Aerosol delivery via a mechanical ventilator remains unregulated with no standards for drug delivery to intubated patients. Bench models predicting drug delivery have not been validated in vivo. For modern ventilator designs, we chose to identify, on the bench, the most important variables affecting aerosol delivery and to correlate in vitro predictions of aerosol delivery with in vivo end points independent of patient response. Test aerosols of albuterol and antibiotics were compared. Bench measurements of inhaled mass (percentage of nebulizer charge, mean ± SEM) ranged from 5.7 ± 0.5% to 37.4 ± 1.6%, with breath-actuated nebulization and humidity identified as the most important factors determining aerosol delivery. In patients, sputum levels of deposited antibiotics varied from 1.10 to 19.6 μg/ml/mg. Variation in sputum levels correlated with predictions from the in vitro model. Aerosol delivery in ventilated patients can be efficient and reproducible only if defined ventilator parameters are tightly controlled. Key parameters can be determined via in vitro bench testing defining delivery standards for clinical trials of drugs with narrow therapeutic/toxicity ratios.
Modern ventilator design does not include standards that relate to aerosol delivery. Bench models predicting nebulized drug delivery during mechanical ventilation have not been validated in vivo. Previous studies have measured patient- and ventilator-related factors affecting aerosol generation and inhalation for nebulizers on the bench (1–4). A few studies have measured deposition of nebulized drugs in ventilated patients, but they did not relate actual deposition to bench predictions (5–7).
Besides the variables already studied in vitro (e.g., humidity, nebulizer type), newer ventilator designs may further complicate predictions of drug delivery. For example, the use of constant flow in the ventilator tubing (e.g., bias flow) during all phases of ventilation may increase aerosol losses. Indeed, adult ventilator systems are becoming similar in design to neonatal ventilators, which are known to be inefficient in aerosol delivery (8). Breath-actuated nebulization, an important factor in spontaneously breathing patients (9), is not a feature of all modern ventilators.
In vivo effects of conventional aerosols can be difficult to relate to aerosol delivery and deposition. For example, responsiveness to bronchodilators may vary between patients, and in the absence of a deposition measurement, changes in airway resistance will not differentiate between patient-related factors and differences in drug delivery.
The purpose of our study was to detail the most important variables affecting aerosol delivery via nebulizer on the bench for modern ventilators employing the newer flow regimes with and without breath actuation and the effects of humidity. For the major variables defined in vitro, predictions of aerosol delivery were correlated to measured levels of antibiotic in suctioned sputum from intubated patients. Sputum levels were chosen as an in vivo end point to the present study because unlike bronchodilators they are independent of patient responsiveness. Some of the results of these studies have been previously reported in the form of an abstract (10).
The bench model is diagramed in Figure 1
. The test ventilator was connected to a test lung (M.I.I. VentAid TTL; Michigan Instruments, Inc., Grand Rapids, MI) via an endotracheal tube (inside diameter = 8.0 mm). Aerosols were sampled just distal to the endotracheal tube with an inhaled mass filter (Pari GmbH, Starnberg, Germany) and a filter in the expiratory line. Aerosols were generated by nebulization with the device located in the inspiratory line 12 inches from the Y piece.For the in vitro study, we decided to use albuterol (2.5 mg in 3 ml of normal saline) labeled with technetium as the test solution, as it was previously characterized in our laboratory. Activity of albuterol aerosol has been shown to correlate with technetium when measured by cascade impaction (11). This allowed for a comparison with previous studies defining important parameters influencing aerosol generation and delivery (1, 2, 12).
The breathing pattern was fixed at a tidal volume of 750 ml, a respiratory rate of 15, a peak flow of 70 L/minute, an inspiratory time of 0.9 seconds, and an inspiratory:expiratory ratio of 1:3.4. For the T-Bird ventilator, bias flow was set at 10, 15, and 20 L/minute.
Ventilatory parameters were monitored with the Bicore, Pulmonary Monitor CP 100 (VIASYS Healthcare, Critical Care, Conshohocken, PA). For experiments using humidification, a Hudson RCI Concha III humidifier (Hudson Respiratory Care Incorporated, Temecula, CA) was set at 35°C. This device was turned off and bypassed for experiments without added humidity.
Aerosol particle distribution was sampled via a cascade impactor (GS 1 IMPAQ; California Measurements, Inc., Sierra Madre, CA) with the device located distal to the endotracheal tube. Aerosols were sampled over a 4-minute period. Radioactivity on the cascade stages was measured by a collimated ratemeter (Ludlum Measurements Inc., Sweetwater, TX) and the distribution plotted on log probability paper. Activity at the median defined the mass median aerodynamic diameter (MMAD).
