Spontaneous recovery of neuromuscular blockade is an independent risk factor for postoperative pulmonary complications after abdominal surgery

A secondary analysis

Abstract

BACKGROUND 

In intermediate-to-high-risk patients, major abdominal surgery is associated with a high incidence of postoperative complications, mainly pulmonary. Neuromuscular blocking drugs have been suggested as a contributing factor, but this remains unproven.

OBJECTIVE 

To define the relationship of neuromuscular blockade management (reversal) with postoperative pulmonary complications (PPCs).

DESIGN 

The individualised PeRioperative Open-lung approach Versus standard protectivE ventilation in abdominal surgery study was a prospective, multicentre, four-arm, randomised controlled trial. This is a secondary analysis of the data.

SETTING 

Twenty-one teaching hospitals in Spain. The study was conducted between 2 January 2015, and 18 May 2016.

PATIENTS 

Age more than 18 years with an intermediate-to-high risk for PPCs, scheduled for major abdominal surgery lasting more than 2 h. Exclusion criteria included pregnancy or breastfeeding, and moderate-to-severe organ diseases.

INTERVENTIONS 

The mode of reversal of neuromuscular blockade determined two patient groups: pharmacological reversal versus spontaneous recovery.

MAIN OUTCOME MEASURES 

The primary outcome was a composite of PPCs during the first 30 postoperative days. The association between categorical variables and PPCs within 30 days was studied. Univariate and multivariable logistic regression modelling and propensity score analyses were performed.

RESULTS 

From the 923 patients included, 596 (64.6%) presented with PPCs within 30 days after surgery. Patients who developed these complications were older with a higher BMI, a lower pre-operative SpO₂, a higher ASA physical status score and a higher incidence of arterial hypertension, diabetes mellitus or chronic obstructive pulmonary disease. Pharmacological neuromuscular blockade reversal was associated with a lower incidence of PPCs (odds ratio 0.62, 95% CI 0.47 to 0.82).

CONCLUSION 

Spontaneous recovery of neuromuscular blockade was an independent risk factor for PPCs in patients with intermediate-to-high risk, undergoing abdominal surgery. We suggest this factor should be included in future studies on PPCs.

TRIAL REGISTRATION 

clinicaltrials.gov identifier: NCT02158923.

Introduction

Major abdominal surgery is associated with a high incidence of severe postoperative complications that negatively affect patient survival and healthcare costs. Recently, in the individualised PeRioperative Open-lung approach Versus standard protectivE ventilation in abdominal surgery (iPROVE) study of patients undergoing abdominal surgery, Ferrando et al.¹ investigated the outcomes of 967 patients who were at intermediate-to-high risk of developing postoperative pulmonary complications (PPCs). Patients were randomised to four ventilatory interventions during and after surgery. The primary outcome was a composite of pulmonary and systemic complications during the first 7 postoperative days. No difference between the groups was found for the primary outcome, although the prevalence of PPCs was significantly lower in the most individualised treatment group when compared with the standard group. However, differences between groups were lower than initially expected.

Studies have suggested an association between the use of neuromuscular blocking agents (NMBAs) and the development of PPCs, although most of the evidence comes from retrospective studies that have not clearly defined prespecified outcomes for PPCs, thus limiting potential causality inference.^2–7 Based on these findings, guidelines and expert recommendations have stressed the importance of neuromuscular monitoring and pharmacological reversal of the effects of NMBAs^8–10 but clinicians are reluctant to follow this advice.

In the iPROVE study,¹ neuromuscular monitoring and reversal were specifically collected as part of the study protocol, but those data were not published.

We hypothesised that muscle weakness associated with spontaneous recovery from NMB at the end of surgery could be a risk factor for the development of PPCs in an intermediate-to-high-risk surgical population. In this unplanned secondary analysis, we sought to elucidate whether PPCs might be related to neuromuscular blockade management and its reversal.

Methods

Study design

The iPROVE study was a prospective, multicentre, four-arm, randomised controlled trial conducted in 21 teaching hospitals in Spain and included 1012 patients: the data from 967 were analysed (Fig. 1). The study was registered with Clinical trials (no. NCT02158923). Ethical approval for this study (Ethical Committee No. CEIC 214/070) was provided by the Ethical Committee CEIC Hospital Clínico Universitario de Valencia, Valencia, Spain (Chairperson M. Labiós Gómez, secretary Ethics Committee) on 26 December 2014, as well as the institutional review boards of all participating centres. A steering committee monitored the study and an independent data and safety monitoring board was constituted. The complete protocol was registered before patient enrolment,¹¹ and the main results have been published.¹ Written informed consent was obtained from each patient before enrolment.

