Introduction

Aerosol therapy, i.e., the delivery of medication particles carried by inhaled gases, constitutes the cornerstone of chronic broncho-dilatory and anti-inflammatory therapy for patients suffering from asthma and chronic obstructive pulmonary disease. It is associated with improved long-term patient-centered outcomes [1–3]. Similarly, antibiotic aerosol therapy has proven effective to treat lung infection in patients suffering from cystic fibrosis [4].

In the acute setting, particularly in the critically ill patients, evaluation of patient-centered outcomes is lacking. Nevertheless, a large body of work has evaluated optimal implementation of aerosol therapy in patients undergoing artificial ventilation in terms of practicability and safety, and has shown significant physiologic efficacy of several inhaled drugs in this setting [5–11]. Significant reductions in respiratory system resistance of ventilated patients have been demonstrated after delivery of bronchodilator using various nebulizer and metered dose inhaler (MDI) set-ups [6, 12, 13]. In ventilator-associated pneumonia, optimized nebulization set-ups such as a low inspiratory peak flow, increased inspiratory time, interrupted humidification and nebulizer placement upstream in the inspiratory limb seem to deliver inhaled antibiotics effectively to treat lung infections [14–17]. Large-scale international studies on ventilatory support have not recorded data about aerosol therapy [18, 19]. In a previous study using an e-mail self-administered survey, we obtained responses from 854 physicians who declared being confident in aerosol therapy efficacy and using it frequently in critically ill patients [20]. In this previous study, knowledge appeared very heterogeneous [20]. A Scandinavian observational study reported the use of aerosol therapy in 50 % of 186 ventilated patients (mainly beta-2-adrenergic receptor agonists), without providing data about implementation modalities [21]. This lack of large-scale prospective data hampers optimal knowledge translation towards the clinical setting and optimal research and educational resources allocation. The aim of the present work was to assess the frequency, modalities and short-term safety of aerosol therapy in critically ill patients either breathing spontaneously or undergoing invasive or noninvasive (NIV) artificial ventilation.

Methods

This prospective cross-sectional point prevalence study was carried out over 14 days in 81 intensive care units in 22 countries (see the list of centers and investigators in the Appendix). Centers were recruited on a voluntary basis among participants of the aforementioned e-mail survey by purposive sampling through e-mail contact of members of the European Society of Intensive Care Medicine, French and Spanish intensive care societies (Société de Réanimation de Langue Française, Revista Electrónica de Medicina Intensiva) and members of the REVA network (Réseau Européen de recherche en Ventilation Artificielle) [20]. The study was approved by the ethics commission of the French intensive care society and additional ethical approval gained at each participating institution if legally required. Given the non-interventional study design, the need for written informed consent was waived by those independent commissions. All patients or their next of kin were informed about the study with the possibility to decline participation. The 2-week participation periods for each unit were staggered over March and April 2013.

All patients present in the unit during the study period and not declining participation were included. Each day, patients’ ventilator statuses were prospectively recorded: (1) “invasive artificial ventilation”: patient breathing or ventilated through a tracheal tube or tracheostomy; (2) “NIV”: patient who underwent at least one NIV session (including continuous positive airway pressure) but no “invasive artificial ventilation”; and (3) “spontaneous breathing” otherwise. Each time a patient received inhaled medication during the study period (aerosol therapy, but also instillation of drugs in the tracheal tube, except 0.9 % sodium chloride instillation for tracheal suctioning), extensive data were recorded (see electronic supplement Tables 1, 2 and 3 for an extensive list of recorded variables). Investigators were invited to report any significant adverse event without specific a priori definition.

Data were entered into a web-based database (ClinInfo, Lyon, France) and analyses performed using R 2.14.1 (R Foundation for Statistical Computing, Vienna, Austria). Quantitative variables were expressed as mean ± standard deviation and compared with Student’s t test, except in cases of non-Gaussian distribution [median (25th, 75th percentile)]. Qualitative variables were expressed as counts (%) and compared between groups using the Chi-square test. The 95 % confidence interval (CI95) of proportions was calculated for the main variables of aerosol therapy (no missing value, no data imputation). A p value lower than 0.05 was considered significant.

