Summary of findings
Description of the condition
Acute respiratory failure is common amongst patients who are hospitalized with an acute exacerbation of their chronic lung disease. Where optimal medical treatment has failed to relieve symptoms, ventilatory support is recommended (Rodriguez-Roisin 2006). Despite advances in non-invasive ventilation strategies (Brochard 1995; Plant 2000), a significant proportion of patients still require invasive ventilation to treat their acute exacerbation. In addition to invasive ventilation, inhaled bronchodilators are an essential component of the treatment and management of this patient group (NICE 2004). Short acting beta2-agonists and ipratropium are widely used to manage symptoms associated with acute exacerbations and are recommended by international guidelines (GOLD 2011).
Description of the intervention
Bronchodilator therapy aims to resolve bronchoconstriction, decrease the work of breathing, potentially relieve dyspnoea (Dhand 2005) and is frequently administered to mechanically ventilated patients(Boucher 1990). Bronchodilators for mechanically ventilated patients may be administered systemically by intravenous infusion, or directly to the lungs through the inhalation of an aerosol (Georgopoulos 2000). There are currently two main methods of delivering aerosol bronchodilation which have been adapted for use in patients receiving mechanical ventilation, the nebulizer and the metered dose inhaler (MDI). Nebulizers deliver bronchodilators to the lower respiratory tract by converting the liquid drug into smaller particle droplets which can then be inhaled. The production of an aerosol may be achieved through the use of compressed gas, ultrasonic sound frequencies, or a vibrating mesh or plate (Dhand 2006a). MDIs contain a pressurized mixture of active drug, surfactants, preservatives and propellants. An aerosol is generated through the actuation of the device, which results in a high speed release of the suspension from the MDI (Jantz 1999). Aerosol delivery offers several advantages over the systemic route, namely painless delivery of the drug directly to the site of action, rapid onset of drug effect, and the resultant reduction in dosage requirements (Dhand 2004; Fink 1999a). As a result, aerosol inhalation is globally recognized as the preferred route of delivery for bronchodilators in chronic lung diseases (GOLD 2011).
Various pharmacological agents with differing modes of action can be deployed for bronchodilation but their overall effect, relaxation of the bronchial smooth muscle, is congruent (Dhand 2006a). Currently, beta2-agonists, anticholinergics and methylxanthines make up the three main pharmacologic classes of agents used for bronchodilation. Methylxanthines can only be administered via enteral or parenteral routes, whereas beta2-agonists and anticholinergics are most frequently utilized through inhalation (BNF 2009) and will therefore be the focus of this review.
Several narrative reviews have attempted to address the issue of which is the most appropriate and effective route of administration of bronchodilator therapy for adult patients receiving mechanical ventilation. Current guidelines endorse either mode of delivery.
The suggested advantages of MDIs have been identified as ease of administration, increased reliability in dosing, cost effectiveness including personnel time to administer the drug, and freedom from contamination risk (Dhand 1996; Dhand 2006a; Dhand 2007a; Fink 1999a; Hess 1991; Hess 2002). Several reviews have concluded that no apparent advantage exists for either MDI or nebulizer if appropriate administration techniques and dose are utilized (Coleman 1996; Dhand 2004; Dhand 2007b; Dhand 2008; Guerin 2008; Jantz 1999; O'Doherty 1997), although the high dose of bronchodilators that is needed for nebulizer delivery may be associated with a higher degree of cardiovascular instability (Dolovich 2005).
How the intervention might work
The success of any aerosol bronchodilation therapy is dependent on satisfactory amounts of active drug reaching the bronchial tree(Dolovich 2005). Aerosol deposition is known to be affected by a number of factors, with specific considerations associated with patients receiving mechanical ventilation that are not present in the ambulatory demographic. These include ventilator, circuit, drug and patient related factors(Dhand 2004). Device related factors are also present, with choice of equipment, position in the ventilator circuit, and timing of drug delivery affecting both nebulizers and MDIs (Fink 1999a).
The efficacy of aerosol drug delivery from nebulizers and MDIs has been shown to be variable in patients receiving mechanical ventilation. Evidence suggests that performance variability is present both in different models of nebulizers (Loffert 1994) and between individual units of the same model (Alvine 1992). The efficacy of bronchodilator delivery from an MDI is also variable, dependent on timing actuation with inspiration (Crogan 1989; Dhand 2003) and rates of inspiratory flow (Fink 1999b). The use of nebulizers for bronchodilator delivery may lead to hypoventilation in mechanically ventilated patients when using older ventilator models (Beaty 1989).
Multi-centre survey data on bronchodilator administration practices in mechanically ventilated neonates highlight variations in practice, with 19% of respondent institutions using MDIs at all times and 43% using nebulizers exclusively (Ballard 2002). Such figures for the adult patient demographic are not available.
Why it is important to do this review
To date, there has not been an international systematic review to determine which method of aerosol bronchodilator delivery system, nebulizer or MDI, is more effective in mechanically ventilated adult patients. This review will therefore attempt to determine which is the most effective delivery system in terms of physiological response and patient outcomes.
