Abstract
Purpose
The relative contributions of factors influencing lung injury immediately after birth are poorly understood. We hypothesized that oxygen content and humidity of inspired air would influence markers of pulmonary inflammation in ventilated lambs.
Methods
Lambs delivered at 140 days gestation (term = 150 days) were assigned to one of five groups (n = 5–6/group): unventilated controls, or ventilation with 21 or 100% O2 that was either heated and humidified or cold and dry. Lambs were ventilated gently for 3 h: blood gases were recorded regularly. Bronchoalveolar lavage and samples of tracheal mucosa and lung were collected post mortem.
Results
Arterial pH was lower [mean difference (95% CI): −0.07 (−0.13, −0.03)], while there was an increase in PaCO2 [mean difference (95% CI): 10.2 (2.4, 17.9)] and fold change in lung pro-inflammatory IL-1β cytokine mRNA [mean difference (95% CI): 28.3 (0.3, 56.2)] or IL-8 [mean difference (95% CI): 27.8 (7.9, 47.7)] cytokine mRNA expression with 100% O2 relative to 21% O2. Cold dry inspired gas did not influence gas exchange or dynamic mechanics at 3 h compared to heated humidified gas. Compared to 100% inspired O2, cold dry inspired gas had less marked effect on fold change in lung pro-inflammatory IL-1β cytokine mRNA [mean difference (95% CI): 27.2 (−0/8, 55.1)] or IL-8 [mean difference (95% CI): 14.5 (5.5, 34.4)] cytokine mRNA expression, although cilial dysfunction/damage was evident on electron microscopy, especially when exposure to cold dry gas was combined with hyperoxia.
Conclusions
In near-term neonatal lambs ventilated for 3 h, hyperoxia was associated a more powerful stimulus for pulmonary dysfunction and upregulation of inflammatory cytokines than cold dry gas.
Similar content being viewed by others
Introduction
Avoidance of hyperoxia during ventilation of newborns has attracted considerable attention over the past 60 years. Whereas attention in the 1950s was focused on avoidance of retinopathy of prematurity [1], there has been subsequent recognition of the potential role of oxidative stress in the pathogenesis of other diseases, including bronchopulmonary dysplasia [2], necrotizing enterocolitis [3], periventricular leukomalacia [4] and dysfunction of regulatory processes such as reactivity of the ductus arteriosus and pulmonary vessels [5, 6]. More recently, the timing of hyperoxic exposure has been addressed, with the adverse effects of hyperoxia during fetal–neonatal transition and resuscitation as the focus of attention [7, 8]. Much remains to be understood about the time frame and conditions under which these adverse effects take place in the newborn.
Gas conditioning may influence injury to the respiratory system in the first few hours of life. The initiation of ventilation in the newborn infant is often undertaken in the delivery room without heating and humidification of the ventilatory gas, and this may be continued for some hours if transport to a NICU is required. Current International Liaison Committee on Resuscitation (ILCOR) guidelines [9] do not mention gas conditioning. Apart from commentaries that state “there is a strong physiologic rationale for delivering the inspiratory gas at or close to core body temperature and saturated with water vapor to infants who have an artificial airway undergoing longer term mechanical ventilatory assistance” [10], very little information exists to identify the effects of over/under humidification and heating of the inspired gas in the newborn infant. We hypothesized that hyperoxia and exposure to cold dry gas would be associated with upregulation of markers of inflammation in the airways and lung parenchyma of the near term newborn lamb.
Methods
Investigations were approved by the animal ethics committees of the University of Western Australia, the Department of Agriculture and Food of Western Australia and Cincinnati Children’s Hospital Medical Center. Studies were performed at the Medina Research Station in Western Australia.