Results

A total of 2808 patients were included (Table 1), predominantly in intensive care units [10,689 (81 %) vs. 2514 (19 %) patient-days in intermediate care], and 9714 inhaled drug administrations were recorded. Follow-up was complete; participating countries and centers are detailed in the electronic supplement (Table 4).

Table 1 Characteristics of patients receiving or not aerosol therapy
Full size table

Frequency of aerosol therapy

A total of 678 patients (24 % CI95 22–26 %) received at least one inhaled medication over the 2-week period [median number of 7 (2, 18) per patient], while only 271 patients (10 %) were taking inhaled medications chronically at home. Frequency of aerosol therapy was heterogeneous between centers (range 0–57 % of patients; see electronic supplement Table 4). Aerosol-generating devices and patients’ ventilation status during aerosol therapy are detailed in Table 2. Overall, aerosols were mainly delivered either to patients breathing spontaneously (n = 4832 aerosols, 50 %) or into the ventilator circuit of intubated patients (n = 4532, 47 %), representing two distinct clinical and therapeutic situations. Aerosols under NIV represented only 3 % of all aerosols.

Table 2 Characteristics of aerosols
Full size table

Spontaneously breathing patients

Among 4832 aerosols performed in patients breathing spontaneously, jet nebulizers were used predominantly (n = 3388, 70 %), followed by MDIs (n = 790, 16 %).

NIV

Among 305 patients who underwent one or several days of NIV, 149 (49 % CI95 40–57 %) received at least one aerosol on such days. Aerosols were predominantly delivered when patients were breathing spontaneously inbetween NIV sessions (n = 1057 aerosols, i.e. 75 % of aerosols in patients undergoing NIV) and infrequently directly into the ventilatory circuit (n = 350 aerosols, i.e. 25 % of aerosols in patients undergoing NIV). Among 1057 aerosols delivered inbetween NIV sessions, only 171 aerosols (16 %) specifically triggered NIV interruption in order to deliver the inhaled therapy.

Intubated patients

Among 1215 patients who underwent invasive artificial ventilation, 262 (22 % CI95 20–24 %) received at least one aerosol while intubated. Aerosols delivered during artificial ventilation were mostly delivered in patients intubated and ventilated with a two-limb ventilatory circuit (n = 4499, 92 %) (Fig. 1; Table 3). Bronchodilators and corticosteroids were mainly delivered using nebulizers (n = 2264 bronchodilator aerosols, 63 %; n = 355 corticosteroid aerosols, 69 %). In intubated patients, antibiotics were delivered using jet, ultrasonic and vibrating mesh nebulizers in 221 (62 %), 105 (29 %) and 31 (9 %) cases, respectively. Ventilator settings were changed for administration of 107 anti-infectious aerosols (30 %) as compared to only 74 (2 %) of bronchodilator aerosols (p < 0.01). Similarly, when using a heated humidifier, the device was turned off for 119 (59 %) anti-infectious aerosols as compared to 249 (15 %) of bronchodilator aerosols (p < 0.01). Placement of the nebulizer upstream in the inspiratory limb at a distance from the Y piece remained infrequent even for administration of anti-infectious aerosols (n = 33, 9 % of anti-infectious aerosols) (Fig. 1). Among 1867 aerosols delivered using a jet nebulizer, a ventilator integrated breath-actuated jet nebulization system was available in 1115 cases (60 %); when available, it was used for nearly all cases (n = 1109, 99 %). Placement of a filter on the expiratory limb to protect the ventilator was done for 2997 (66 %) aerosol administrations; this filter was infrequently changed in relation to nebulization (Fig. 1).

Fig. 1
figure 1

Main determinants of aerosol set-ups used in intubated patients. In intubated patients, aerosol therapy was predominantly performed using jet nebulizers placed close to the Y piece while ventilator settings were left unchanged