Fig. 1: Flowchart of the individualised PeRioperative Open-lung approach Versus standard protectivE ventilation in abdominal surgery secondary analysis.

Participants

Inclusion and exclusion criteria were previously described.^1,11 In brief, patients were eligible for inclusion if they were aged 18 years or older, had an intermediate-to-high risk for the development of PPCs (as defined by the ARISCAT score),¹² who were scheduled for major abdominal surgery with an expected surgical time longer than 2 h, and who had a BMI less than 35 kg m⁻². Patients were excluded if they were aged less than 18 years old, were pregnant or breastfeeding, or had moderate or severe acute respiratory distress syndrome, cardiac failure, a diagnosis or suspicion of intracranial hypertension, mechanical ventilation during the previous 15 days [including continuous positive airway pressure (CPAP)], pneumothorax or giant bullae, chronic obstructive pulmonary disease (COPD) requiring oxygen or CPAP, or were participating in another interventional study.

In the original study, the patients were randomised to four ventilatory interventions during and after surgery: individualised ventilation with high positive end expiratory pressure after a lung recruitment manoeuvre – open lung approach (OLA) – followed by individualised continuous positive airway pressure (OLA-iCPAP); individualised ventilation followed by standard CPAP (OLA-CPAP); standard ventilation followed by standard CPAP (STD-CPAP); and standard ventilation followed by standard postoperative oxygen therapy (STD-O₂).

Outcomes

The original protocol included a recommendation about the intra-operative use of NMBAs according to guidelines.⁹

The primary outcome was a composite of PPCs (hypoxemia requiring supplemental oxygen, a positive air-test,¹³ pleural effusion, acute respiratory failure requiring noninvasive or invasive ventilation, pneumonia and acute respiratory distress syndrome) during the first 30 postoperative days. Pulmonary outcomes were defined according to the joint task force of the European Society of Anaesthesiology and the European Society of Intensive Care Medicine¹⁴ (Appendix 1, https://links.lww.com/EJA/A247). Length of stay (LOS) in the ICU and in the hospital as well as death were recorded.

The primary and secondary data outcomes were recorded by an investigator 15 min and 3 h after postanaesthesia care unit (PACU) or ICU admission and at 1, 2, 7 and 30 days after surgery with a follow-up for mortality at 180 and 365 days after surgery.

Predictors of postoperative pulmonary complications

We assessed several peri-operative variables as possible confounders: age, sex, BMI, obstructive sleep apnoea syndrome, COPD, oncologic surgery, ARISCAT score, duration of surgery, type of surgery (laparotomy or laparoscopy), total intra-operative fluid administration, intra-operative bleeding, blood transfusion, oliguria (defined as intra-operative diuresis <0.5 ml kg⁻¹ h⁻¹), whether a patient's trachea was extubated in the operating room at the end of surgery, epidural analgesia use, intra-operative respiratory driving pressure (Δp), intra-operative respiratory system dynamic compliance (Crs), intra-operative mean arterial pressure, intra-operative partial pressure of arterial oxygen to inspiratory oxygen fraction ratio (PaO₂/FiO₂), and a need for surgical re-intervention.

Neuromuscular blockade management

Neuromuscular blockade management was at the discretion of the attending anaesthesiologist. The administration of NMBAs, reversal drugs and neuromuscular monitoring (yes/no) were recorded. An adequate neuromuscular blockade was encouraged (0 responses to train-of-four stimulation), meaning that at least the first dose of the selected NMBA might be twice the ED95 to provide a ‘deep’ or ‘intense’ neuromuscular blockade at the start of the procedure. Continuous intra-operative monitoring of the neuromuscular blockade was also suggested.¹¹ Selection of the NMBAs (intermediate acting) and reversal drugs (neostigmine or sugammadex) were again at the discretion of the attending anaesthesiologist. In the protocol, NMBAs drug reversal was stressed, the dose to be guided by neuromuscular monitoring. Despite this, the data revealed two groups: patients who received pharmacological reversal and patients who had spontaneous recovery from the NMB. Postoperative management in the PACU followed the established protocols at each participating centre.

