At last! A quality ICU study!

Journals Reviewed: NEJM May 2000
Abstracted by: Dr J van Schalkwyk; FCP (Critical Care) (Consultant, University of the Witwatersrand)

Summary of abstract

All too often, we encounter studies in critical care that are tiny, poorly designed, and equivocal. The ARDSnet study appears to be quite different. Read on! (You may wish to briefly browse our editorial comment ).

Ventilation with lower tidal volumes in Acute Lung Injury


The Acute Respiratory Distress Syndrome (ARDS, née Adult Respiratory Distress Syndrome) is an acute clinical syndrome characterised by bilateral pulmonary infiltrates on frontal chest x-ray, arterial hypoxaemia (the cut-off being taken as a PaO 2 /FiO 2 ratio of under 200), all not attributable to left atrial or pulmonary capillary hypertension. The American-European Consensus Conference is well worth re-reading from time to time (Am J Respir Crit Care Med 1994 149 818-24) - note that although the table provided in that document states that pulmonary capillary wedge pressure must be <= 18 mmHg, or there must be no clinical evidence of left atrial hypertension, in the text the authors state that the clinical findings of ARDS "cannot be explained by, but may coexist with, left atrial or pulmonary capillary hypertension"!

Intensivists have long suspected that our ventilatory intervention in such ill patients is far from benign. A variety of supportive strategies have been put forward to support ARDS patients while they recover, but studies of these approaches have been dogged by controversy, and results have generally been equivocal. This most recent study is thus timeous and should be welcomed by all physicians keen on not harming their ventilated, hypoxic and critically ill patients.

The Study

This is a substantial and well-designed study, which was terminated prematurely when it became clear that a large-volume ventilatory strategy was harming those subjects subjected to this conventional ventilatory approach. Patient selection and management appears impeccable, and the number of patients in this multi-centre trial (861) are impressive. Although two other ARDS-related trials were conducted simultaneously on the same patients (ketoconazole versus placebo, and lisofylline versus placebo) these do not appear to have biased the study in any way. (As an aside, the ketoconazole showed no benefit, and the lisofylline trial is awaited).

Certain definitions within the protocols deserve close examination. Wisely, the authors related ventilatory parameters to a predicted body weight. This is appropriate because critically ill patients often have weights that are grossly divergent from their predicted weight, while one would expect pulmonary volumes to correlate well with predicted weight.

The main difference between the two ventilatory strategies was (as indicated in the title of the article) to use different tidal volumes - 6 ml/kg in the 'experimental' group versus 12 ml/kg in the 'conventional' group. To prevent outrageous ventilatory parameters, pause pressures in the conventional group were limited to 50 cmH 2 O. In the study group, pressures were limited to 30 cmH 2 O. In addition, in no patient was the tidal volume permitted to drop under 4 ml/kg. In the study group, administration of sodium ions (as sodium bicarbonate) was permitted to correct pH abnormalities.

A bold step was the acceptance in the study patients of PaO 2 s of 55 mmHg or more, and arterial saturation of 88% or more, rather than more conventional higher values.

Two 'primary' outcomes were assessed:

  • death before discharge;
  • ventilator-free days.
Other variables that were assessed were days without organ failure, evidence of barotrauma, and plasma interleukin-6 levels.

A considerable merit of the study is that the entire protocol is available on the internet.

Another interesting wrinkle is that in centres at high altitude (over 1000m), criteria for ARDS/ALI were adjusted accordingly (by multiplying the PaO 2 /FiO 2 by barometric pressure in mmHg divided by 760) This makes sense in terms of appropriately assessing the severity of the pulmonary insult.

Study Results

Baseline characteristics of the two groups were similar. There were significant differences in outcome between the two groups. Of greatest interest was the 22% reduction in mortality seen with the small volume ventilation group, when compared with the conventionally ventilated group. In addition, the other primary end-point reached significance - with significantly fewer ventilator days in the 'trial' group. A precis of their Table 1 is worth reproducing:

  Variable Group Receiving Lower Tidal volumes Group Receiving Traditional Tidal volumes P value
Death before discharge (%) 31.0 39.8 0.007
Breathing without assistance by day 28 (%) 65.7 55.0 <0.001
Ventilator-free days 12 ± 11 10 ± 11 0.007
Days without failure of non-pulmonary organs days 1 to 28 15 ± 11 12 ± 11 0.006

The respiratory measurements in the two groups were substantially different. As expected, tidal volumes were close to 6 ml/kg in the study group as compared with 12 ml/kg in controls. Peak and plateau pressures were of course lower in the former group.