To compare nebulizers to MDIs for bronchodilator delivery for invasively ventilated critically ill adult patients in terms of physiological response and patient outcomes. Subgroup analyses were planned according to other ventilation and bronchodilation strategies, ventilator settings and administration variables.
Criteria for considering studies for this review
Types of studies
We included randomized controlled trials (RCTs), including randomized cross-over trials where the order of the intervention was randomized, comparing the nebulizer and MDI for aerosol bronchodilation in mechanically ventilated adult patients.
Types of participants
We included adult patients (as defined by the trialists) receiving invasive mechanical ventilation in critical care units. If no definition was available, we assumed that the participants were adults unless identified as paediatric patients in the studies.
Types of interventions
We excluded studies in which aerosol bronchodilation agents were delivered via the same MDI or nebulizer device simultaneously with another drug group. Combination administration of bronchodilators of differing drug groups (for example beta2-agonists and anticholinergics) was allowed. We excluded any studies in which bronchodilator agents were administered by any route other than aerosol. Other ventilation and bronchodilation strategies such as heated humidification, use of spacer devices, helium oxygen, and nitric oxide mixtures were allowed if equally distributed between the intervention and control groups. We also excluded studies where different bronchodilation agents were used between the intervention and control groups.
Types of outcome measures
- Reduction in airway resistance, measured as a reduction in interrupter resistance (Rint) and additional effective resistance (ΔRrs)
- Patient outcome, mortality during critical care unit admission
- Patient outcome, duration of mechanical ventilation
- Adverse changes to haemodynamic observations
- Reduction in wheezing
- Freedom from contamination
- Quality of life
- Practitioner satisfaction including ease of use and convenience
Search methods for identification of studies
We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2012, Issue 5) (Appendix 1); Ovid MEDLINE (1950 to Week 19 2012) (Appendix 2); Ovid EMBASE (1980 to Week 19 2012) (Appendix 3); and CINAHL via EBSCOhost (1982 to Week 19 2012) (Appendix 4).
Searching other resources
We did not limit the search by language or publication status.
We contacted manufacturers of MDIs and nebulizers that have been adapted for use within a ventilator circuit (for example Philips Respironics, Cardinal Health and Trudell Medical) to identify any published, unpublished or ongoing studies which met the inclusion criteria.
We reviewed conference proceedings available online for relevant trials (American Thoracic Society International Conference (2006 to 2012); European Society of Intensive Care Medicine (2003 to 2012); and the Respiratory Drug Delivery Conference (2000 to 2012)).
We screened reference lists within relevant trials to identify any further potential papers worthy of review.
Data collection and analysis
Selection of studies
We undertook the systematic review using the methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Two authors (AH and LV) independently examined the titles and abstracts identified by the search strategy to remove any duplicate records and obviously irrelevant reports. We retrieved and evaluated the full text versions of potentially relevant studies identified by at least one author. Two authors (AH and LV) independently assessed each study to determine if they met the eligibility criteria outlined above in the section Criteria for considering studies for this review. We resolved any disagreements by discussion between the authors (AH and LV), with a further author (FS) acting as arbiter. We have provided details of both included and excluded studies in the respective tables of this review.
Data extraction and management
AH and FS extracted data independently utilizing a standardized data extraction form based on the Cochrane Anaesthesia Review Group recommendations (see Appendix 5). We resolved any disagreements by discussion between the authors (AH and FS), with a further author (LV) acting as arbiter. The data extraction form included the following.
- General information: author(s), title, source, contact address, year of study, country of study, language of publication, year of publication.
- Trial characteristics: design (RCT) and risk of bias assessment criteria as outlined below in the section Assessment of risk of bias in included studies.
- Participants: baseline characteristics (including other ventilation and bronchodilation strategies outlined above in the section Types of interventions), inclusion and exclusion criteria, sample size and number of patients allocated to each intervention group, co-morbidity.
- Interventions: detailed description of the comparison devices and administration methods, bronchodilator administered.
- Outcomes. Primary outcomes: reduction in airway resistance, measured as a reduction in interrupter resistance (Rint) and additional effective resistance (ΔRrs); patient outcome including mortality during critical care unit admission and duration of mechanical ventilation. Secondary outcomes: adverse changes to haemodynamic observations; reduction in wheezing; freedom from contamination; quality of life; and practitioner satisfaction including ease of use and convenience.
- Other: sources of funding, conflicts of interest, unexpected findings.
We used the statistical package Review Manager software RevMan 5.1, utilizing double data entry with two authors (AH and FS) to control and correct data entry errors.
Assessment of risk of bias in included studies
We assessed the risk of bias of included studies using The Cochrane Collaboration's tool for assessing risk of bias as outlined by Higgins 2011. The standard components in this tool include adequacy of allocation generation, allocation concealment, blinding, completeness of outcome data, possible selective outcome reporting and any other potential sources of bias. In addition, a further four components were considered as potential sources of bias for cross-over trials; namely appropriateness of the cross-over design, randomization of the order of treatments received, freedom from the bias of carry over effects, and the availability of unbiased data. Each component was judged: 'Yes' for low risk of bias, 'No' for high risk of bias, or 'Unclear'. We have included a 'Risk of bias' table as part of the 'Characteristics of included studies' and a 'Risk of bias summary' figure which details all of the judgements made for all included studies in this review.
