Prone positioning of patients with acute respiratory failure was first suggested over 30 years ago. In the present issue of Clinical Science, Reutershan and co-workers have studied the changes in end-expiratory lung volume in 12 patients with ARDS (acute respiratory distress syndrome) over an 8 h period following manual turning from the supine to prone position. From the data presented, the authors suggest that baseline end-expiratory lung volume could be used to identify responders, and serial measurements would permit appropriate ‘dosing’ of the therapy. Although this is an interesting study that provides data that have rarely been collected when assessing the response to prone positioning, there are a number of limitations that need to be considered. However, despite the limitations, the study does stimulate a number of important questions related not only to the use of the prone position, but also to the management of patients with ARDS in general.
Prone positioning of patients with acute respiratory failure was first suggested over 30 years ago . Although studies have repeatedly shown improvement in arterial oxygenation in approx. two-thirds of patients with ARDS (acute respiratory distress syndrome) , no clear outcome benefit has been demonstrated, largely because nearly all the studies have been small, directed at discovering the mechanism of the benefit and at attempting to predict which patients would benefit.
Central to these studies has been the assumption that improvement in oxygenation would equate with outcome benefit. The only two large randomized controlled studies of prone positioning in ALI (acute lung injury) from Italy  and France  showed the anticipated improvement in oxygenation, but there was no survival benefit. Both of these studies have been criticized on the grounds of their design, the heterogeneity of the patients enrolled, protocol violations, for only providing prone position for brief periods (≤8 h/day) and for failing to adjust the ventilator settings in response to the improvement in gas exchange (not ‘using the gain’). A recent Spanish study  achieved 17 h/day in the prone position, did ‘use the gain’ to modify the ventilator settings [reduce FIO2 (fraction of inhaled oxygen) and PEEP (positive end-expiratory pressure)] and demonstrated a 25% relative reduction in ICU (Intensive Care Unit) mortality and a 19% reduction in hospital mortality, but only 136 patients were enrolled and this did not achieve conventional significance (P=0.12). However, the prone group had a higher severity of illness score as reflected by the simplified acute physiology score (SAPS II) at inclusion. Multivariate analysis showed that the days from the diagnosis of ARDS to study entry, the SAPS II score and randomization to the supine position were all independent risk factors for mortality.
In the present issue of Clinical Science, Reutershan and co-workers  have studied the changes in EELV (end-expiratory lung volume) in 12 patients with ARDS over an 8 h period following manual turning from the supine to prone position. To their credit the authors  give a clear description of how the patients were supported when in the prone position. Since change in thoraco-abdominal compliance is a major factor in the change in distribution of ventilation that results from proning, the way in which the thorax and abdomen are supported can be of great significance. EELV was measured by manually ventilating the patient on a rebreathing bag containing 1% (v/v) sulphur hexafluoride in oxygen and values were derived for relative FRC (functional residual capacity; EELV/predicted FRC). Even in this small study , the baseline data reveal a group of patients that are heterogeneous in regard to the aetiology of lung injury (pulmonary compared with extrapulmonary), the time from onset of ARDS and the level of PEEP used. There was also a surprising variation in the FRC, with the majority showing the expected significant volume loss, but with some patients at or above predicted values. Eight patients (66%) were defined as ‘responders’ on the basis of ≥30% increase in oxygenation index [PaO2 (arterial partial pressure of oxygen)/FIO2], but the time course of this improvement and the associated increase in EELV was variable, with three patients achieving maximum recruitment within a few hours, whereas in the other five patients recruitment continued through the entire proning period. Of the baseline parameters measured, it was only EELV that predicted the gas exchange response to proning, but the time course of the changes was patient-specific and not predictable. The authors  suggest that baseline EELV could be used to identify responders and serial measurements would permit appropriate ‘dosing’ of the therapy.
This is an interesting study that provides data that have rarely been collected when assessing the response to prone positioning, but some issues need to be highlighted. The technique described to measure EELV is neither practical in routine clinical practice nor arguably desirable in view of the risk of de-recruitment associated with removing patients with severe lung injury from the ventilator, notwithstanding the authors'  practice of clamping the endotracheal tube at end-expiration in an attempt to maintain PEEP. The patients were also paralysed for the procedure. The fact that duplicate measurements showed no difference between the first and second measurement does not necessarily exclude the possibility that the procedure influenced the measurement, only that the effect, however large or small, is consistent. Most of the previous studies that have explored the baseline characteristics that may be predictive of benefit from proning have used a lesser (approx. 20%) improvement in oxygenation to define a ‘responder’. As an observational study  it is understandable that the clinically determined ventilator settings were not changed during the study period, but the tidal volumes used were consistently above the 6 ml/kg demonstrated to have survival benefit in the ARDSNet study  published in 2000, and the higher levels of PEEP used in some of the patients would not be supported by a more recent study from the same group . Furthermore, improvements in gas exchange achieved in the responders were not used to permit more lung-friendly ventilation by reducing the inspired oxygen concentration, tidal volume and PEEP. Although no mortality benefit would be expected in such a small study, it is interesting to note that survival was higher (75%) in the ‘non-responders’ than in the ‘responders’ (63%).
Despite its limitations, this study  stimulates a number of important questions related not only to the use of the prone position, but also to our management of patients with ARDS in general. (i) How should we determine the ‘optimum’ ventilator settings particularly with regard to PEEP? Should we be guided by gas exchange benefit alone or, as the evidence increasingly suggests, beware of producing high airway pressures and overinflating normal lung? Would measurement of EELV be an appropriate way of assessing the need for and the impact of recruitment manoeuvres? (ii) Which measurement should be used to define a ‘responder’ to a therapy such as proning and which intermediate measurements are the best surrogates for outcome benefit? If we become obsessed with an inappropriate intermediary endpoint the assessment of new therapies and the design of subsequent outcome studies will be seriously compromised. Could our fixation on improvement in indices of oxygenation be flawed? A previous therapy for ARDS, inhaled nitric oxide, produced dramatic improvements in gas exchange in the short term, but proved to have no long-term benefit  and may even have been harmful. The concept of VILI (ventilator-induced lung injury) has now focussed our management not only on avoiding mechanical trauma due to high alveolar pressures and lung overdistension, but also on avoiding ventilatory strategies that may cause increased cytokine release, which may produce distant organ damage and contribute to multiple organ failure [10,11]. Surely any gain in gas exchange obtained with proning should be used to reduce both the inspired oxygen concentration and PEEP to the level at which maximum recruitment of collapsed lung is achieved, but without causing damaging overdistension of more normal lung. (iii) Should we rely on large randomized controlled studies to decide how to manage patients with ARDS and thereby adopt a ‘one size fits all’ policy ignoring the heterogeneity of this patient population or attempt to tailor management to the individual patient using measurements such as EELV?
The recently introduced technique of electrical impedance tomography promises the opportunity of the bedside measurement not only of lung volumes, but also the distribution of ventilation and perfusion within the lung in the ventilated patient and may prove to be a more feasible way of collecting the data that this paper  suggests are important in assessing and titrating the use of prone positioning and other therapies that will hopefully improve the outcome from ARDS.