Power calculation

We performed a power analysis of the available sample based on the previously reported incidence of PPCs for this cohort of patients, which was around 25%¹ and 19% in another observational study.¹⁵ With a significance level of 5% and assuming an incidence of PPCs of 20% a sample size of 700 patients was calculated as necessary to obtain a power of 86% to detect an absolute reduction of 10% in patients with NMBA drug reversal.

Statistical analysis

Normally distributed data were described as mean ± SD and nonnormally distributed data were reported as median values [interquartile range]. Categorical variables were reported as proportions. The association between categorical variables and PPCs in the first 30 postoperative days was studied using Pearson's χ² test. For numerical variables, the t test or the Mann–Whitney U test were used, depending on whether the data were normally distributed.

Both univariate and multivariable logistic regression models were performed to study the strength of the association between NMBA reversal and the development of PPCs in the first 30 days. As independent variables, those with P value less than 0.15 in the univariate analysis were introduced in the model. Subsequently, the effect of these variables in the number of patients in whom reversal was used or not used was evaluated (as stated, a change of at least 10% in the odds ratio (OR) was considered clinically relevant).

Only patients without missing values in the variables of interest were included in the final analysis.

In addition, a propensity score analysis (PSA) was developed to determine the probability of belonging to each group and was calculated using logistic regression with stepwise variable selection. In this model, age, sex, estimated blood loss, baseline PaO₂/FiO₂, duration of surgery and Crs at the end of surgery were included as independent variables. The validity of the results was studied by applying the PSA under different approaches: using the PSA as an adjustment variable; stratifying the PSA into five layers; using the PSA to match patients with and without reversal; and weighting the data according to the PSA.

SPSS Statistics for Windows, Version 22.0. (SPSS, IBM Corp., Armonk, New York, USA) and Stata Statistical Software version 13 (Stata Corp. 2013, College Station, Texas, USA) were used for calculations. All statistical tests were performed two sided and a P value less than 0.05 was considered significant.

Results

Nine hundred and twenty-three patients were included in this secondary analysis of the data. Of these, 596 (64.6%) suffered from PPCs during the first 30 postoperative days. The demographic and pre-operative characteristics of all patients are shown in Table 1 and Appendix 2 (Appendix 2, Supplemental Digital Content, https://links.lww.com/EJA/A247). Patients who developed PPCs were older, with a higher BMI, lower pre-operative SpO₂, higher ASA physical status score, and higher incidence of arterial hypertension, diabetes mellitus or COPD.

Table 1: Demographic and pre-operative data

Table 2 lists the intra-operative data for patients with and without PPCs. In summary, patients with PPCs showed higher airway plateau pressure (ΔP) and they had a lower static compliance, as well as a lower PaO₂/FiO₂ and a higher PaCO₂. Moreover, these patients suffered higher blood losses and showed higher volume of diuresis. The duration of surgery and mechanical ventilation were both longer in patients with PPCs. The proportion of patients receiving NMBA reversal drugs was lower in the group who developed PPCs (Table 3): out of 596 patients suffering PPCs, 287 (59.4%) patients whose neuromuscular blockade was pharmacologically reversed suffered PPCs, versus 309 (70.2%) patients with spontaneous reversal (P = 0.001) (Table 3). More nonreversed patients suffered pleural effusions and showed a positive air test (Table 3). No other differences were observed between patients with or without PPCs (Supplemental Digital Content, Appendices 3 and 4, https://links.lww.com/EJA/A247).

Table 2: Intra-operative data

Table 3: Postoperative data

The comparison of drug-induced with spontaneous neuromuscular blockade recovery groups regarding other postoperative outcomes is shown in Table 3. Pharmacological neuromuscular blockade reversal was associated with a lower prevalence of PPCs: OR 0.6 (95% CI, 0.46 to 0.69), and OR 0.62 (95% CI, 0.47 to 0.82), during the first 7 and the first 30 postoperative days, respectively. Although median hospital LOS was longer in patients suffering PPCs (8 [6 to 14] vs. 7 [5 to 11] days, P < 0.001) as was ICU LOS (1 [1 to 3] vs. 1 [1 to 2] days, P = 0.03) and mortality at day 30 was 2.1% in patients with PPCs vs. 0.3% in patients without PPCs, P = 0.023 (Table 3, Appendices 2 to 4, https://links.lww.com/EJA/A247), there was no influence of NMBAs management on these aspects