Assessment of risk of bias was carried out independently by two authors (AH and FS). We resolved any disagreements by discussion between the authors, with a further author (LV) acting as arbiter.
Measures of treatment effect
We intended to use the risk ratio (RR) as the effect measure for dichotomous data, and to calculate the mean difference (MD) or the standardized mean difference (SMD) with a 95% confidence interval (CI), as appropriate, for continuous outcomes.
Unit of analysis issues
If suitable data were available from cross-over trials, we intended to adopt the approach recommended by Elbourne 2002. We intended to include data using results from paired analyses where estimates of within patient differences and means and standard errors were either available, obtainable from the trialists or could be calculated.
Dealing with missing data
Where data were missing, we contacted the original investigators to request the missing data. We intended to perform intention-to-treat (ITT) analysis for dichotomous data. For continuous data we intended to perform ITT analyses if sufficient results were available from the included studies.
Assessment of heterogeneity
We intended to assess clinical heterogeneity using a three step approach. We initially intended to assess graphical depictions of confidence intervals generated by Review Manager software (RevMan 5.1) for the amount of overlap present. Statistical heterogeneity is indicated if there is poor overlap of confidence intervals (Higgins 2011). We intended to explore the presence of heterogeneity formally using the Chi
Assessment of reporting biases
We intended to generate funnels pots using the mean differences and standard errors for each primary outcome to visually assess the impact of study size on treatment estimates. If more than 10 studies were included in a meta-analysis, we intended to also use the regression asymmetry test to test for funnel plot asymmetry, as described by Egger 1997. Where the intervention effect was measured in terms of odds ratios for binary data, we intended to test funnel plot asymmetry using the arcsine test proposed by Rücker 2008.
We intended to combine data from parallel group and cross-over trials for meta-analysis. In case of bias due to carry over effect in cross-over trials, we intended to incorporate data from the first time period only if the necessary information was available. For cross-over trials when both time periods were used and no standard deviation of the mean difference was available, we intended to impute this using the correlation coefficient from other studies. We intended to calculate this from as many other studies as possible. We intended to analyse the results using inverse variance meta-analysis.
We intended to also meta-analyse data from parallel group and cross-over trials separately. If there was a discrepancy between the two we intended to report the results separately, otherwise the results of the meta-analyses would be reported together.
We intended to employ both a fixed-effect model and a random-effects model to combine data. If there was a discrepancy between the two, we intended to report results from both models. If there was no discrepancy, we intended to report the results from the fixed-effect model if the I
Subgroup analysis and investigation of heterogeneity
We planned to perform subgroup analyses to assess the impact of other ventilation and bronchodilation strategies such as heated humidification, use of spacer devices, helium oxygen mixtures, and nitric oxide mixtures for ventilation. Additionally, we planned subgroup analyses to estimate the impact of differing doses of bronchodilator agents.
We did not perform subgroup analyses as there were inadequate data available from the studies or the study authors to enable the groupings to be made. Additionally, the low number of included studies did not allow for any subgroups to be large enough to enable meaningful analyses.
We intended to perform a sensitivity analysis comparing the intervention effect in trials judged to have a low risk of bias (that is, trials in which all components of The Cochrane Collaboration's tool for assessing risk of bias have been judged as 'Yes') to trials which have been judged as having a moderate to high risk of bias (that is, trials in which one or more of the components of The Cochrane Collaboration's tool for assessing risk of bias have been judged as 'Unclear' or 'No').
We intended to perform a sensitivity analysis comparing the intervention effect in trials that based the decision to discontinue mechanical ventilation on pre-specified standardized criteria within the study compared to studies that based this decision on clinicians' judgements alone. This was to estimate the potential for a biased effect when the duration of mechanical ventilation was determined by a subjective judgement.
We intended to perform a sensitivity analysis comparing the intervention effect in trials that used combination administration of bronchodilators of differing drug groups to studies that administered a single bronchodilator agent. This would provide an estimate of the potential for a biased treatment effect when combination bronchodilator therapy was utilized.
Description of studies
The studies were prospective, randomized, cross-over trials conducted on mechanically ventilated adult patients in intensive care units (ICUs). The trials compared nebulizers and MDIs for aerosol bronchodilation, and the order of the interventions were randomized.
Results of the search
Our search identified 2080 titles and abstracts. A total of 18 abstracts were potentially relevant and we obtained the full publications of these. Two authors (AH and LV) independently read the full text publications and referred to a third author (FS) regarding four studies. From this, 11 studies were initially identified as having met the inclusion criteria. The 'study selection algorithm' in the 'Study Quality Assessment and Data Extraction form' (Appendix 5) was then applied to these studies. Eight studies were subsequently excluded (see 'Characteristics of excluded studies' table) and three studies were identified as having met the inclusion criteria (see 'Characteristics of included studies' table). A flow diagram of the search results is shown in Figure 1.
|Figure 1. Study flow diagram.|
We included three studies in this review with a total of 46 patients, which are described in the 'Characteristics of included studies' table. Individual sample sizes of each study were 18 (Gay 1991), 10 (Manthous 1993) and 18 (Guerin 1999) participants, and the studies were conducted in ICUs in America (Gay 1991; Manthous 1993) and France (Guerin 1999) respectively. Participants were recruited from a single medical ICU (Guerin 1999) and multiple ICUs from within a single institution (Manthous 1993). The age range of the participants for the Manthous 1993 and Guerin 1999 studies was 44 to 78 years, with a mean age of 69 years given for the Gay 1991 study. There were a higher number of males (32) than females (14).