Aerosol production was quantified by measuring radioactivity on the filters via a well counter (CRC-10; Capintec, Inc., Montvale, NJ), and all activity (filters plus nebulizer) was summed in a “mass balance” designed to trace aerosol losses.
The experimental configurations are outlined on Figure 2
. There were multiple configurations/combinations that were designed to test major variables. These included the ventilator type, nebulizer brand, form of nebulizer activation (continuous versus breath actuation), and presence of humidification and magnitude of bias flow. Some devices could not perform all functions; for example, the Evita ventilator does not have adjustable bias flow, and the T-Bird ventilator does not have nebulizer breath actuation.Ventilators were chosen based on properties that may affect aerosol delivery.
PB 7200 (Puritan Bennett, Pleasanton, CA): breath-actuated nebulization, no bias flow, previously tested in our laboratory
Evita 4 (Drager Inc., Critical Care Systems, Telford, PA): newer, breath-actuated nebulization
T-Bird (VIASYS Healthcare, Inc., Conshohocken, PA); no breath actuation, mandatory bias flow
The nebulizers that were tested included the AeroTechII Aerosol Delivery System (CIS-US, Bedford, MA), which was previously tested in our laboratory, and the Portex Small Volume Medication Nebulizer Kit (SIMS Portex, Inc., Fort Myers, FL), which is a nebulizer that was used for routine aerosol therapy in our hospital. Nebulizers, run to dryness, were driven directly by the ventilator (breath actuation) or continuously via compressed air or wall oxygen at a flow of 8 L/minute (continuous nebulization).
The variables evaluated for each configuration were as follows:
Humidified versus nonhumidified ventilator circuit (heated wire circuits were not used)
Breath-actuated nebulization versus continuous nebulization
Bias flow set at flow rates of 10, 15, and 20 L/minute
Nebulizer brand
Parameters measured for each of the variables evaluated were as follows:
Inhaled mass (%), the amount of drug on the filter as a percentage of the nebulizer charge
Mass balance (percentage of recovery), the sum of both filters plus remnant activity in the nebulizer
MMAD
Data obtained from the configurations listed in Figure 2 were combined by the test parameter when reported in the results. For example, “breath-actuated humidification” included data from all devices that satisfied that specific criteria.
Based on our bench data, the effects of major parameters predicting aerosol delivery were tested in vivo by measuring sputum levels of aerosolized antibiotics sampled from intubated patients. Patients were participating in a parallel protocol designed to measure the effectiveness of aerosolized antibiotics. The routine consent for the institutional review board–approved antibiotic protocol was modified to allow variation in the mode and conditions of nebulization, that is, breath actuated or continuous as well as humidified or nonhumidified. Institutional review board approval was obtained for these modifications. For these parameters, paired data for each patient were obtained for sputum suctioned from the proximal airways 1 hour after aerosol therapy with an antibiotic. The protocol incorporated a standardized suction routine (13) and avoidance of instilled saline. Patients were treated with Gentamicin, Amikacin, or Vancomycin via the AeroTech II nebulizer (the clinical “dose” or nebulizer charge was 80 mg for Gentamicin, 400 mg for Amikacin, and 120 mg for Vancomycin). Sputum was sampled after ventilator parameters were set for 24 hours (humidification was maintained throughout except as noted later here). The antibiotic treatment regimen consisted of aerosolized antibiotics given three times using an every 8 hours regimen. It is important to note that for the “nonhumidified” treatment regimen, the humidifier was turned off and bypassed only during the actual nebulizer treatment (approximately 1 hour). After the 24-hour test period, sputum was obtained for the 1-hour period between 1 and 2 hours after an aerosol treatment. The sample was weighed, and the volume was standardized with normal saline to 4.0 ml and centrifuged at 40,000 rpm for 60 minutes at 4°C. The supernatant phase was diluted 1:100 for Gentamicin and Vancomycin and 1:1,000 for Amikacin to allow analysis within the reference range of the assay (Roche Integra; Roche Diagnostics, Somerville, NJ). Results are reported in micrograms per milliliter sputum per milligram of drug placed in the nebulizer.
Results are reported as mean ± SE. The Mann-Whitney and unpaired t tests were used in the statistical analysis of the in vitro data. In vivo data were analyzed as pairs using the Wilcoxin signed rank test and paired t tests (Stat View 4.5; Abacus, Inc., Berkley, CA). When the n was small, the nonparametric p values were reported.