Animals, delivery and postnatal care
A cesarean section was performed on anesthetized, date-mated Merino ewes bearing singletons or twins at 140 days gestation (term is ≈150 days). The fetal head was exteriorized, and, following tracheostomy, the fetus was intubated with a 4.5-mm cuffed tracheal tube, and fetal lung fluid was removed. The fetus was delivered, weighed, dried and randomized to tissue collection [unventilated controls (UVC), n = 5] or to one of four ventilation groups (n = 6/group, Table 1): (1) fractional inspired oxygen (F i O2) of 0.21 and heated humidified gas, (2) F i O2 0.21 and cold dry gas, (3) F i O2 1.0 and heated humidified gas or (4) F i O2 1.0 and cold dry gas. Intermittent positive pressure ventilation was commenced using a low tidal volume (V T 7–8 ml/kg). Initial ventilator settings consisted of a peak inspiratory pressure (PIP) of 25 cmH2O, positive end-expiratory pressure (PEEP) of 5 cmH2O, rate of 40/min and inspiratory time (t I) of 0.6 s. PIP was adjusted to target a V T of 7–8 ml/kg and an arterial partial pressure of carbon dioxide (PaCO2) of 50–60 mmHg, for a total of 3 h. Heating and humidification of gas were achieved using an MR850 humidifier (Fisher & Paykel Healthcare, New Zealand). UVC fetuses were euthanized with pentobarbitone overdose (100 mg/kg pentobarbitone IV) immediately prior to delivery and post-mortem tissue harvest.
Ventilated lambs were positioned prone, covered with transparent occlusive wrap (NeoWrap, Fisher & Paykel Healthcare, New Zealand) and kept warm on an infant radiant warmer (CozyCot, Fisher & Paykel Healthcare, New Zealand). Sedation and analgesia were achieved with propofol (0.1 mg/kg/h: Repose®, Norbrook Laboratories Ltd, Victoria, Australia) and remifentanil (0.05 μg/kg/h; Ultiva®, Glaxo Smith Kline Ltd, Victoria, Australia) by continuous infusion through an umbilical venous catheter.
Blood gas samples, obtained at regular intervals from an umbilical arterial catheter, were used to measure pH, partial pressure of oxygen (PaO2) and PaCO2. Tidal volume (V T) was monitored continuously (Florian, Acutronics, CH). At the end of the 3-h ventilation period, lambs were killed (100 mg Pentobarbitone IV) and exsanguinated prior to tissue collection. Ventilator efficiency index (VEI) was calculated as 3,800/∆P·f·PaCO2 where ΔP is the change in pressure between inspiration and expiration, and f is the respiratory frequency.
Lung processing
The thorax was opened and the lungs were excised. Lung tissue from the right lower lobe was frozen in liquid nitrogen for cytokine mRNA measurements. Total RNA isolated from frozen lung and liver was used for RNAse protection assays using sheep-specific riboprobes for IL-1β, IL-6 and IL-8 [11–13]. Three repeated saline lavages of the left lung were performed using 0.9% NaCl at 4°C and combined for bronchoalveolar lavage (BAL) [14]. BAL aliquots were saved for cell counts and measurement of protein quantified with the Bradford method [15] adapted for micro-titer plate based assays [16].
Data analysis and statistics
Data were analyzed using SPSS v14.0 (SPSS Inc, USA). One-way ANOVA or ANOVA on ranks were used to compare ventilated groups to the unventilated controls; where significant differences were found, post hoc analysis was performed using Tukey’s multiple comparison test. A mixed-model linear regression analysis was used to assess the effect of F i O2 and temperature and humidity (cold dry vs. heated humidified), with body temperature included as a covariate. Significance was accepted for P < 0.05. Comparisons between subgroups are displayed as mean difference (95% CI) for F i O2 0.21 versus 1.0 or cold and dry versus heated and humidified gas.
Results
Physiological measurements
The baseline characteristics (birth weight, cord blood gas variables) of the groups of animals were not different (Table 1).
Figure 1 illustrates differences at the completion of the study in pH, PaCO2 and tidal volume for lambs exposed to F i O2 0.21 or F i O2 1.0, with or without heating and humidification of the inspired gas. Lambs exposed to a F i O2 of 1.0 had lower pH and higher PaCO2 despite a slightly larger V T than lambs ventilated with F i O2 of 0.21. Statistical measures [mean differences (95% CI)] for arterial blood gas, ventilator and dynamic mechanical variables are summarized in Table 2. Lambs exposed to F i O2 of 1.0 had higher PaO2 compared to those exposed to F i O2 of 0.21. There were no significant effects of gas conditioning on measures of gas exchange or dynamic mechanics at 3 h compared to heated humidified gas. There were no significant interactions between O2 content and gas temperature/humidity.