The studies compared an MDI and nebulizer for aerosol bronchodilator delivery, with two studies utilizing a single short acting beta2-agonist (albuterol) delivered in either a single dose (Gay 1991) or successively increasing doses (Manthous 1993). The other study used a combination therapy of beta2-agonist and anticholinergic, delivered as a single dose (Guerin 1999). Participants were sedated in one study (Manthous 1993) and sedated and paralysed in another (Guerin 1999). The study by Gay 1991 did not give any information regarding the sedation or anaesthesia of the participants.
Guerin 1999 attached the MDI adapter 15 to 20 centimetres from the Y-piece, actuating the MDI on the onset of the mechanical breath and applying a four second inflation hold. Gay 1991 delivered the MDI bronchodilator dose in three breaths, using a slow manual inflation of the lungs and an inflation hold prior to recommencing mechanical ventilation. Manthous 1993 attached the adapter directly to the endotracheal (ET) tube, timed actuation with inspiration, and did not use an inflation hold. Two studies placed the nebulizer between 10 and 20 centimetres from the ET tube (Guerin 1999; Manthous 1993), with Gay 1991 describing the nebulizer as being placed near the Y junction between the ventilator tubing and ET tube. When stated, gas flows of five (Guerin 1999) and six litres per minute (Manthous 1993) were used to deliver the bronchodilator dose over 20 (Gay 1991) to 30 minutes (Guerin 1999; Manthous 1993). The wash out period between crossing over to the alternative method of administration was four (Gay 1991; Manthous 1993) and 10 hours (Guerin 1999). Respiratory mechanics were obtained using the end-inspiratory interruption technique under constant flow inflation in two studies (Guerin 1999; Manthous 1993). The third study (Gay 1991) obtained recordings during stepwise deflations of a relaxed respiratory system.
The studies by Manthous 1993 and Guerin 1999 both reported on the review's primary outcome measure of reduction in airway resistance, measured as a reduction in additional effective resistance (ΔRrs). All of the studies reported on adverse changes to haemodynamic observations, one of the secondary outcomes of this review. None of the studies reported the patient outcome of mortality during critical care unit admission, or duration of mechanical ventilation.
We excluded eight studies. Three studies did not meet our criteria because they compared different types of MDI with no nebulizer comparison (Fernandez 1990; Fuller 1994; Waugh 1998). A further study by Marik 1999 compared the pulmonary bioavailability of bronchodilators when delivered by MDI and nebulizer using urinary analysis of drug levels, and was excluded as it did not record any of the review's outcomes. Similarly, the study by Fuller 1990, which compared lung deposition of aerosolized bronchodilator therapy administered through MDI and nebulizer, did not record any of the review's outcomes and was also excluded. The study by Duarte 2000 included a participant who had received intravenous bronchodilator in the overall data analysis. One trial was conducted on patients who were breathing spontaneously and not receiving mechanical ventilation (Gervais 1987). The final excluded study by Gutierrez 1988 provided only limited data. We contacted the author but were unable to obtain any further study reports or data.
Risk of bias in included studies
We used The Cochrane Collaboration’s domain-based evaluation tool available in RevMan 5.1 for assessing risk of bias. In addition, we added four further domains based on the recommendations outlined by Higgins 2011 for assessing risk of bias in cross–over trials. Most of the trials had a low risk of bias across the 10 domains (Figure 2). Study authors were contacted to supplement information, where needed, to permit judgements on the classification of each risk of bias item. One of the authors did not respond, and for two of the trials the authors responded but the data were not available.
|Figure 2. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
Allocation sequence was adequately generated in two of the trials (Gay 1991; Guerin 1999) and was not reported in the third (Manthous 1993). None of the trials reported the methods used to conceal allocation, and we were unable to obtain this information from the authors direct. Given the nature of the intervention, blinding of participants and investigators delivering the interventions was not possible. We therefore assessed the risk of bias based on the blinding of the outcome assessors. In two of the trials, investigators completing post-sampling analysis (Guerin 1999) in addition to data acquisition (Gay 1991) were blinded to treatment modality. Blinding of outcome assessors was unclear in the third study (Manthous 1993). No incomplete data were apparent in any of the studies; however Gay 1991 excluded two participants from the original recruited sample as they did not meet the study inclusion criterion of a diagnosis of airways obstruction. In two trials (Guerin 1999; Manthous 1993) there was insufficient information to enable a judgement. The third study identified and reported on all three pre–set outcomes (Gay 1991). Other potential sources of bias included tracheal suctioning 'as required' prior to data collection (Gay 1991), patients receiving other inhaled bronchodilators that were not under study prior to onset of the investigation (Guerin 1999), and variations in participants and bronchodilator dose (Manthous 1993).