Full size image
Table 3 Aerosol therapy in intubated patients
Full size table

Drugs delivered

Drugs were frequently delivered as a combination (n = 4131 aerosols, 42 %; Table 2). This mainly concerned association of a short acting beta-2-adrenergic agonist and an anticholinergic drug (n = 2317, 56 % of combined aerosols). Bronchodilators (n = 7960 aerosols) represented 82 % of administrations and concerned 89 % of patients receiving aerosols (Table 4). Corticosteroids were the second most frequent inhaled drugs (n = 1233, i.e. 13 % of aerosols and 26 % of patients receiving aerosols). Together, bronchodilators and corticosteroids represented 88 % of aerosols. These drugs were delivered far beyond the patients suffering chronic obstructive pulmonary diseases or asthma, who accounted for 312 patients among the 626 receiving bronchodilators and/or corticosteroids (50 %). Indeed, in a majority of cases, bronchodilator and corticosteroid aerosols were delivered to treat exacerbation of COPD, acute asthma or acute bronchospasm of another origin (n = 2204, 51 % of aerosols with only one molecule), but various other heterogeneous indications were observed such as infection (n = 579, 13 %) or wheezing of undetermined origin (n = 293, 7 %) (see Table 5 of the electronic supplement).

Table 4 Drugs delivered as aerosols
Full size table

A total of 509 anti-infectious aerosols were recorded, predominantly colistin (n = 400, 79 % of anti-infectious aerosols) and amikacin (n = 49, 10 %). Anti-infectious aerosols were primarily indicated to treat nosocomial pneumonia (n = 342, 67 %) and to a lesser extent tracheobronchitis/bronchial colonization (n = 94, 19 %). Prophylactic anti-infectious aerosols accounted for a smaller proportion (n = 31, 6 %). Overall, anti-infectious aerosols concerned 31 patients (1 %) in 14 centers (17 %).

Side effects

A total of 106 administrations (<1 %) prompted notification of a side effect, mainly tachycardia and arterial hypertension (n = 39), arterial hypotension (n = 16), hypoxemia (n = 20) and cough (n = 23). Bronchospasm was reported three times (colistin nebulization in all cases).

Discussion

The main results of this large-scale prospective international cross-sectional prevalence study is that aerosol therapy is used in one-fourth of critically ill patients and in every fifth intubated patient, confirming smaller-scale observations and declarative data [20, 21]. Aerosol therapy appeared even more frequent in patients undergoing NIV, as half of those patients received aerosols, mainly inbetween ventilation sessions. Bronchodilators and corticosteroids were the overwhelmingly predominant drugs delivered as aerosols (88 %); anti-infectious aerosols, even though representing a smaller proportion of aerosols (5 %), were frequently recorded over the 14-day study period and almost exclusively delivered to treat nosocomial infections; only 3 % of aerosols were mucus-modulating drugs. Albeit only a limited number of side effects were recorded in the present study, the high prevalence of aerosol therapy observed raises questions about the optimization of technical implementation and long-term safety in the critical care setting.

Spontaneous breathing

The predominant use of nebulizers to deliver aerosols in critically ill patients is in accordance with guidelines addressing aerosol therapy for severe asthma and chronic obstructive pulmonary disease exacerbation in the emergency department as proper use of MDIs may be difficult for those patients [11].

NIV

Interestingly, about a quarter of aerosols delivered to patients breathing spontaneously concerned patients otherwise undergoing NIV. This may suggest poor knowledge translation given existing data on the efficacy of inhaled bronchodilators delivered within NIV circuits [23–26]. Conversely, one may hypothesize that clinicians and nursing staff consider aerosol delivery into ventilator circuits too cumbersome, thus calling for progress in equipment simplification.

Intubated patients

Safety and efficacy issues may be discussed based on the current literature (briefly summarized in the electronic supplement Table 6) [5–17]. Regarding safety, the predominant use of nebulizers to deliver bronchodilators and corticosteroids in ventilated patients seems intriguing, as they are available as MDIs. In fact, as aerosols were predominantly delivered using jet nebulizers, with breath-actuated ventilator integrated systems frequently unavailable, about every fourth aerosol administration exposed intubated patients to uncontrolled tidal volumes (the jet nebulizer being supplied by an external gas source) [27]. The use of MDIs, when available, might be preferred. Actually, only about 9 % of bronchodilator and/or corticosteroids aerosols in intubated patients were delivered with a MDI connected to an inhalation chamber, whereas this simple technique is the one with the most extensively evaluated efficacy [5–13]. The second important safety issue relates to particles cleared through the expiratory limb, which may interfere with the proper function of the ventilator expiratory block, particularly when nebulizing antibiotics or performing continuous nebulization [10]. One-third of aerosols (n = 1502) were administered in intubated patients with no filter protecting the expiratory block. No dysfunction was documented over the 2-week study period, in part due to the predominant delivery of bronchodilators and corticosteroids; nevertheless, given the very severe complications reported, including pneumothorax and cardiac arrest, additional educational efforts are warranted in order to promote better practice [10, 14, 28–30].

