Correspondence
Catheters and the Treatment of Acute Lung Injury
N Engl J Med 2006; 355:956-958August 31, 2006
- Article
To the Editor:
The National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network reports an increased incidence of arrhythmias among patients who received a pulmonary-artery catheter (PAC), as compared with a central venous catheter (CVC) (May 25 issue).1 The authors attribute this difference to the arrhythmogenic effects of the PAC and its insertion. In Table 1 of the article, they report that vasopressor use was 36 percent in the PAC group, as compared with 30 percent in the CVC group. The difference was significant. It is well known that vasopressors can cause both atrial and ventricular arrhythmias. It is therefore possible that a significant proportion of the arrhythmias encountered in the PAC group was related to the use of vasopressors and was not necessarily due to the use of the catheter itself.
1 ReferencesAndrew A. Pastewski, M.D.
Yizhak Kupfer, M.D.
Sidney Tessler, M.D.
Maimonides Medical Center, Brooklyn, NY 112191
The National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network
Full Text | Web of Science | Medline
To the Editor:
The National Heart, Lung, and Blood Institute ARDS Clinical Trials Network submitted two analyses related to the factorial Fluid and Catheter Treatment Trial (FACTT), one a comparison of the use of PACs and CVCs in patients with acute lung injury and the other a comparison of two fluid-management strategies in such patients (June 15 issue).1 However, the only information given regarding the interaction between the type of fluid management and the type of catheter is a P value of 0.26. Important data necessary for the interpretation of this factorial study include the sample sizes and mortality rates in the four factorial groups (Table 1Table 1
Factorial Table with Data Queries and Estimation of Predicted Mortality Rates from the Baseline Mortality Rate of 31 Percent in the Intensive Care Unit.).My analysis suggests that these trials were underpowered. Given the hypothesis of both reports that each intervention would decrease mortality by 33 percent (from 31 percent to 21 percent), the decrease in the mortality rate among patients receiving both interventions — the conservative fluid-management strategy and a PAC — would be from 21 percent to 14 percent (Table 1). The hypothesized mortality rate in these reports would thus be 26 percent, as compared with 17.5 percent when combined factorially. The planned enrollment of 1000 patients seems to have been based on these estimated rates of death (combined expected mortality rate, 26 percent vs. 17.5 percent; P=0.05, given a statistical power of 90 percent and a required enrollment of 986 to 1001 patients). The trial appears to have been powered for an arguably unrealistic mortality outcome of −14 percent in one group. Careful interpretation of these data is required.
1 ReferencesMark R. Daley, B.Med.
Royal Prince Alfred Hospital, Sydney 2050, Australia
mark.daley@email. cs. nsw. gov. au 1
The National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network
Full Text | Web of Science | Medline
To the Editor:
Both the FACTT, comparing the use of PACs and CVCs to guide treatment of acute lung injury, and the accompanying editorial (May 25 issue)1 cite a study by Sandham and colleagues,2 which found no benefit from the use of the PAC in high-risk surgical patients. Surprisingly, a 2002 meta-analysis by Kern and Shoemaker3 is not cited. This meta-analysis showed that only really high-risk surgical patients whose condition was optimized before organ failure occurred could benefit from PAC-guided therapy. Sandham and colleagues did not study patients at such high risk nor patients whose condition was optimized before surgery.4,5 As expected on the basis of Kern and Shoemaker's meta-analysis, acute lung injury has not been shown to improve with the use of a PAC.
A clear distinction needs to be made between critical care patients who have established organ failure, such as those with acute lung injury, and patients for whom high-risk surgery is recommended in order to decide in which patients to use PAC-guided therapy.
5 ReferencesGonzalo Tornero-Campello, M.D.
Hospital General Universitario de Elche, 03203 Elche, Spain
gtorcam@hotmail.com 1
Shure D . Pulmonary-artery catheters -- peace at last? N Engl J Med 2006;354:2273-2274
Full Text | Web of Science | Medline2
Sandham JD ,Hull RD ,Brant RF , et al. A randomized, controlled trial of the use of pulmonary-artery catheters in high-risk surgical patients. N Engl J Med 2003;348:5-14
Full Text | Web of Science | Medline3
Kern JW ,Shoemaker WC . Meta-analysis of hemodynamic optimization in high-risk patients. Crit Care Med 2002;30:1686-1692
CrossRef | Web of Science | Medline4
Cholley BP ,Payen D . Pulmonary-artery catheters in high-risk surgical patients. N Engl J Med 2003;348:2035-2036
Full Text | Web of Science | Medline5
Karkouti K ,Wijeysundera DN ,Beattie SW . Pulmonary-artery catheters in high-risk surgical patients. N Engl J Med 2003;348:2036-2036
Web of Science
Author/Editor Response
We discussed some of the factors complicating the interpretation of arrhythmia rates, including the greater number of catheter insertions in patients in the PAC group and the potential for ascertainment bias. We agree with Pastewski et al. that other factors, including vasopressor use, could cause or potentiate tachyarrhythmias. However, we doubt that the difference in the rate of vasopressor use of 6 percentage points (P=0.05) we reported accounts for the observed incidence of arrhythmias in the PAC group, which was six times that in the CVC group. Furthermore, it is unlikely that vasopressors were responsible for the higher frequency of conduction block in the PAC group.
In response to Tornero-Campello, we did not cite the meta-analysis of Kern and Shoemaker1 because our study focused on patients with established acute lung injury during the first 48 hours of illness, rather than patients in the perioperative period. Despite the differences in the types of patients, the conclusion of Kern and Shoemaker that the PAC is not of benefit in surgical patients with established organ failure is consistent with our findings. In our Discussion section, we cautioned readers regarding generalizing our findings to groups we did not study, and we reiterate this admonition.
The table provided by Daley does not correspond to the assumptions of factorial design. Conventionally in such trials, treatment combinations are not compared separately unless a significant interaction exists.2 Similar analyses occur in single-intervention studies where other factors are not controlled. For example, patients involved in a fluid-management study might have therapy directed on the basis of either a PAC or a CVC as part of usual practice. The results of such a trial would be analyzed without regard to the type of catheter unless evidence were to suggest that different types of catheters significantly altered the effectiveness of the fluid-management strategy. In our study, the catheter assignment made no difference in the aggregate, and the test for interaction was not significant (P=0.26); therefore, we conducted the pooled analysis, planned a priori, comparing both the fluid-management strategies stratified according to the type of catheter and the types of catheter stratified according to fluid-management strategy. With a factorial design, our study had 90 percent power to detect a 33 percent reduction in mortality for either intervention.
2 ReferencesArthur P. Wheeler, M.D.
Vanderbilt University, Nashville, TN 37232
art.wheeler@vanderbilt. edu Herbert P. Wiedemann, M.D.
Cleveland Clinic, Cleveland, OH 44195David A. Schoenfeld, Ph.D.
Massachusetts General Hospital, Boston, MA 021141
Kern JW ,Shoemaker WC . Meta-analysis of hemodynamic optimization in high-risk patients. Crit Care Med 2002;30:1686-1692
CrossRef | Web of Science | Medline2
Van Belle G. Statistical rules of thumb. New York: Wiley-Interscience, 2002:11.