The cross-over design was suitable for all three trials as all participants were deemed clinically stable, with a history of chronic respiratory disease and absence of haemodynamic instability during the study period. The order of receiving treatments was randomized in two studies, using a coin flip (Gay 1991) and a random order table (Guerin 1999). There was insufficient reporting to enable a judgement in the third trial (Manthous 1993). All of the trials were judged to be free from carry over effects with wash out periods ranging from four (Gay 1991; Manthous 1993) to 10 hours (Guerin 1999). We looked for unbiased data to be available with some form of paired analysis, as recommended by Elbourne 2002. No dropouts or systematic differences between the two study periods were present in the trials. Paired analysis (Student’s t-test) was available for both interventions in two trials (Gay 1991; Guerin 1999) but only for the nebulizer treatment response in the third (Manthous 1993). The judgement on the classification of risk of bias is shown in Figure 3.
|Figure 3. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
Effects of interventions
Reduction in airway resistance
Reduction in airway resistance, measured as a reduction in interrupter resistance (Rint) and a reduction in additional effective resistance (ΔRrs), was reported in one trial (Guerin 1999). Guerin 1999 presented results as mean and standard error of the mean (SEM) at baseline and mean and SEM after cross-over. Manthous 1993 reported resistive pressure drop as a reduction in either total resistance of the respiratory system (Rrs) or Rint. Results were presented as mean at baseline and after cross-over, but they did not report whether further reported figures were SEM or standard deviation (SD). See Appendix 6.
The correlation coefficient of the patients' baseline and follow-up measurements, the paired t-test statistics, exact P values or confidence intervals were not available either in the study reports or from the authors direct. Therefore, estimates of the SE of the differences could not be calculated (Elbourne 2002; Higgins 2011) and it was not feasible to combine the study findings.
Results from Guerin 1999 suggest the choice of delivery device affects each component of total respiratory system resistance differently, with the MDI resulting in a significant reduction in Rint and the nebulizer a significant reduction in ΔRrs. Results from Manthous 1993 suggest that a significant reduction in Rrs or Rint is achieved when the bronchodilator is administered via the nebulizer.
Manthous 1993 demonstrated methodological adequacy in terms of incomplete outcome data reporting; an unclear randomization process, allocation concealment, blinding and selective outcome reporting alongside a small sample mean it is difficult to place much weight on these results. The authors did not state exactly which measure of respiratory system resistance was being used. From the description, this could be Rrs or Rint. Exact figures for post-treatment MDI were not provided, the authors stated that these “had no significant effect” (Manthous 1993, p1568). All results were presented as a figure (Figure 2, p1568) in the published paper. The exact data were not obtainable from the authors direct, and we could not confirm with the authors or publisher if the figure had been altered to fit within the published manuscript. No other reports of this study were available.
Guerin 1999 demonstrated methodological adequacy in terms of randomization, blinding and incomplete outcome data reporting. An unclear allocation concealment and selective outcome reporting mean results must be interpreted with caution. Additionally, all of the participants had chronic obstructive pulmonary disease (COPD).
We used the approach recommended by Elbourne 2002 to further analyse the study results. Making the assumption that baseline differences in airway resistance prior to administration of either method of delivery were not considered very different, comparisons of post-inhalation measures were carried out. For each study, we estimated the SE for the difference in post-inhalation resistances between MDI and nebulizer using the formula for continuous data provided by Elbourne 2002 and using an arbitrary range of correlation coefficients.
There was no observed difference in post-inhalation ΔRrs between the two administration methods in the Guerin 1999 study. The assumption that the baseline measures were similar did not hold in this study; the post-inhalation measures were equal hence it was not surprising that there was no statistical evidence of a difference in post-inhalation measures between the two treatments. In terms of Rint, there was no statistical evidence of a difference in the post-inhalation resistance between the two administration methods, even if the differences were assumed to be highly correlated. The results from Manthous 1993 demonstrated no statistical evidence of a difference in post-inhalation resistance between the two administration methods, even if the correlation coefficient between the differences was assumed to be high.
We also tested the differences in mean reduction between nebulizer and MDI, again using the methods outlined by Elbourne 2002. We estimated the paired SEs using the bounded P value in one arm of the study and assuming the SE of the reductions would be the same in the other arm as each patient acted as their own control.
Assuming a level of significance at P = 0.05, the results from Guerin 1999 suggested that a statistically significant change in resistance was achieved in ΔRrs when the correlation coefficient was 0.6 or above. These estimates were based on yet a further assumption and therefore were less likely to be reliable. Further results demonstrated no statistical evidence of a change in Rint resistance between the two administration methods, even if the correlation coefficient between the differences was assumed to be high.
The Manthous 1993 results suggested that a statistically significant change in resistance was achieved at all levels of correlation. These estimates were based on yet a further assumption and therefore were less likely to be reliable. Additionally, it was not possible to accurately identify the outcome measure used by Manthous 1993 (See Included studies) and therefore these results should be interpreted with extreme caution.