Inhaled Mass % | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
Nebulization Mode | Nonhumidified | n | Humidified | n | NH/H | p Value | ||||
Breath-actuated nebulization† | 37.4 ± 1.6 | 8 | 9.6 ± 1.0 | 19 | 3.84 | < 0.0001 | ||||
Continuous nebulization‡ | 10.4 ± 0.8 | 21 | 5.7 ± 0.5 | 17 | 1.81 | < 0.0001 | ||||
All ventilators | 17.9 ± 2.4 | 29 | 7.7 ± 0.7 | 36 | 2.09 | < 0.0001 |
The effect of bias flow on inhaled mass percentage is shown in Table 2
Inhaled Mass % | |||||||
---|---|---|---|---|---|---|---|
Bias Flow (L/min) | Nonhumidified | n | Humidified | n | |||
10 | 10.0 ± 0.7 | 7 | 5.3 ± 0.4 | 9 | |||
15 | 8.3 ± 1.0 | 12 | 5.3 ± 0.5 | 15 | |||
20 | 8.6 ± 0.7 | 10 | 3.8 ± 0.8 | 5 |
Aerosol delivery was not a strong function of the brand of nebulizer. As shown in Table 3
Inhaled Mass % | ||||||||
---|---|---|---|---|---|---|---|---|
Ventilator Mode | AeroTech II | n | Portex | n | p Value | |||
Nonhumidified | 20.8 ± 3.0 | 19 | 12.3 ± 3.6 | 10 | 0.0929 | |||
Humidified | 10.7 ± 1.3 | 13 | 6.1 ± 0.5 | 23 | 0.0003 |
The aerodynamic behavior of the particles is summarized in Table 4
MMAD (μm) | ||||||||
---|---|---|---|---|---|---|---|---|
Ventilator Mode | AeroTech II | n | Portex | n | p Value | |||
Nonhumidified | 1.2 ± 0.1 | 16 | 1.9 ± 0.2 | 8 | 0.0065 | |||
Humidified | 2.3 ± 0.3 | 13 | 2.2 ± 0.2 | 8 | 0.6435 |
Tubing losses with added humidity were also suggested by the recovery data. When the humidifier was turned off and bypassed, the sum of filters plus residual nebulizer activity averaged 88.2 ± 0.9% (n = 29). With humidification, recovery dropped to 73.9 ± 1.8% (n = 36).
In summary, under typical clinical conditions of aerosol delivery, for example, continuous nebulization in a humidified circuit, only 5.7 ± 0.5% of the nebulizer charge was delivered. On the other hand, an optimized system (breath-actuated nebulization, nonhumidified) delivered 37.4 ± 1.6% to the filter (Table 1).
Six patients were studied. Diagnoses, ventilators, settings, endotracheal and tracheostomy tube sizes, and peak airway pressures are listed in Table 5
Ventilator | Sputum Levels (μg/ml) | Sputum Levels (μg/ml/mg) | PIP | |||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Case No. | Diagnosis | Type | Mode | Setting | Tube
I.D. | Antibiotic | No. of
Pairs | Humidified | Nonhumidified | Humidified | Nonhumidified | Before | After | |||||
1 | Respiratory failure/ Quadraplegia | T-Bird* | AC | 12 500 40% | 8.0† | Amikacin | 6 | 263 ± 46 | 553 ± 108 | 0.66 ± 0.12 | 1.38 ± 0.27 | 26 | 28 | |||||
2 | Pneumonia | T-Bird* | AC | 10 600 50% | 8.0† | Gentamicin | 4 | 86 ± 10 | 200 ± 34 | 1.08 ± 0.12 | 2.51 ± 0.43 | 29 | 29 | |||||
3 | Severe COPD/ Pneumonia | Bear 3 | IMV | 6 700 35% | 8.0† | Amikacin | 4 | 1,777 ± 519 | 5,790 ± 2,140 | 4.44 ± 1.30 | 14.48 ± 5.35 | 28 | 30 | |||||
4 | Pneumonia | PB-7200 | AC | 14 700 50% | 7.5 | Gentamicin | 4 | 133 ± 21 | 734 ± 99 | 1.66 ± 0.27 | 9.18 ± 1.24 | 30 | 30 | |||||
5 | Pneumonia | PB-7200 | AC | 15 450 40% | 7.5 | Vancomycin | 1 | 132 | 2,352 | 1.10 | 19.60 | 24 | 24 | |||||
6 | Pancreatitis | PB-7200 | PC | 18 450 30% | 7.5 | Amikacin | 5 | 1,576 ± 280 | 4,947 ± 941 | 3.94 ± 0.70 | 12.37 ± 2.35 | 36 | 36 |
Sputum levels (μg/ml/mg) | |||||||
---|---|---|---|---|---|---|---|
Nebulizer Activation | n | Nonhumidified | Humidified | NH/H | p Value | ||
Breath actuation | 14 | 12.6 ± 1.8 | 3.2 ± 0.5 | 3.89 | < 0.001 | ||
Continuous | 10 | 1.8 ± 0.3 | 0.8 ± 0.1 | 2.20 | 0.0005 | ||
All ventilators | 24 | 8.1 ± 1.5 | 2.2 ± 0.4 | 3.63 | 0.0002 |
When compared with the bench data, there were similarities as well as systematic differences between the predicted aerosol delivery (Table 1) and in vivo sputum levels (Table 6). The relative effects of humidity predicted on the bench were reflected in the in vivo levels of antibiotics (predicted ratios of nonhumidified to humidified values breath actuation and continuous nebulization 3.84 and 1.81; ratio of nonhumidified to humidified values observed 3.89 and 2.20, respectively). However, breath-actuated nebulization compared with continuous nebulization was more effective in raising sputum levels in vivo than predicted on the bench. In the bench study, depending on humidification, inhaled mass was 1.7 to 3.6 times higher for breath actuation compared with continuous nebulization (Table 1). In vivo levels of sputum antibiotics were four to seven times higher (Table 6), indicating that other in vivo factors were important (see Discussion).
