The acute respiratory distress syndrome: definitions, severity and clinical outcome. An analysis of 101 clinical investigations

Intensive Care Med. 1996 Jun;22(6):519-29. doi: 10.1007/BF01708091.


Objective: To determine possible changes in outcome from acute respiratory distress syndrome (ARDS) and to compare severity of lung injury and methods of treatment from 1967 to 1994.

Data sources: Computerized (Medline, Current Contents) and manual (Cumulated Index Medicus) literature search using the key word and/or title ARDS.

Study selection: Only clinical studies published as full papers reporting data on both patient mortality (survival) and oxygenation index (PaO2/FIO2) were included. Single case reports, abstracts, reviews and editorials were excluded from evaluation.

Data extraction: Relevant data were extracted in duplicate, followed by quality checks on approximately 80% of data extracted.

Data synthesis: 101 papers reporting on 3264 patients were included: 48 studies (2207 patients) were performed in the USA, 43 studies (742 patients) in Europe and 10 studies (315 patients) elsewhere. Mortality reported in these studies was 53 +/- 22% (mean +/- SD), with no apparent trend towards a higher survival (1994: 22 studies, mortality 51 +/- 19%). The mean PaO2/FIO2 ratio remained unchanged throughout the observation period (118 +/- 47 mmHg). No correlation could be established between outcome and PaO2/FIO2 or lung injury score. Patients who underwent pressure-limited ventilation had a significantly lower mortality (35 +/- 20%) than patients on volume-cycled ventilation (54 +/- 22%) or patients for whom there was no precise information on ventilatory support (59 +/- 19%). Significantly lower PaO2/FIO2 ratios (61 +/- 17 mmHg) were observed in patients prior to extracorporeal lung assist, together with mortality rates in the range of those for conventionally treated patients (55 +/- 22%).

Conclusions: The mortality of ARDS patients remained constant throughout the period studied. Therefore, the standard for outcome in ARDS should be a mortality in the 50% range. Neither PaO2/FIO2 ratio nor lung injury score was a reliable predictor for outcome in ARDS. Patients might benefit from pressure-limited ventilatory support, as well as extracorporeal lung assist. Since crucial data were missing in most clinical studies, thus preventing direct comparison, we emphasize the importance of using standardized definitions and study entry criteria.

Publication types

  • Meta-Analysis

MeSH terms

  • Analysis of Variance
  • Humans
  • Mortality / trends
  • Oxygen / blood
  • Patient Selection
  • Respiration, Artificial / methods
  • Respiratory Distress Syndrome / classification*
  • Respiratory Distress Syndrome / mortality*
  • Respiratory Distress Syndrome / therapy
  • Severity of Illness Index
  • Statistics, Nonparametric
  • Treatment Outcome


  • Oxygen