Adverse changes to haemodynamic observations
Adverse changes to haemodynamic observations were measured as a change in heart rate (beats per minute) in two trials (Gay 1991; Guerin 1999). Guerin 1999 presented results as mean and SE at baseline and mean and SE after cross-over. Gay 1991 presented results as the difference in means and SD. The correlation coefficient of the patients' baseline and follow-up measurements, the paired t-test statistics, and the exact P values or confidence intervals were not available in the Guerin 1999 study report or from the authors direct. Therefore, estimates of the SE of the differences could not be calculated (Elbourne 2002; Higgins 2011) and it was not feasible to combine the study findings.
Results suggested that heart rate was not significantly altered with either method of administration.
Guerin 1999 demonstrated methodological adequacy in terms of randomization, blinding and incomplete outcome data reporting. An unclear allocation concealment and selective outcome reporting meant results must be interpreted with caution. Additionally, all of the participants had COPD.
The increase noted in the heart rate was not statistically different for either method of administration.
Gay 1991 demonstrated methodological adequacy in terms of randomization, blinding, incomplete and selective outcome reporting. Unclear allocation concealment and the endotracheal suctioning of patients "when necessary" (Gay 1991, p68) prior to data collection may have influenced the results as previous studies have demonstrated that a significant rise in heart rate is associated with this procedure (Johnson 1994).
Summary of main results
This review included three trials, two addressing the primary outcome measure of a reduction in airway resistance (Guerin 1999; Manthous 1993) (n = 10, n = 18) and two the secondary outcome measure of adverse changes to haemodynamic observations (Gay 1991; Guerin 1999) (n = 18, n = 18). Limitations in data availability and reporting in the included trials precluded meta-analysis and therefore the present review consisted of a descriptive analysis.
Results from Guerin 1999 suggest that a significant reduction in interrupter resistance (Rint) is achieved when the bronchodilator is administered via a MDI and a significant reduction in additional effective resistance (ΔRrs) is achieved when the bronchodilator is administered via a nebulizer. Manthous 1993 suggest that a significant reduction in Rrs or Rint is achieved when the bronchodilator is administered via the nebulizer. The exact measure of respiratory system resistance used in this study is unclear. Additionally, post-treatment data for the MDI from Manthous 1993 had to be estimated from a published figure and therefore all results need to be interpreted with caution.
Further analysis of the study results, using the approach recommended by Elbourne 2002 for the estimation of the SEs for the difference of post-inhalation resistances, resulted in no statistical evidence of a difference, even if the correlation coefficient between the differences was assumed to be high. Testing the differences in mean reduction between nebulizer and MDI using the Elbourne 2002 methods, a statistically significant change in resistance is achieved with nebulizer delivery. Results from Guerin 1999 demonstrate this effect only in ΔRrs and when the correlation coefficient is 0.6 or above. Results from Manthous 1993 demonstrate this effect across all levels of correlation. However, these results must be interpreted with extreme caution as all further analyses were based on increasing levels of assumption. Additionally, it is not possible to correctly identify the outcome measure used by Manthous 1993 (see Characteristics of included studies). Cautious interpretation of the included study results suggests that nebulizers could be a more effective method of bronchodilator administration than MDIs in terms of a change in resistance.
Heart rate was not significantly altered with either method of administration, however non-standardized respiratory care prior to data collection was present in Gay 1991. Additionally, all participants in Guerin 1999 had chronic obstructive pulmonary disease. No further eligible trials were found that addressed any of the other outcomes of the review.
Due to missing data issues, meta-analysis was not possible. Further analyses of included study results in relation to the primary outcome measure of a reduction in airway resistance had to be based on several levels of assumption about the study design. Additionally, small sample sizes and variability between the studies with regards to patient diagnoses, bronchodilator agent and administration technique mean that it would be speculative to infer definitive recommendations based on these results at this time. This is insufficient evidence to determine which is the most effective delivery system between nebulizer and MDI for aerosol bronchodilation in adult patients receiving mechanical ventilation.
Overall completeness and applicability of evidence
Three relevant studies were identified for inclusion. The studies identified were not sufficient to address the objectives of the review as variations in the patient diagnoses, bronchodilator agent and administration technique were present. Some primary and secondary objectives were not addressed at all in the evidence.
Quality of the evidence
The body of evidence that has been identified does not allow a robust conclusion regarding the objectives of the review. Three studies were included with a total of 46 participants (n = 10, n = 18, n = 18). Key methodological limitations were small sample sizes and variability between the studies with regards to patient diagnoses, bronchodilator agent and administration technique. The two studies which addressed the primary outcome measure of a reduction in airway resistance were consistent in their finding that nebulizer delivery was associated with the greater, statistically significant reduction. The two studies which addressed the secondary outcome measure of adverse changes to haemodynamic observations were consistent in their finding that no significant rise in heart rate was observed with either mode of delivery. The overall rating of the quality of the body of evidence was moderate. Reasons for downgrading the evidence by one level are the high risk of bias identified in two of the studies (Gay 1991; Manthous 1993) and the potential for imprecision of results due to the small sample sizes and estimation of one study result from a published figure (Manthous 1993).