Tracheal and lung tissue cytokine mRNA
Figure 2 shows mRNA expression for IL-1β in the trachea (Fig. 2a) and lung tissue (Fig. 2b). Two-way multiple linear regression results are tabulated in the electronic supplementary material (ESM) Table 1 (trachea) and ESM Table 2 (lung tissue). In the trachea and lung tissue, the IL-1β, IL-6 and IL-8 mRNA levels were increased in all ventilated groups compared to UVC. Exposure to an FIO2 of 1.0 for 3 h resulted in increased IL-1β and IL-8 mRNA expression in the lung tissue relative to the FIO2 0.21 group. There was a strong trend towards increased IL-1β mRNA expression in the lung tissue for animals exposed to cold dry gas relative to warm humidified gas. There were no statistically significant effects of either oxygen or gas conditioning on IL-1β and IL-8 mRNA expression in the trachea.
Tracheal mucosa
Representative scanning electron micrographs of tracheal cilia obtained from the proximal trachea of lambs in each of the study groups are shown in Figure 1 in the ESM. Denudation of the cilia was most evident in lambs ventilated with cold, dry gas and an FIO2 of 1.0.
Bronchoalveolar lavage
There were few inflammatory cells in the BAL of ventilated groups (ESM Figure 2). There was a trend (P = 0.056) towards differences in inflammatory cells between groups on one-way ANOVA; however, there were no differences between UVC and ventilated groups for BAL total protein. An increased proportion of inflammatory cells were observed in the heated humidified group compared to cold dry gas groups; however, this difference was predominantly due to increased presence of inflammatory cells in the FIO2 1.0 heated humidified group (ESM Table 3).
Discussion
Summary of findings
This study was designed to assess the relative contributions of O2 content and conditioning of inspired gas on lung physiology and airway and pulmonary tissue injury in near-term newborn lambs. Ventilation increased mRNA expression for the pro-inflammatory cytokines IL-1β and IL-8 in the tracheal mucosa relative to UVC; the increases in IL-1β, IL-6 and IL-8 mRNA expression were of a greater magnitude in lung tissue. Exposure to hyperoxia (FIO2 of 1.0) was associated with poorer gas exchange and increased expression of inflammatory cytokines in lung tissue, with a trend towards higher IL-1β mRNA expression in the tracheal mucosa. In contrast, we found that 3-h exposure to cold dry air had no effect on gas exchange or lung mechanics, and was associated only with a trend towards increased IL-1βmRNA expression in lung tissue.
The association between hyperoxia and the development of acute pulmonary injury and subsequent chronic lung disease is widely acknowledged [2]. Hyperoxic lung injury is characterized by severe endothelial damage, alveolar epithelial injury and increased pulmonary permeability [17], in addition to upregulation of IL-1β, IL-6 and IL-8, which precedes the influx of inflammatory cells [17] and lung tissue damage. Whereas we observed increases in cytokine mRNA, the increase in BAL protein was not significantly different between groups in this study, suggesting that increased pulmonary permeability may not have been present. Hyperoxic neonates can survive longer than adult animals, with a later onset of inflammation [17]; however, ventilation appears to increase susceptibility to the effects of hyperoxia.
We targeted mild hypercapnia (45–55 mmHg) and tidal volumes of 7–8 ml/kg during ventilation of the lambs. A small but statistically significant difference in V T between hyperoxic and air-exposed lambs was evident at 3 h as a result of our strategy to change tidal volume in response to arterial blood gases with the goal of achieving a target PaCO2 of 45–55 mmHg. Despite the increased tidal volume (and hence also minute volume given the constant respiratory rate), PaCO2 was significantly higher in the hyperoxic groups as target PaCO2 could not be achieved within our V T upper limit of 10 ml/kg. It is unlikely that the increased tidal volume reflects a significant increase in compliance, as there was no significant change in PIP, dynamic compliance or VEI. Whereas cyclic volutrauma contributes to the development of lung injury [18], the absolute difference in V T was small.
