circumvented without any change in the ventilator settings by using a larger spacer that allows longer times for the aerosol to evaporate. In fact, the present study shows that drug delivery in a ventilator with a correctly dimensioned spacer can be even more efficient than under normal ambient conditions. Acknowledgment : The laboratory help of H. Orszanska is gratefully acknowledged. References 1. Fuller, H. D., M. B. Dolovich, G. Posmituck, W. W. Pack, and M. T. Newhouse. 1990. Pressurized aerosol versus jet aerosol delivery to mechanically ventilated patients. Am. Rev. Respir. Dis. 141:440 444. 2. Fuller, H. D., M. B. Dolovich, C. Chambers, and M. T. Newhouse. 1992. Aerosol delivery during mechanical ventilation: a predictive in-vitro lung model. J. Aerosol Med. 5:251 259. 3. Fok, T. F., S. Monkman, M. Dolovich, S. Gray, G. Coates, B. Paes, F. Rashid, M. Newhouse, and H. Kirpalani. 1996. Efficiency of aerosol medication delivery from a metered dose inhaler versus jet nebulizer in infants with bronchopulmonary dysplasia. Pediatr. Pulmonol. 21: 301 309. 4. Newhouse, M. T., and H. D. Fuller. 1993. Rose is a rose is a rose? Aerosol therapy in ventilated patients: nebulizers versus metered dose inhalers a continuing controversy. Am. Rev. Respir. Dis. 148:1444 1446. 5. Diot, P., L. Morra, and G. C. Smaldone. 1995. Albuterol delivery in a model of mechanical ventilation. Am. J. Respir. Crit. Care Med. 152: 1391 1394. 6. Dhand, R., and M. J. Tobin. 1996. Bronchodilator delivery with metered-dose inhalers in mechanically-ventilated patients. Eur. Respir. J. 9:585 595. 7. Wildhaber, J. H., M. J. Hayden, N. D. Dore, S. G. Devadason, and P. N. LeSouëf. 1998. Salbutamol delivery from a hydrofluoralkane pressurized metered-dose inhaler in pediatric ventilator circuits. Chest 113: 186 191. 8. Fink, J. B., R. Dhand, A. G. Duarte, J. W. Jenne, and M. J. Tobin. 1996. Aerosol delivery from a metered-dose inhaler during mechanical ventilation. Am. J. Respir. Crit. Care Med. 154:382 387. 9. Fink, J. B., R. Dhand, J. Grychowsky, P. J. Fahey, and M. J. Tobin. 1996. Reconciling in vitro and in vivo measurements of aerosol delivery from a metered-dose inhaler during mechanical ventilation and defining efficiency-enhancing factors. Am. J. Respir. Crit. Care Med. 159: 63 68. 10. Garner, S. S., D. B. Wiest, and J. W. Bradley. 1994. Albuterol delivery by metered-dose inhaler with a pediatric mechanical ventilatory circuit model. Pharmacotherapy 14:210 214. 11. Garner, S. S., D. B. Wiest, J. W. Bradley, B. A. Lesher, and D. M. Habib. 1996. Albuterol delivery by metered-dose inhaler in a mechanically ventilated pediatric model. Crit. Care Med. 24:870 874. 12. Finlay, W. H., and K. W. Stapleton. 1999. Undersizing of droplets from a vented nebulizer caused by aerosol heating during transit through an Anderson impactor. J. Aerosol Sci. 30:105 109. 13. Mouloudi, E., K. Katsanoulas, M. Anastasaki, E. Askitopoulou, and D. Georgopoulos. 1998. Bronchodilator delivery by metered-dose inhaler in mechanically ventilated COPD patients: influence of end-inspiratory pause. Eur. Respir. J. 12:165 169. 14. Wilkinson, L., M. Hill, and E. Vang. 1992. SYSTAT: Statistics, Version 5.2 ed. SYSTAT, Inc., Evanston, IL. 15. Reid, R. C., J. M. Prausnitz, and B. E. Poling. 1987. The Properties of Gases and Liquids, 4th ed. McGraw-Hill, New York. 16. Saunders, P. A. 1979. Handbook of Aerosol Technology, 2nd ed. Van Nostrand Reinhold, New York. 294 295. 17. Kim, C. S., D. Trujillo, and M. A. Sackner. 1985. Size aspects of metereddose inhaler aerosols. Am. Rev. Respir. Dis. 132:137 142. 18. Corr, D., M. Dolovich, D. McCormack, R. Rufin, G. Obminski, M. Newhouse. 1982. Design and characteristics of a portable breathing actuated, particle size selective medical aerosol inhaler. J. Aerosol Sci. 13:1 7. 19. Farr, S. J., I. W. Kellaway, and B. Carman-Meakin. 1987. Assessing the potential of aerosol-generated liposomes from pressurized pack formulations. J. Contr. Rel. 5:119 127. 20. Mitchell, J. P., and M. W. Nagel. 1997. Medical aerosols: techniques for particle size evaluation. Part. Sci. Technol. 15(3 4, special issue):217 241. 21. Dunbar, C. A., A. P. Watkins, and J. F. Miller. 1997. An experimental investigation of the spray issued from a pmdi using laser diagnostic techniques. J. Aerosol Med. 10:351 368. 22. Dhand, R., S. K. Malik, M. Balakrishnan, and S. R. Verma. 1988. High speed photographic analysis of aerosols produced by metered dose inhalers. J. Pharm. Pharmacol. 40:429 430. 23. Ahrens, R., C. Lux, T. Bahl, and S.-H. Han. 1995. Choosing the metereddose inhaler spacer or holding chamber that matches the patient s need: evidence that the specific drug being delivered is an important consideration. J. Allergy Clin. Immunol. 96:288 294.