Potential biases in the review process
Most of the review authors were familiar with the two delivery methods under comparison in this review, from their previous clinical experience. However, this did not influence the assessment of data. To our knowledge, no additional sources of bias were present in the review process.
Agreements and disagreements with other studies or reviews
Duarte 2000 found no difference in either Rint or ΔRrs when comparing the MDI and nebulizer for bronchodilator administration. This study was excluded from this review due to the administration of intravenous steroids, which may in part explain these conflicting results. Previous narrative reviews have had conflicting results. Fink 1999b strongly advocate MDIs, however this recommendation is based on drug deposition and aerosol delivery not on patient response assessed via respiratory variables. Additionally, no nebulizer comparison is considered. Hess 1991 also recommends the use of an MDI but, in agreement with Jantz 1999, highlights the importance of optimal administration techniques to achieve the benefits associated with this delivery route. If an optimal administration technique is used, equal physiological end points may be achieved and either method of administration is purported by Guerin 2008 and Dhand 2007b. Dhand 1997 also advocate MDI based on a cost effectiveness analysis completed by Bowton 1992, which demonstrated potential annual cost savings of $396,000 when MDIs were substituted for nebulizer therapy. Cost effectiveness analysis was not an outcome of this review, therefore these claims cannot be substantiated or refuted. Further narrative reviews (Dolovich 2005; O'Doherty 1997) concluded that no advantage exists with either method of administration and both MDI and nebulizer can be used to achieve successful bronchodilation in patients receiving mechanical ventilation.
Implications for practice
Existing randomized controlled trials, including randomized cross-over trials where the order of the intervention was randomized, comparing a nebulizer and MDI for aerosol bronchodilation in mechanically ventilated adult patients do not provide sufficient evidence to support either delivery method at this time.
Implications for research
Further large, randomized cross-over trials are required to assess which is the most effective delivery system for aerosol bronchodilation in adult patients requiring invasive mechanical ventilation. Additionally, there are currently not enough studies that measure the respiratory mechanics of the resistance of the respiratory system to gas flow, despite this appearing to be physiologically the most appropriate measure to assess bronchodilator response to MDIs and nebulizers in this patient group. Future studies should also address patient outcome measures such as mortality during critical care unit admission and duration of mechanical ventilation, the other primary outcomes of this review. In addition, future studies should address the secondary outcome measures of this review.
Any future research evaluating two interventions utilizing a cross-over study design should ensure that findings are reported as a difference in means ± SE or SDM, or provide sufficient data in the study report to enable this to be calculated. This will enable any subsequent meta-analysis of study findings.
We thank Prof Harald Herkner (content editor), Prof Nathan Pace (statistical editor), Prof Claude Guerin, and Dr Mark D Neuman, (peer reviewers) for their help and editorial advice. We also thank Dr Sandra Bonellie for her work on the protocol for this review prior to her retirement.
Data and analyses
This review has no analyses.
Appendix 1. Search strategy for CENTRAL, T he Cochrane Library
#1 MeSH descriptor Metered Dose Inhalers explode all trees
#2 MeSH descriptor Nebulizers and Vaporizers explode all trees
#3 MeSH descriptor Bronchodilator Agents explode all trees
#4 MeSH descriptor Administration, Inhalation explode all trees
#5 MeSH descriptor Drug Delivery Systems explode all trees
#6 MeSH descriptor Nitric Oxide explode all trees
#7 metered-dose inhaler*
#10 (bronchodilat* near (therap* or strateg*))
#11 (heated near humidific*)
#12 (spacer near devic*)
#13 (helium near oxygen)
#14 ((nitric oxide or NO) near mixture*)
#15 (bronchodilator* near delivery)
#16 (aerosol near bronchodilat*)
#17 (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16)
#18 MeSH descriptor Respiration, Artificial explode all trees
#19 mechanical near ventilat*
#20 (#18 OR #19)
#21 (#17 AND #20)
Appendix 2. Search strategy for MEDLINE (OvidSP)
1. exp Metered Dose Inhalers/
2. exp "Nebulizers and Vaporizers"/ or Bronchodilator Agents/
3. Administration, Inhalation/
4. Drug Delivery Systems/
5. Nitric Oxide/ad, tu, sd [Administration & Dosage, Therapeutic Use, Supply & Distribution]
6. metered-dose inhaler*.mp.
9. (bronchodilat* adj6 (therap* or strateg*)).mp.
10. (heated adj3 humidific*).mp.
11. (spacer adj3 devic*).mp.
12. (helium adj3 oxygen).mp.
13. ((nitric oxide or NO) adj3 mixture*).ti,ab.
14. (bronchodilator* adj3 delivery).mp.
15. (aerosol adj6 bronchodilat*).mp.
16. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15
17. exp Respiration, Artificial/
18. (mechanical adj3 ventilat*).mp.
19. 18 or 17
20. 19 and 16
21 ((randomized controlled trial or controlled clinical trial).pt. or randomized.ab. or placebo.ab. or clinical trials as topic.sh. or randomly.ab. or trial.ti.) and humans.sh.
