Free
Correspondence  |   February 2009
Residual Confounding in Observational Studies
Author Notes
  • Royal Perth Hospital, Perth, Australia.
Article Information
Correspondence
Correspondence   |   February 2009
Residual Confounding in Observational Studies
Anesthesiology 2 2009, Vol.110, 430. doi:10.1097/ALN.0b013e3181943242
Anesthesiology 2 2009, Vol.110, 430. doi:10.1097/ALN.0b013e3181943242
To the Editor:—
I read the study by Fellahi JL et al.  titled “Perioperative use of dobutamine in cardiac surgery and adverse cardiac outcome” with great interest.1 This study showed that the use of dobutamine was associated with adverse outcomes in adult cardiac surgery, and the authors concluded that these results suggest that dobutamine should only be administered when the benefit is judged to outweigh the risks. Although the study is interesting and the results are provocative, we need to be careful how to examine these results before we conclude that these results suggest that dobutamine may be harmful. First, we can be sure from the data of this observational study that the group of patients who received dobutamine was sicker in both left ventricular function and EURO score. Therefore, it is vital to know whether the propensity score (probability of using dobutamine) was a significant variable in the second method of logistic regression analysis when the propensity score was used as a separate covariate. If propensity score was a significant variable in determining adverse outcome after cardiac surgery in the logistic regression analysis, it would suggest that residual confounding due to “selection bias” caused by using dobutamine in a sicker group of patients was still present, despite the propensity score was used in the multivariate analysis.2 We definitely need this data to be sure whether any residual confounding was present, and if residual confounding is present, it may at least, in part, explain the results why the use of dobutamine was associated with a poorer outcome. Furthermore, the authors should also discuss the possibility of residual confounding from omitting confounders as a limitation of an observational study.3 Second, the conclusion of only using dobutamine when benefit is judged to outweigh the risks is not helpful to both practicing anesthesiologists and researchers. The data presented in this study showed that dobutamine was used predominantly in patients with a lower ejection fraction or those with left ventricular dysfunction in their cohort. This practice will remain unchanged in many cardiothoracic centers after knowing the results of this study because this is the group of cardiac patients most likely to have benefits from the use of dobutamine. I believe a more appropriate, balanced, and constructive conclusion from the data of this study is that an adequately powered randomized controlled trial is needed to clarify the risks and benefits of dobutamine in patients with different degrees of left ventricular function in adult cardiac surgery.4 
Royal Perth Hospital, Perth, Australia.
References
Fellahi JL, Parienti JJ, Hanouz JL, Plaud B, Riou B, Ouattara A: Perioperative use of dobutamine in cardiac surgery and adverse cardiac outcome: propensity-adjusted analyses. Anesthesiology 2008; 108:979–87Fellahi, JL Parienti, JJ Hanouz, JL Plaud, B Riou, B Ouattara, A
Joffe MM, Rosenbaum PR: Invited commentary: propensity scores. Am J Epidemiol 1999; 150:327–33Joffe, MM Rosenbaum, PR
Weitzen S, Lapane KL, Toledano AY, Hume AL, Mor V: Weaknesses of goodness-of-fit tests for evaluating propensity score models: the case of the omitted confounder. Pharmacoepidemiol Drug Saf 2005; 14:227–38Weitzen, S Lapane, KL Toledano, AY Hume, AL Mor, V
Deeks JJ, Dinnes J, D’Amico R, Sowden AJ, Sakarovitch C, Song F, Petticrew M, Altman DG: International Stroke Trial Collaborative Group; European Carotid Surgery Trial Collaborative Group. Evaluating non-randomised intervention studies. Health Technol Assess 2003; 7:iii–x, 1–173