Utilization of pulmonary artery catheterization (PAC) remains controversial primarily because of lack of appropriately performed clinical outcome trials. Unfortunately, this controversy has been intensified by a recent call for a moratorium on PAC1 triggered by publication of an observational study plagued with limitations in design and analysis.2 In the SUPPORT investigation2, more patients managed with PAC entered the study with multiorgan system failure, acute respiratory failure, congestive heart failure, higher APACHE III scores, lower blood pressure and a lower a priori estimated survival than patients managed without PAC. Subgroup analysis indicated that the relative risk of death was 1.5 times greater in postsurgical patients with PAC than in all other groups, but this has little meaning since risk factors were not quantitated for this subpopulation. Despite attempts to equalize severity of illness using extensive statistical manipulation, only a randomized trial with a large sample size can "assure" fair distribution of unknown factors which affect outcome. Case-matching and multivariate analysis can account for only a few known measured variables.
The SUPPORT study2, like other studies reporting no change in outcome with PAC, failed to standardize treatment algorithms, prohibiting firm epidemiologic recommendations regarding PAC. Conversely, when information from PAC has been applied in a prespecified manner, improved outcome has been observed in specific patient populations.3-5 It is not PAC and measurement of physiological variables but rather the therapeutic interventions guided by PAC which affect outcome. The only definitive conclusion permitted by the SUPPORT data2 is that the treatment protocols applied to patients with PAC in 15 ICUs were not more effective than the protocols applied to patients managed without PAC. Certain patients may have received under-treatment while others received relative over-treatment with a potentially negative impact. The greater average intensity of care4-7 in the PAC group in the SUPPORT protocol2 may not have been driven by PAC, but rather, motivated by patient status and physician preference. Without tightly-controlled therapeutic protocols, debate over the correct interpretation is futile. Control of adjunctive cointerventions (e.g. antibiotics, nutrition, ventilation, transfusion, etc.) is also essential since these aspects of management can either obscure or exaggerate the impact of PAC. In addition, existing data on the effects of PAC are flawed by the presence of learning-contamination bias, since clinicians who have used PAC and have learned from it probably better manage similar patients even when not using PAC. The ethical dilemma of testing the null hypothesis by continuing patient care without PAC when hemodynamic instability occurs is a principle reason that appropriate randomized, well controlled clinical trials of PAC have not been conducted. A previous attempt to perform such a study6 was terminated prematurely with nearly half of enrolled controls crossing over to PAC because clinicians deemed it unethical to withhold PAC in the face of clinical deterioration. These difficulties should not serve as discouragement but rather, should illustrate the need for increased support of such studies from human research ethics committees as well as from funding sources.
In the absence of adequately controlled, large-scale randomized trials, there is presently no justification to call for a moratorium on PAC. PAC does carry some risks, although there is no epidemiologic evidence that PAC endangers patients or represents a safety issue.7 Based upon the large cumulative experience of clinicians caring for critically ill patients with PAC over the last 25 years, it is more likely that a moratorium on PAC would endanger the lives of large numbers of patients. A more productive and rational approach involves vigorous pursuit of the initiation and conduct of properly designed, adequately powered, randomized trials of the effectiveness of PAC, without breakdown of randomization assignment, and with incorporation of strategies for identifying optimal treatment algorithms. Most studies to date, especially the recently published SUPPORT study2 are scientifically flawed and provide no basis for any immediate change in PAC use. Rather than debate the value of PAC based upon incomplete and flawed information, we will better serve our patients by engaging the support of organizations such as APSF for the development and implementation of studies which will better define the optimal application of PAC, using preexisting expert consensus7 about the requirements for such studies.
Dr. Tuman is Professor and Vice Chair of Anesthesiology at the Rush-Presbyterian-St.LukeÕs Medical Center in Chicago.