The higher minute volume in the hyperoxic groups, compared to the air groups was not matched by improved ventilatory efficiency. This suggests the hyperoxic lambs had impaired CO2 exchange, indicating either reduced alveolar minute ventilation or increased ventilation-perfusion mismatch. Given that there were no differences between groups in the tidal volume over the first 30 min of life (data not shown), the need for an increased V T to achieve target PaCO2 reflects impairment to CO2 removal that evolved over the period of the injury exposure. This is consistent with the increased cytokine mRNA levels in lung tissue and the observed trend towards increased inflammatory cells in the BAL fluid in the hyperoxic groups. However, the removal of CO2 at rest is normally perfusion limited, and acute hyperoxia reduces pulmonary perfusion in awake sheep [19]; hence, ventilation/perfusion mismatching is the likely underlying explanation for the increased ventilation requirement in the hyperoxic groups.
The effect of an FIO2 of 1.0 was most marked in the lung tissue, but we saw a similar trend towards increased expression of proinflammatory cytokine mRNA in the tracheal mucosa, demonstrating the inflammatory effect of hyperoxia combined with ventilation throughout the respiratory tree. Compared to the lung, however, the magnitude of increases in tracheal mucosa pro-inflammatory cytokine mRNAs was modest. Taken together with recent observations of injury to the small airways in association with brief (15 min) volutrauma [20], our findings suggest that the distal respiratory system is more susceptible to hyperoxic injury than the proximal airways.
Although the increase in inflammatory markers was modest in the tracheal tissue, denuded and flattened cilia were more evident in the hyperoxic groups compared to those ventilated in air (ESM Figure 1), suggesting a direct effect of an FIO2 of 1.0 on cilial integrity. It was interesting to note, however, that the most marked effect on the cilia was seen in the hyperoxic group treated with cold, dry air.
Preconditioning of the inspired gas is thought to be important during mechanical ventilation, because the tracheal tube bypasses the upper respiratory tract. The temperature and humidity of the gas inspired during ventilation affects a number of different airway functions including: gas exchange (lung mechanics, gas conditioning), conservation (heat and moisture recovery) and airway defense (clearance) [21]. Heat and moisture are lost from the mucosa to the inspired air with each breath, until the air reaches core body temperature and 100% relative humidity. The point in the lung at which this condition is reached (isothermic saturation boundary) may be influenced by the initial condition of the inspired gas [22], as well as the pattern of ventilation [23–25], and disease states [26]. It follows that the amount of mucosa suffering a heat and water deficit, and the magnitude of that loss will also vary according to these variables [21]: only about 25% of the heat and moisture is recovered by the mucosa on expiration [27]. In adult males breathing quietly, the isothermic saturation boundary is reached at the level of the main bronchus [21]. It is therefore normal in the adult for the upper airway to be in a water and heat energy deficit [21].
The newborn infant may be exposed to cold dry air for brief periods (~15 min) during resuscitation and prior to arrival in the NICU, a practice supported by current resuscitation guidelines [9]. It is common for neonates to be exposed to cold dry air for more extended periods during transport. The use of heat and moisture exchangers to avoid heat and water loss from the respiratory tract has not been widely used for neonates due to problems associated with increased deadspace, but recent reports of (at least partial) effectiveness are emerging [28]. Mucosal damage and acute inflammation associated with inhalation of cold dry air have been reported in several adult animal species [29–31]. The effect of this exposure on airway and parenchymal injury in the respiratory system of the neonate has not been studied extensively: the optimal temperature and relative humidity of inspired gas in ventilated infants remains controversial [10].
We observed a strong trend to increased IL-1β in lung tissue after 3 h of ventilation with cold dry gas. However, non-conditioning of the gas did not result in abnormal gas exchange or pulmonary mechanics. Our study duration was relatively short, and we cannot rule out an effect of non-conditioning of the gas on lung tissue injury. Nonetheless, it is clear that hyperoxia is the more potent injury stimulus for the peripheral lung tissue.
We saw evidence of impaired cilial integrity in the trachea. The mucociliary transport system is considered particularly sensitive to inhalation of cold dry air [32], and others have shown that damage to the mucocilliary layer in dogs occurs as soon as 60 s after exposure to unheated and unhumidified gas [33]. In contrast, there was no evidence of upregulation of inflammatory cytokines in the tracheal mucosa that was independently associated with the administration of cold dry air. Given that IL-1β is upregulated before the other interleukins, it may be that longer periods of exposure of cold dry gas (greater than 3 h) are required before substantive inflammatory changes occur in the proximal airways.