The present study illustrates the major factors affecting aerosol delivery to intubated patients via conventional nebulization. The bench model predicted that breath-actuated nebulization and humidification would be the major determinants of drug delivery. In patients, after inhalation of antibiotics, sputum values ranged over an order of magnitude, but much of this variability was explained when the data were normalized for breath actuation and humidification. Sputum levels of deposited antibiotics provided a direct index of drug delivery (as opposed to an indirect index such as a change in airway resistance). When reported as absolute values, antibiotic levels were high (e.g., a mean gentamicin level of 12.6 μg per ml of sputum per mg of drug multiplied by 80 mg = 1,008 μg per ml of sputum) if the humidifier was turned off and bypassed. These results are consistent with previously reported values (5). However, if a clinical response were dependent on drug level, the simple addition of humidification would reduce the average gentamicin level to 258 μg per ml of sputum.
The recovery data coupled with the changes seen in MMAD suggest hygroscopic growth of particles and losses via tubing impaction in the presence of humidification of inspired gases. The humidifier supersaturates the ventilator gas and promotes rainout as the humid air cools. Aerosol particles can serve as nuclei and losses are enhanced.
Early studies from our group suggested that nebulizer type may be important (12, 14), but later bench studies showed that running the nebulizer to the point of sputtering (loosely termed “dryness” in the literature) reduced device-related variability (2). The present protocol ran the devices to dryness (run time of approximately 60 minutes). When the data were corrected for breath actuation and humidification, the AeroTechII was more efficient than the Portex device (Table 3).
The bench model predicted that breath-actuated nebulization and humidification would have the greatest effects in vivo. Bench predictions for humidity effects appeared to be the most accurate. However, the bench model seemed to underestimate the effects of breath-actuated nebulization. As presented in the Results, the in vivo levels of antibiotic between breath actuated and continuous nebulization were approximately twice as high as predicted. We believe that these observations can be explained by the differences in the concepts of inhaled mass versus actual aerosol deposition. In a bench model, it is possible to measure variables that affect the quantity of aerosol presented to the patient and “inhaled,” that is, the “inhaled mass.” Once the particles are inhaled, a certain fraction of the inhaled particles are deposited (the “deposition fraction”), and sites of deposition are determined by the breathing pattern, airway geometry, and the MMAD. Although we did not measure actual deposition in our patients, we did control many of the parameters that affect the deposition of inhaled particles. For example, all of our in vivo experimental conditions were paired, except for changes in nebulizer actuation and humidification. In our clinical study using aerosolized antibiotics, only the AeroTechII nebulizer was used. A comparison of the data between Tables 1 and 6 suggests that for breath-actuated nebulization, either the deposition fraction was increased or particle deposition was more central than during treatment with continuous nebulization. A parameter linked to increases in both deposition fraction and central deposition is the MMAD. To isolate this factor, we returned to our bench studies. Table 7
MMAD | MMAD | |||||||
---|---|---|---|---|---|---|---|---|
Mode | n | Nonhumidified | p Value | n | Humidified | p Value | ||
Breath actuated | 9 | 1.5 ± 0.1 | 7 | 3.0 ± 0.2 | ||||
0.0006 | 0.0177 | |||||||
Continuous | 7 | 0.9 ± 0.1 | 6 | 1.6 ± 0.5 |
Our findings have important implications for clinical trials. Future studies will be needed to define the safety and efficacy of a given drug and mode of delivery (such as using a nonhumidified circuit). For drugs such as antibiotics, control of dose requires strict control of the aerosol delivery protocol. Failure to specify the ventilator type, the presence or absence of humidity, and/or breath actuation will prevent control of drug dosing and may affect the assessment of clinical effects.