After multivariate logistic regression modelling [that included ARISCAT index (high), BMI, median intra-operative ΔP, type of surgery, ASA physical status classification, volume of fluids and intra-operative diuresis <0.5 ml kg⁻¹ h⁻¹], the OR of suffering PPCs increased: OR 0.67 (95% CI 0.49 to 0.91). After calculating a PSA with the variables associated with neuromuscular blockade reversal, the OR did not significantly change, no matter the analytical approach selected. The OR values ranged between 0.59 and 0.65 using the paired PSA considering patients with and without neuromuscular blockade pharmacological reversal or by weighting patients for their PSA (Table 4).

Table 4: Odds ratio of postoperative pulmonary complications at day 30 for patients with neuromuscular blocking agents reversal with different methods of propensity score analysis

Discussion

In this secondary analysis of the iPROVE study we showed that spontaneous recovery from neuromuscular blockade is an independent risk factor for the development of PPCs in intermediate-to-high-risk patients undergoing abdominal surgery. Moreover, the individual analysis showed that these patients had a higher incidence of a positive air-test in the PACU, a higher incidence of atelectases and pleural effusions.

As stated before, expert recommendations and guidelines of scientific societies have stressed the benefits of drug reversal of the neuromuscular blockade at the end of surgery. However, less than one-third of practitioners routinely administer anticholinesterases.^16,17 Several reasons may explain this fact: studies reporting an increase in PPCs associated with these drugs^4–6,18 possibly related to its paradoxical muscle relaxant effects,^19–21 the cardiovascular adverse effects of anticholinesterases and the need to administer anticholinergics to prevent this, and finally, the lack of awareness about the prevalence of postoperative residual neuromuscular blockade.¹⁶ This can be observed in the iPROVE study¹ where only 30% of the patients were monitored or reversed, despite the recommendations in the study protocol.

Intra-operative neuromuscular blockade management and the reversal of residual neuromuscular blockade are rarely taken into account in studies of strategies aiming to reduce PPCs. Because of the relationship found between NMBAs and PPCs^2–7 and studies showing the high prevalence of residual NMBA effects, even when appropriate neostigmine doses and a nerve stimulator were used, this has been considered as an investigational gap.²² As an example, Thilen et al.,²³ in a randomised controlled trial, observed an incidence of train-of-four (TOF) ratio less than 0.9 of 35% after tracheal extubation. Along this line, although not statistically significant (P = 0.053), we found that the prevalence of PPCs is higher in nonreversed patients, reinforcing previous findings.^5,24

Factors related to PPC development have been reviewed recently.²⁵ Residual block could contribute to PPCs not only because muscle weakness affects the muscles of the thoracic cage and the diaphragm, but also because of a lack of coordination of the upper airway muscles and protective airway reflexes, breathing and breathing control, swallowing and coughing.^25–29

Accordingly, we found a high incidence of hypoxaemia after breathing room air for at least 5 min (air test), possibly related to increased postoperative atelectasis if the neuromuscular blockade is not reversed adequately. The air-test detects postoperative atelectasis that otherwise may go undetected due to the routine use of supplemental oxygen in the PACU. Atelectases increased when a neuromuscular blockade was not reversed.³⁰ Also, it has been demonstrated that the effect of incomplete recovery of neuromuscular blockade becomes more evident when forced tests are performed to measure inspiratory capacity rather than measuring passive expiratory capacity,³¹ since the former is the most useful for assessing the capacity for lung re-expansion.²⁵ A degree of residual neuromuscular blockade would favour the formation of new atelectasis or hinder reopening of poorly ventilated areas. These effects can be observed immediately or in the days following surgery, as we also observed in our patients. It is also likely that effects of manoeuvres performed to improve oxygenation during and after the surgical procedure (summarised as ‘individualised protective ventilation’ or an ‘open lung approach’ strategy) could be diminished by the residual effects of the NMBAs. CPAP application,¹ and perhaps other noninvasive ventilation strategies, might counteract the deleterious effects of residual NMB as well.