Conclusions
Over a 3-h ventilation period, hyperoxia was a more powerful stimulus for the upregulation of inflammatory cytokine mRNA in lung tissue than ventilation with cold dry gas. Hyperoxia was associated also with a need for slightly increased ventilation to avoid hypercapnia, which is consistent with increased deadspace and/or ventilation perfusion imbalance. The combination of an FIO2 of 1.0 and cold dry inspired gas had a marked effect on cilial integrity in sections of the proximal trachea. Information about the effects of cold dry gas on the respiratory system of newborns is lacking and warrants further research.
References
Ashton N, Ward B, Serpell G (1954) Effect of oxygen on developing retinal vessels with particular reference to the problem of retrolental fibroplasia. Br J Ophthalmol 38:397–432
Saugstad OD (2003) Bronchopulmonary dysplasia—oxidative stress and antioxidants. Semin Neonatol 8:39–49
Saugstad OD (1988) Hypoxanthine as an indicator of hypoxia: its role in health and disease through free radical production. Pediatr Res 23:143–150
Haynes RL, Folkerth RD, Keefe RJ, Sung I, Swzeda LI, Rosenberg PA, Volpe JJ, Kinney HC (2003) Nitrosative and oxidative injury to premyelinating oligodendrocytes in periventricular leukomalacia. J Neuropathol Exp Neurol 62:441–450
Archer SL, Peterson D, Nelson DP, DeMaster EG, Kelly B, Eaton JW, Weir EK (1989) Oxygen radicals and antioxidant enzymes alter pulmonary vascular reactivity in the rat lung. J Appl Physiol 66:102–111
Clyman RI, Saugstad OD, Mauray F (1989) Reactive oxygen metabolites relax the lamb ductus arteriosus by stimulating prostaglandin production. Circ Res 64:1–8
Saugstad OD, Ramji S, Vento M (2006) Oxygen for newborn resuscitation: how much is enough? Pediatrics 118:789–792
Vento M, Asensi M, Sastre J, Garcia-Sala F, Pallardo FV, Vina J (2001) Resuscitation with room air instead of 100% oxygen prevents oxidative stress in moderately asphyxiated term neonates. Pediatrics 107:642–647
The International Liaison Committee on Resuscitation (ILCOR) (2006) Consensus on science with treatment recommendations for pediatric and neonatal patients: pediatric basic and advanced life support. Pediatrics 117:e955–e977
Schulze A (2007) Respiratory gas conditioning and humidification. Clin Perinatol 34:19–33
Wilson TC, Bachurski CJ, Ikegami M, Jobe AH, Kallapur SG (2005) Pulmonary and systemic induction of SAA3 after ventilation and endotoxin in preterm lambs. Pediatr Res 58:1204–1209
Chomczynski P, Mackey K (1995) Modification of the TRI reagent (TM) procedure for isolation of RNA from polysaccharide- and proteoglycan-rich sources. Biotechniques 19:942–945
Kallapur SG, Willet KE, Jobe AH, Ikegami M, Bachurski C (2001) Intra-amniotic endotoxin: chorioamnionitis precedes lung maturation in preterm lambs. Am J Physiol 280:L527–L536
Jobe AH, Kramer BW, Moss TJ, Newnham JP, Ikegami M (2002) Decreased indicators of lung injury with continuous positive expiratory pressure in preterm lambs. Pediatr Res 52:387–392
Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding. Anal Biochem 72:248–254
Cheah FC, Jobe AH, Moss TJ, Newnham JP, Kallapur SG (2008) Oxidative stress in fetal lambs exposed to intra-amniotic endotoxin in a chorioamnionitis model. Pediatr Res 63:274–279
Bhandari V, Elias JA (2006) Cytokines in tolerance to hyperoxia-induced injury in the developing and adult lung. Free Radic Biol Med 41:4–18
Hillman NH, Moss TJ, Kallapur SG, Bachurski C, Pillow JJ, Polglase GR, Nitsos I, Kramer BW, Jobe AH (2007) Brief, large tidal volume ventilation initiates lung injury and a systemic response in fetal sheep. Am J Respir Crit Care Med 176:575–581
Melsom MN, Flatebo T, Nicolaysen G (1999) Hypoxia and hyperoxia both transiently affect distribution of pulmonary perfusion but not ventilation in awake sheep. Acta Physiol Scand 166:151–158