Not so intuitively obvious were the other parameters recorded. Mean airway pressure was the same in the two groups. This was in part due to a slightly higher PEEP (9.4 cmH 2 O) in the study group compared to controls (8.6) cmH 2 O, despite the higher plateau pressures in the latter. Arterial pH was in the normal range in both groups, although slightly lower in the study group. Respiratory rates in the low volume group were nearly twice those in the conventional group (29/min vs 16/min).


There is a clear cut benefit of lower tidal volumes, despite a slightly lower arterial pH, slightly higher PEEP requirements, and faster respiratory rates in this group when compared with the group that was ventilated in a traditional fashion. There do not appear to be confounding variables that might explain away this difference.

The discussion provided by the authors is coherent and appropriate. Their analysis and explanation of the failure of other studies to achieve the same results is well worth reading. They draw particular attention to their correction of acidosis (by using fast ventilator rates, and NaHCO3 where required), and how this variable may have confounded other studies.

The finding by the researchers that IL-6 levels were lower in the study group provides some justification for the improved outcome in these patients. The hypothesis that traumatic ventilation of critically ill lungs with large volumes results in systemic inflammation and poor outcomes appears to be supported, as it has been by so much animal work.

It is a real pleasure to read an intensive care study that reached its primary end points, where the authors didn't have to resort (as so many other ICU studies do) to sub-analysis and statistical fiddling to achieve something like significance. Quite frankly, this study took my breath away! Surely it is now medically indefensible to ventilate patients with ARDS with volumes substantially greater than 6 ml/kg?

Article 1: Ventilation with Lower Tidal Volumes as Compared with Traditional Tidal Volumes for Acute Lung Injury and the Acute Respiratory Distress Syndrome.
New Engl J Med 2000 May 342 (18) 1301-9
Article type: Clinical study
Authors: The Acute Respiratory Distress Syndrome Network (Brower RG et al)

Editorial quibbles!

Before we discuss the above review, I feel it is appropriate to consider the editorial that accompanied the excellent ARDSnet study.

I find this editorial most unfortunate. Perhaps most startling is the statement "Survival increased only when patients treated with conventional or traditional tidal volumes had mean plateau pressures of more than 32 cm of water" [italics ours]! One would hope that in a journal of this reputation, editorial typographical errors (which we assume the above is) of this magnitude would be avoided.

There are other problems with the editorial. Dr Tobin conjures up the magical number of a plateau pressure under 32 cm of water, dismissing chest wall compliance with the comment "Plateau pressure is influenced not only by the lungs but also by the chest wall.."! Such a brief dismissal is simplistic - for we know from the few studies that have addressed the matter that inter- and intra-individual variations in the chest wall compliance of patients in ICU are substantial. If this is the case, then patients with identical pause pressures may have vastly different trans-alveolar pressures, and consequently greater or lesser tendencies to develop ventilator-induced lung injury. The primary intervention in the ARDSnet study was one of limiting tidal volumes, and this is surely what we should aim for (in the absence of evidence to the contrary).

Dr Van Schalkwyk's assessment of the study is perhaps a shade too enthusiastic. Although the study is indeed excellent, a few minor quibbles might be raised:

  • The ventilatory protocol is rather complex, perhaps unnecessarily so.
  • The additional limitation that pressures should not exceed 30cmH 2 O in the study group muddies the water a little, especially as the authors do not indicate in what percentage of study patients (or what proportion of the time) volumes had to be dropped to under 6 ml/kg in order not to exceed this critical pressure limit. It is however unlikely that such an intervention contributed substantially to the results in view of the close adherence to 6 ml/kg reflected in the readings recorded on days 1, 3 and 7.
  • Although the authors are to be commended for making their entire protocol available on the internet, it would have been wonderful if they could have provided the raw data resulting from their study at the same site. Far too many studies in the past have failed to provide raw data, making subsequent assessment a problem, years down the line.
  • The web-site contains the protocol information, but more attention could have been paid to the design of the site. Blind application of automated tools (to convert LaTeX to HTML) has not enhanced the quality of the documents.

The above are small issues. Overall the study is brilliant, and should substantially change our ICU practice. Undoubtedly there will be physicians who bash on with conventional ventilation out of ignorance or bloody-mindedness, but we hope that even they will ameliorate their ventilatory insults.

An important (and often life-saving maneuver) not addressed by the study is that of lung recruitment. We look forward to studies that address this issue, which could conceivably have even more impact on outcome in 'ARDS' than use of low tidal volumes. Prone positioning of patients is likewise not addressed.

One small point in conclusion. The study does not consider the issue of patients who have been inappropriately ventilated with large volumes, have sustained ventilator-induced injury, and are then changed to a low-volume strategy. It would seem prudent to tread carefully with such cases, where stepping in at the last minute with 6 ml/kg could conceivably not have the same desirable consequences as starting this gentle strategy right from the start.


scopes.gif (3289 bytes)