22. 21 and 20
Appendix 3. Search strategy for EMBASE (OvidSP)
1 exp Metered Dose Inhaler/
2 exp Nebulizer/ or exp Medical Nebulizer/
3 exp Vaporizer/
4 exp Bronchodilating Agent/
5 exp Inhalational Drug Administration/
6 exp Drug Delivery System/
7 exp Nitric Oxide/dt, ad, do, ih [Drug Therapy, Drug Administration, Drug Dose, Inhalational Drug Administration]
8 metered-dose inhaler*.mp.
11 (bronchodilat* adj6 (therap* or strateg*)).mp.
12 (heated adj3 humidific*).mp.
13 (spacer adj3 devic*).mp.
14 (helium adj3 oxygen).mp.
15 ((nitric oxide or NO) adj3 mixture*).ti,ab.
16 (bronchodilator* adj3 delivery).mp.
17 (aerosol adj6 bronchodilat*).mp.
18 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17
19 exp Artificial Ventilation/
20 (mechanical adj3 ventilat*).mp.
21 19 or 20
22 21 and 18
Appendix 4. Search strategy for CINAHL (EBSCOhost)
S26 S19 and S25
S25 S20 or S21 or S22 or S23 or S24
S24 AB trial* or random*
S23 (MM "Multicenter Studies")
S22 (MM "Placebos")
S21 (MM "Double-Blind Studies") or (MM "Single-Blind Studies") or (MM "Triple-Blind Studies")
S20 (MM "Random Assignment") or (MH "Clinical Trials+")
S19 S15 and S18
S18 S16 or S17
S17 TX mechanical and ventilat*
S16 (MH "Respiration, Artificial+")
S15 S1 or S2 or S3 or S4 or S5 or S6 or S7 or S8 or S9 or S10 or S11 or S12 or S13 or S14
S14 TX aerosol and bronchodilat*
S13 TX bronchodilator* and delivery
S12 AB nitric oxide or NO
S11 TX helium and oxygen*
S10 AB spacer*
S9 TX heated and humidific*
S8 AB bronchodilat* and therap*
S7 TX Nebuliser
S6 TX metered-dose inhaler*
S5 (MH "Nitric Oxide")
S4 (MH "Drug Delivery Systems+")
S3 (MM "Administration, Inhalation")
S2 (MH "Bronchodilator Agents+")
S1 (MM "Nebulizers and Vaporizers")
Appendix 5. Study quality assessment and data extraction form
Any other published versions/reports of this trial?
All references to a trial need to be linked under one Study ID both on this form (p1) and in RevMan.
Add other additional lines/codes as required
Trial characteristics – Risk of bias assessment
∗ measures to include airway resistance (Rrs min, Rrs max, ΔRrs, Rint) Remember – we are looking for recording of these outcomes; not reporting.
Appendix 6. Summary of primary outcome measures
Contributions of authors
Conceiving the review: Agi Holland (AH)
Co-ordinating the review: AH
Undertaking manual searches: AH and Gill McCrossan (GM)
Screening search results: AH and Linda Veitch (LV)
Organizing retrieval of papers: GM and LV
Screening retrieved papers against inclusion criteria: AH and LV
Appraising quality of papers: AH and Fiona Smith (FS)
Abstracting data from papers: AH and FS
Writing to authors of papers for additional information: GM and LV
Providing additional data about papers: GM and LV
Obtaining and screening data on unpublished studies: Not required
Data management for the review: AH and FS
Entering data into Review Manager (RevMan 5.1): AH and FS
RevMan statistical data: Not required
Other statistical analysis not using RevMan: Kay Penny (KP)
Double entry of data: (data entered by person one: AH; data entered by person two: FS)
Interpretation of data: AH, LV, FS, GM, KP
Statistical inferences: KP
Writing the review: AH, FS, KP, GM
Securing funding for the review: AH
Performing previous work that was the foundation of the present study: AH
Guarantor for the review (one author): AH
Person responsible for reading and checking review before submission: AH
Declarations of interest
Agi Holland: see Sources of support
Fiona Smith: see Sources of support
Kay Penny: none known
Gill McCrossan: see Sources of support
Linda Veitch: none known
Caroline Nicholson: none known
Sources of support
- Edinburgh Napier University, UK.
- Karen Hovhannisyan, Denmark.Help with search strategies
- The Chief Scientist Office of The Scottish Government, UK.Financial support for Agi Holland, Gill McCrossan and Fiona Smith to undertake the Review through Grant number CZG/2/417
Differences between protocol and review
Dr Kay Penny joined the review team for statistical advice and support following Dr Sandra Bonnellie's retirement from Edinburgh Napier University.
Medical Subject Headings (MeSH)
*Critical Illness; *Metered Dose Inhalers; *Nebulizers and Vaporizers; *Respiration, Artificial; Aerosols; Airway Resistance [*drug effects; physiology]; Bronchodilator Agents [*administration & dosage]; Heart Rate [drug effects; physiology]; Intensive Care Units; Randomized Controlled Trials as Topic
MeSH check words
* Indicates the major publication for the study