Polglase GR, Hillman N, Pillow JJ, Cheah FC, Nitsos I, Moss TJ, Kramer BW, Ikegami M, Kallapur SG, Jobe AH (2008) Positive end-expiratory pressure and tidal volume during initial ventilation of preterm lambs. Pediatr Res 64:517–522
Williams RB (1998) The effects of excessive humidity. Respir Care Clin N Am 4:215–228
Ingelstedt S (1956) Studies on the conditioning of air in the respiratory tract. Acta Otolaryngol Suppl 131:1–80
Caldwell PR, Gomez DM, Fritts HW Jr (1969) Respiratory heat exchange in normal subjects and in patients with pulmonary disease. J Appl Physiol 26:82–88
Ferrus L, Guenard H, Vardon G, Varene P (1980) Respiratory water loss. Respir Physiol 39:367–381
McFadden ER Jr, Pichurko BM, Bowman HF, Ingenito E, Burns S, Dowling N, Solway J (1985) Thermal mapping of the airways in humans. J Appl Physiol 58:564–570
Primiano FP Jr, Saidel GM, Montague FW Jr, Kruse KL, Green CG, Horowitz JG (1988) Water vapour and temperature dynamics in the upper airways of normal and CF subjects. Eur Respir J 1:407–414
Walker JE, Wells RE Jr, Merrill EW (1961) Heat and water exchange in the respiratory tract. Am J Med 30:259–267
Fassassi M, Michel F, Thomachot L, Nicaise C, Vialet R, Jammes Y, Lagier P, Martin C (2007) Airway humidification with a heat and moisture exchanger in mechanically ventilated neonates: a preliminary evaluation. Intensive Care Med 33:336–343
Barbet JP, Chauveau M, Labbe S, Lockhart A (1988) Breathing dry air causes acute epithelial damage and inflammation of the guinea pig trachea. J Appl Physiol 64:1851–1857
Ingenito EP, Pliss LB, Ingram RH Jr, Pichurko BM (1990) Bronchoalveolar lavage cell and mediator responses to hyperpnea-induced bronchoconstriction in the guinea pig. Am Rev Respir Dis 141:1162–1166
Van Oostdam JC, Walker DC, Knudson K, Dirks P, Dahlby RW, Hogg JC (1986) Effect of breathing dry air on structure and function of airways. J Appl Physiol 61:312–317
Williams R, Rankin N, Smith T, Galler D, Seakins P (1996) Relationship between the humidity and temperature of inspired gas and the function of the airway mucosa. Crit Care Med 24:1920–1929
Omori C, Schofield BH, Mitzner W, Freed AN (1995) Hyperpnea with dry air causes time-dependent alterations in mucosal morphology and bronchovascular permeability. J Appl Physiol 78:1043–1051
Acknowledgments
We gratefully acknowledge the technical assistance of Dr. Zahra Champion, Mr. Martin Leckie, Dr. Ilias Nitsos and Mr. Christian Saville in the delivery and care of the animals used in the study. This research was supported by HD 12714 from the National Institute of Health, USA, a NHMRC Career Development Awards (JJP: 404102; TJMM: 303261) and the Women and Infants Research Foundation. Ventilator circuits, humidifiers and radiant warmer beds used in the studies were provided by Fisher & Paykel Healthcare™ (New Zealand).
Conflict of interest statement
This research was supported in part by Fisher & Paykel Healthcare™ who market neonatal equipment including ventilator circuits, humidifiers and radiant warmer beds (Fisher & Paykel Healthcare, New Zealand), as used in the study. Representatives from Fisher & Paykel Healthcare assisted with care of the newborn lambs under the supervision of Prof. Jobe and Prof. Pillow. Fisher & Paykel were not involved in analysis or interpretation of the study measurements, or the decision to publish the data. They were given the opportunity to read the manuscript prior to submission; however, their input to the manuscript was limited to comments relating to description of their products.
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Pillow, J.J., Hillman, N.H., Polglase, G.R. et al. Oxygen, temperature and humidity of inspired gases and their influences on airway and lung tissue in near-term lambs. Intensive Care Med 35, 2157–2163 (2009). https://doi.org/10.1007/s00134-009-1624-z
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00134-009-1624-z