Pleural effusions in nearly 15% of patients with no reversal were observed in our study. Although very frequent after abdominal surgery, pleural effusion is generally considered as a minor problem, with minimal consequence on postoperative management and prognosis.³² In previous studies this element was not included. The loss of volume of the atelectatic lung can increase pleural negative pressure, favouring interpleural accumulation of fluid.³³

Classically, postoperative pneumonia has been related to residual block,⁴ even when a TOF ratio of 0.9 was reached: this is related to pharyngeal dysfunction and swallowing alterations facilitating unnoticed microaspirations. In our study, as in that of Sasaki et al.³⁴ and Ledowski et al.,⁶ the incidence of pneumonia was not reduced by the use of reversal drugs.

Several investigations have analysed the association between NMBAs reversal and PPCs, but few were prospective. While some support the deleterious effects of not using pharmacological reversal,^3,4 others do not.^6,7,24 In addition, other studies have suggested a higher rate of PPCs when anticholinesterase agents were used, due to their muscle relaxant effects.^2,5,21,34 However, this should not occur when neostigmine is properly administered, that is, in a bolus dose of less than 0.06 mg kg⁻¹ with adequate neuromuscular monitoring (>2 TOF responses).^2,18,35 We hypothesised that one of the main reasons explaining the differences between our results and those of previous studies is the duration of surgery, as a longer duration could result in a higher requirement of repeated and cumulative doses of NMBAs.³⁶ However, other factors might contribute to differences as well (e.g. hypothermia).

When compared with neostigmine or no drug reversal, the incidence of residual effects (e.g. respiratory events in PACU, tracheal reintubation, oxygen saturation) are significantly lower after sugammadex,^6,36 but an effect on PPCs has not yet been demonstrated. Indeed, the POPULAR study observed no difference in PPCs whether NMB was reversed with sugammadex or neostigmine.⁷ This finding might be related to inadequate neuromuscular monitoring, suboptimal use of sugammadex, or to different teaching and implementation of neuromuscular blockade management between caregivers, settings or countries. These differences are illustrated by the results of a recent survey where experts did not think that ‘specific drugs or techniques to limit residual neuromuscular blockade’ were part of a care bundle but, surprisingly, it was recommended that both ‘pre-operative inspiratory muscle training’ and ‘intra-operative neuromuscular blockade limitation’ should be considered.³⁷ Thus, it is not surprising that, in the iPROVE study, compliance with neuromuscular block monitoring was low and thus its influence on PPCs remains unclear.

Limitations

The iPROVE study was not specifically designed to analyse the potential association between muscle relaxant management and PPCs. However, the study was exhaustive with respect to prospective data collection and the two main variables that potentially can be related to PPCs (monitoring and reversal) were recorded. Other limitations that might affect the interpretation of the results are that objective TOF ratio monitoring was not performed in the PACU^4,5,34 and that doses of NMBAs or reversal drugs were not recorded. Neither the stage of NMB recovery at which patients received the antagonist nor the TOF ratio when the patient's trachea was extubated were recorded. These are important limitations but, in the author's opinion, reflect real-world scenarios regarding neuromuscular blockade management. This could explain (at least in part) the relatively small difference (around 10%) we found in the incidence of PPCs whether the NMBA was reversed or not. However, despite this apparently low percentage, because of the large number of surgical procedures, this translates into a significant number of patients per year. In addition, we should remember that this was not a study intended to find a causal relationship.

Although optimisation of neuromuscular blockade (defined in the iPROVE study as ‘patients in whom neuromuscular block was monitored or reversed before extubation’ was not different among study groups [79 (33.2%), 66 (28.2%), 76 (31.8%), 66 (27.7%), P = 0·49, Table 2], a mixed effect of drug reversal, neuromuscular monitoring and postoperative CPAP could be considered (or cannot be discarded) (Table 2). Nevertheless, this pragmatic secondary analysis was performed to generate new hypotheses and not to establish recommendations.

We observed that despite the increase in PPCs in nonreversed patients, this had no apparent effect on important outcomes, such as LOS, unplanned ICU admission and mortality. This has diverse explanations, for instance that the PPCs composite score is too sensitive and not relevant to long-term outcomes, that the score considers items of variable severity (as arterial oxygen desaturation or atelectases) and the clinical outcome were quite variable, and that postoperative treatment prevents such deleterious outcomes.