Regarding efficacy, unlike for bronchodilator therapy, nebulization/ventilation set-up is a key factor for success of inhaled anti-infectious therapy, in particular when aiming to treat pneumonia, which was the case for 73 % of anti-infective aerosol deliveries [31]. Indeed, delivering inhaled antibiotics to the infected, poorly aerated, distal alveolar compartment of intubated patients may be challenging [32]. In this regard, jet nebulizer, the most frequently used type of nebulizer for antibiotic administration, is well known for a high residual volume (amount of drug which remains in the nebulizer at the end of nebulization) as compared to vibrating mesh and ultrasonic nebulizers [6]. This may influence aerosol therapy efficacy. Lu et al. observed that nebulizing 400 mg of colistimethate using a mesh nebulizer enabled the treatment of nosocomial pneumonia, while the same dose placed in a jet nebulizer results in a much lower dose of drug actually deposited in the patient [14]. Similarly, Palmer et al. reported positive results nebulizing aminoglycosides and/or vancomycin using a breath-actuated jet nebulizer in patients suffering nosocomial tracheobronchitis or pneumonia [16, 17]. Again, the common practice observed in the present study, consisting in placing the nebulizer at the Y piece (Fig. 1), may be counterproductive by favoring aerosol loss in the expiratory limb and preventing the replication of favorable results in daily clinical practice [14–17, 27, 33]. Such dose/nebulizer issues may, in part, explain some discrepancies among studies evaluating the potential benefit of inhaled antibiotics to treat multidrug-resistant lung infections [34]. Furthermore, unlike in the aforementioned prospective interventional studies, ventilator settings were left unchanged and heated humidifiers kept active during, respectively, 70 and 40 % of anti-infectious aerosols recorded [14–17].

While anti-infective aerosols concerned a limited number of patients (1 %), bronchodilators and corticosteroids were extensively delivered (every fifth critically ill patient). Beyond bronchodilation, unlike in the outpatient setting, no long-term patient-centered outcomes have been evaluated in critically ill patients [2, 3, 5–13, 35]. Given potential side effects, one may question the value of their large use, far beyond the population of patients receiving it at home and with obstructive pulmonary disease, with indications such as infections which may need specific evaluation [36].

Study limits

Beyond capturing only a low incidence of side effects not defined a priori, which may be underestimated, the study design restricting observation on two consecutive weeks did not enable the capture of seasonal variations in practice. Aerosol therapy may be more frequent in the winter months due to increased respiratory infections. Albeit including a high number of centers in several countries on all continents, the international scope of the study was damped by the predominance of European centers, especially in France and Spain, and by the absence of North American centers. Thus, results cannot be extrapolated worldwide. Interestingly, some practice heterogeneity was observed (see electronic supplement Table 4) calling for additional evaluation in regions not covered by the present work. Specific case mix within each center, not captured by the present study, may in part explain the observed aerosol therapy practices. Centers participated in the study on a voluntary basis, and one cannot exclude a bias towards more experienced or expert units, physicians’ knowledge being not assessed in this study. As aerosol efficacy was not evaluated in the present study, observed practice can only be put into perspective with existing knowledge and recommendations, without drawing conclusions about the efficacy of aerosol therapy in individual patients. Similarly, staff protection from potential aerosol toxicity and the types of NIV interfaces were not recorded in this study. Finally, given the non-interventional design, one cannot exclude that the study by itself induced some changes in aerosol therapy practice during the observation period.

Conclusions

Aerosol therapy is a common practice concerning a fifth to a quarter of intensive care and intermediate care patients despite the lack of proven benefit on patients centered outcome. The frequent implementation of aerosol therapy during invasive artificial ventilation seemed suboptimal in a significant number of cases and almost never performed during NIV, calling for actions on the educational level such as issuing guidelines specifically dedicated to aerosol therapy in critically ill patients.