For commonly used drugs (bronchodilators), the lowest doses delivered in this study exceed those given in clinical trials (16), and thus, potent safe drugs should still be effective in most patients even under conditions that promote inefficient aerosol delivery (e.g., continuous nebulization and added humidification); however, the potential remains that with efficient delivery toxic levels can be achieved and with inefficient delivery some patients (e.g., those with severe asthma) may be undertreated.
In clinical situations different from those studied with our model, there may be important differences in drug delivery. Aerosol delivery is notoriously difficult during neonatal ventilation primarily because of high bias flow (8). In adults, the use of other modes of ventilation may have an impact on delivery. If important to the investigator, bench studies under clinically relevant conditions will help in understanding and controlling potential confounding factors.
In conclusion, bench models can determine factors important in aerosol delivery to intubated patients. Clinical trials of aerosolized drugs require strict control of the ventilator and conditions of nebulization if the dose to the patient is important in assessing clinical response.
The authors thank Drager Inc., Critical Care Systems, Telford, PA, for the loan of the Evita 4 ventilator.
1. | Diot P, Morra L, Smaldone GC. Albuterol delivery in a model of mechanical ventilation: comparison of metered-dose inhaler and nebulizer efficiency. Am J Respir Crit Care Med 1995;152:1391–1394. |
2. | O'Riordan TG, Palmer LB, Smaldone GC. Aerosol deposition in mechanically ventilated patients: optimizing nebulizer delivery. Am J Respir Crit Care Med 1994;150:1474–1475. |
3. | O'Doherty MJ, Thomas SHL, Page CJ, Treacher DF, Nunan TO. Delivery of a nebulized aerosol to a lung model during mechanical ventilation: effect of ventilator settings and nebulizer type, position, and volume of fill. Am Rev Respir Dis 1992;146:383–388. |
4. | Dhand R, Tobin MJ. Inhaled bronchodilator therapy in mechanically ventilated patients. Am J Respir Crit Care Med 1997;156:3–10. |
5. | Palmer LB, Smaldone GC, Simon SR, O'Riordan TG, Cuccia A. Aerosolized antibiotics in mechanically ventilated patients: delivery and response. Crit Care Med 1998;26:31–39. |
6. | Thomas SH, O'Doherty MJ, Fidler HM, Page CJ, Treacher DF, Nunan TO. Pulmonary deposition of a nebulised aerosol during mechanical ventilation. Thorax 1993;48:154–159. |
7. | Fuller HD, Dolovich MB, Posmituck G, Wong Pack W, Newhouse MT. Pressurized aerosol verses jet aerosol delivery to mechanically ventilated patients. Am Rev Respir Dis 1990;141:440–444. |
8. | O'Riordan TG, Kleinman LI, Hughes K, Smaldone GC. Predicting aerosol deposition during neonatal ventilation: feasibility of bench testing. Respir Care 1994;39:1162–1168. |
9. | Smaldone GC. Enhanced in vitro delivery of budesonide via continuous and breath-actuated nebulization. Eur Respir J 2000;16:540s. |
10. | Miller D, Smith S, Hughes K, Smaldone GC. Drug delivery via nebulizer during mechanical ventilation: an update. Am J Respir Crit Care Med 2002;165:A377. |
11. | McPeck M, Tandon R, Hughes K, Smaldone GC. Aerosol delivery during continuous nebulization. Chest 1997;111:1200–1205. |
12. | O'Riordan T, Greco MJ, Perry RJ, Smaldone GC. Nebulizer function during mechanical ventilation. Am J Respir Crit Care Med 1992;145:1117–1122. |
13. | Palmer LB, Smaldone GC, O'Riordan TG, Morra L. Tracheal aspirates in chronically mechanically ventilated patients: a human model of gram-negative infection and airway inflammation. Chest 1995;108:1326–1332. |
14. | McPeck M, O'Riordan TG, Smaldone GC. Choice of mechanical ventilator: influence on nebulizer performance. Respir Care 1993;38:887–895. |
15. | Ilowite JS, Gorvoy JD, Smaldone GC. Quantitative deposition of aerosolized gentamicin in cystic fibrosis. Am Rev Respir Dis 1987;136:1445–1449. |
16. | Dhand R, Tobin MJ. Bronchodilator delivery with metered dose inhalers in mechanically ventilated patients. Eur Respir J 1996;9:585–595. |