The authors are to be commended for their efforts given the limitations of these studies. What can we conclude from this systematic review? The answer is that com- bining a variety of studies with different diagnostic criteria, variable quality, differ- ent designs (matched and unmatched), and high degree of heterogeneity does not pro- duce results that we can be confident of. Furthermore, the finding that VAP was not associated with attributable mortality in the ARDS and trauma populations, in which there was little heterogeneity, should give rise to further thought. At the minimum, this systematic review should lead us to question accepted dogma and ask whether appropriately treated VAP does cause significant attributable mortality. In this respect, the debate continues (7). Because observational data in regard to the attributable mortality of VAP are all that we will ever have, studies of more methodological rigor are required if we are going to answer this important ques- tion. Case control studies need to be bet- ter matched for the prognostic factors of VAP outcome. No studies have conducted a propensity analysis of VAP mortality, and this may be useful in view of the large number of factors that may influence mortality (8). Furthermore, better defini- tion of the groups being compared, along with better diagnostic modalities, are re- quired. Without these, the controversy will continue without resolve. John Muscedere Queen’s University Kingston, Ontario, Canada REFERENCES 1. Schumacher M, Wangler M, Wolkewitz M, et al: Attributable mortality due to nosocomial infections: A simple and useful application of multistate models. Methods Inf Med 2007; 46:595– 600 2. Melsen W, Rovers M, Bonten M: Ventilator- associated pneumonia and mortality: A sys- tematic review of observational studies. Crit Care Med 2009; 37:2709 –2718 3. Stroup D, Berlin J, Morton S, et al: Meta- analysis of observational studies in epidemi- ology: A proposal for reporting. JAMA 2000; 283:2008 –2012 4. Guidelines for the management of adults with hospital-acquired, ventilator-associ- ated and healthcare associated pneumonia. Am J Respir Crit Care Med 2005; 171: 388 – 416 5. Kollef MH: Broad-spectrum antimicrobials and the treatment of serious bacterial infec- tions: Getting it right up front. Clin Infect Dis 2008; 47(suppl 1):S3–S13 6. Safdar N, Dezfulian C, Collard HR, et al: Clinical and economic consequences of ven- tilator-associated pneumonia: A systematic review. Crit Care Med 2005; 33:2184 –2193 7. Carlet J: Dying from or with a nosocomial pneumonia in the intensive care unit? Crit Care Med 2001; 29:2392–2394 8. Rubin DB: Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997; 127:757–763 Pulmonary artery catheter redux: Physical findings in acute respiratory distress syndrome/acute lung injury* U sing a physical examination to gain information that oth- erwise would require a pul- monary artery catheter (PAC) would avoid the cost and the complica- tions associated with a PAC. In lieu of a PAC, central venous pressure and central venous saturation (ScvO 2 ) obtained via a central venous catheter (CVC) might sup- plement the examination sufficiently to obviate any need for a PAC, although the cannulation complications are the same for both types of catheters. In this issue of Critical Care Medicine, Grissom and col- leagues (1) used data from patients with acute respiratory distress syndrome (ARDS) or acute lung injury (ALI) culled from the PAC arm of the Fluid and Cath- eter Treatment Trial (FACTT) (2), to test three hypothesis: 1) whether capillary re- fill time 2 secs, skin mottling over the knees, and cool extremities could predict a cardiac index (CI) 2.5 or a mixed venous saturation (SvO 2 ) 60%; 2) whether this prediction would be en- hanced by adding urine output and cen- tral venous pressure; and 3) whether ScvO 2 was a useful predictor of SvO 2 . Their results indicate that these three clinical findings, clinical findings (CF), and ScvO 2 “are not useful predictors of a low CI or low SvO 2 ” and that ScvO 2 could not reliably predict SvO 2 . These results engender three questions: 1) Could these results be spurious because of study de- sign or the patient population? 2) In view of the repeatedly negative results from studies of the contribution of PAC to out- come, why try to predict a low CI or SvO 2 from CF, urine output, and central ve- nous pressure? 3) Are there physiologic reasons why these values for CI and SvO 2 should be important? Before fully accepting these negative findings, we should recognize that they may reflect an underpowered study. Al- though the number of patients is ade- quate, there are numerous limitations in the data that potentially could cause large variability and bias. Most of these limita- tions are discussed by the authors. Par- ticularly important are the potential sources of bias in data collection: The examiners were not blinded to the CI or SvO 2 , there were no objective criteria for the CF, and interrater reliability was not determined. Imprecision in the analyses could arise from the small number of patients who had either a CI 2.5 (8%) or ScvO 2 70% (16%) so that small er- rors in measurements or CF leading to misclassifications could markedly affect the results. Further loss of power might have occurred because ARDS/ALI patients comprise a heterogeneous group. Al- though the criteria for ARDS/ALI are widely accepted, they identify a syn- drome, not a disease, with diverse etiolo- gies and phenotypes. It is well known that clinical estimates of left atrial pressure are inaccurate and that there is consider- able disagreement in interpreting chest radiographs, both leading to imprecise *See also p. 2720. Key Words: pulmonary artery catheter; ARDS/ALI; clinical examination; cardiac index; central venous oxygen saturation The author has not disclosed any potential con- flicts of interest. Copyright © 2009 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0b013e3181b3a06a 2846 Crit Care Med 2009 Vol. 37, No. 10