Finally, the statistical analyses used have their own limitations. Secondary analysis of the data depends in part on the sample size of the original iPROVE study but, as stated in the methods section, on power,³⁸ both sample size and power were met by our study design. Propensity scores could be biased due to unknown confounders.³⁹ However, one feature of the iPROVE study was the wide range of parameters and measurements included, minimising the possibility of unknown confounders.

We conclude that allowing the spontaneous recovery of NMBAs is an independent risk factor for the appearance of PPCs. Based on the results of this analysis we highlight the potential benefits of following the guidelines recommending the use of pharmacological reversal of NMBAs, with the possible exception of situations where the objective exclusion of residual block has been ascertained by adequate neuromuscular monitoring. The results of this secondary analysis of the iPROVE study supports the need for well powered randomised controlled trials to study the potential benefit of reversing residual neuromuscular block on the incidence of PPCs. In addition, it appears that educational interventions regarding neuromuscular blockade management are still necessary.

Acknowledgements relating to this article

Assistance with the study: none.

Financial support and sponsorship: the original article was funded by the Instituto de Salud Carlos III of the Spanish Ministry of Economy and Competitiveness (grant PI14/00829, cofinanced by the European Regional Development Fund), and received the support of the Grants Programme of the European Society of Anaesthesiology.

Conflicts of interest: OD-C received two Merck Investigator Studies Program Review Committee (MISP-RC) Merck Sharp & Dohme medical grants for international research projects, payment for medical advice and travel not related to this research; CLE is Medical Director Surgical Services at the Consorcio Hospital General Universitario de Valencia, Valencia, Spain.

Presentation: none.

Members of the Individualized PeRioperative Open-lung VEntilation (iPROVE) Network and their affiliations:

Marina Soro (1), Carmen Unzueta (2), Andrea Brunelli (3), Natividad Pozo (4), Alicia Llombart (5), Irene León (1), Cesar Aldecoa (6), Tania Franco (7), Francisco J Redondo (8), Amalia Alcón (9), Jose I. García-Sánchez (10), Maite Ibáñez (11), Manuel Granell (12), Aurelio Rodríguez-Pérez (13), Francisco Sandín (14), Manuel de la Matta (15), Rafael González (16), Javier García (17), Samuel Hernández (18), Francisco Barrios (19), Lucas Rovira (1), Patricia Piñeiro (20), Nuria García (21), Javier Belda (1,22), and Carlos Ferrando (9, 23).

(1) Hospital Clínico Universitario, Valencia, Spain. (2) Hospital Universitario Sant Pau, Barcelona, Spain. (3) Hospital Universitario Germans Trias i Pujol, Badalona, Spain. (4) INCLIVA Clinical Research Institute, Hospital Clinico Universitario de Valencia, Valencia, Spain. (5) Clínica Corachán, Barcelona, Spain. (6) Hospital Universitario Río Hortega, Valladolid, Spain. (7) Hospital Universitario Ramón y Cajal, Madrid, Spain. (8) Hospital General Universitario de Ciudad Real, Ciudad Real, Spain. (9) Hospital Clínic i Provincial, Barcelona, Spain. (10) Fundación Hospital Alcorcón, Alcorcón, Spain. (11) Hospital de la Marina Baixa, Vila Joiosa, Alicante, Spain. (12) Consorcio Hospital General Universitario, Valencia, Spain. (13) Hospital Universitario Doctor Negrín, Las Palmas de Gran Canaria, Spain. (14) Hospital Universitario Miguel Servet, Zaragoza, Spain. (14) Hospital Universitario Miguel Servet, Zaragoza, Spain. (15) Hospital Universitario Virgen del Rocio, Sevilla, Spain. (16) Hospital Universitario de León, León, Spain. (17) Hospital Universitario Puerta de Hierro, Majadahonda, Madrid, Spain. (18) Hospital Universitario Nuestra Señora de Candelaria, Santa Cruz de Tenerife, Spain. (19) Hospital Universitario Príncipe de Asturias, Oviedo, Spain. (20) Hospital Universitario Gregorio Marañón, Madrid, Spain. (21) Hospital Universitario Politécnico La Fe, Valencia, Spain. (22) Department of Surgery, Facultad de Medicina, Universidad de Valencia, Valencia, Spain. (23) CIBER de Enfermedades Respiratorias. Instituto de Salud Carlos III, Madrid, Spain.