Despite more than 30 years of clinical experience in tens of millions of patients, the pulmonary artery flow-directed catheter (PAC) has yet to save a single life. The PAC is a diagnostic tool--no more and no less. In properly trained and experienced hands, it provides information on right and left ventricular filling pressures, thermodilution cardiac output, and perhaps, right ventricular ejection fraction or mixed venous oxygen saturation. But commonly, the catheter increases hospital costs while exposing the patient to potential infectious, arrhythmogenic, mechanical, and therapeutic misadventures. Current issues involving the PAC include:
Issue 1: The PAC may cause patient injury. The literature is replete with nearly 200 published citations of serious or fatal complications arising from the insertion and use of the PAC. 1-4 Countless problems and misadventures go unreported. Recognized complications include ventricular tachycardia or fibrillation, asystole, atrial arrhythmias, right ventricular or pulmonary artery perforation, thrombosis, catheter knotting, embolism, endocarditis and other nosocomial infections,5, 6 and a host of additional problems related specifically to central venous cannulation (i.e., pneumothorax, hemothorax, chylothorax, carotid or subclavian artery puncture, and so forth).
Issue 2: The PAC may be unreliable. The accuracy of intermittent thermodilution cardiac output determinations is reported to range from +3% to +30%.7 Basic issues such as timing of the thermal injection (during the respiratory cycle), number of determinations per measurement, and injectate volume and temperature are still debated.7, 8 Vascular pressure measurements are likewise prone to over or underestimates9, being subjected to transmission of thoracic airway pressures.10 Due to variability in ventricular compliance11,Tuman has warned practitioners of the "many artifacts negating the relationship among pulmonary artery occlusion pressure (PAOP), [left ventricular end-diastolic pressure] LVEDP, and [left ventricular end-diastolic volume] LVEDV."10 Even when the PAOP can be determined accurately, the diagnostic value is limited as PAOP is an unreliable indicator of myocardial ischemia, with a positive predictive value of only 15%.12
Issue 3: The PAC requires a high degree of skill and experience to use and interpret correctly. Many physicians are inadequately skilled to properly utilize the PAC. When Iberti tested understanding and application of PAC principles and measurements among 496 US and Canadian physicians, test scores averaged <70%.13 Thus, despite 25 years of education, training, and widespread use, many physicians have demonstrable difficulty with safely inserting and troubleshooting the PAC, and may simultaneously misinterpret the hemodynamic data it generates.
Issue 4: The PAC does not improve outcome. Tuman prospectively studied over 1,000 cardiac surgery patients, and demonstrated no difference in outcome whether patients were randomized to perioperative management with a PAC or a central venous pressure (CVP) monitor.14 Furthermore, two large studies have demonstrated not only a lack of efficacy using the PAC in critically ill patients, but suggest they may do more harm than good. Gore, in a community study of over 3,000 admissions for acute myocardial infarction (MI), found patients with complicated MI (defined as congestive failure or hypotension) fared significantlv worse if a PAC was inserted.15 More recently, Connors et al. reached the same conclusion in 5,735 critically-ill patients treated with or without a PAC.16 PAC use was associated with significantly greater mortality, greater length of stay in the ICU, and of course, much greater overall cost.16 In addition, recent studies by Gattinoni17 and Hayes18 have failed to demonstrate any benefit when the PAC was utilized in a sophisticated algorithm to prospectively optimize hemodynamics and oxygen delivery in critically ill patients. Again, some patients were actually harmed.18 While the study design and statistical validity of these investigations may be questioned by some, isn't it time to acknowledge that the burden of proof is now on those who advocate routine use of this device? We believe PAC supporters must define what patient populations benefit from its use, and under what clinical circumstances (if any).
Issue 5: The PAC has high direct and indirect costs. Pearson demonstrated the PAC to be significantly more expensive than other alternatives, and no more effective.19 The PAC has many direct and indirect financial costs, as well as significant professional fees for insertion. The total exceeds $2 billion in health care expenditures in the U.S. alone.2, 16
Issue 6: Safer, less-invasive alternatives are becoming more readily available. Shoemaker reviewed recent advances in non-invasive bioimpedance technology for cardiac output determinations.20 Likewise, the use of surface, or transesophageal echocardiography (TEE), has become more common in the OR and the ICU.21 This technology provides real time, 2-D, color flow, and doppler technology to interrogate anatomic and physiologic status of the myocardium, ventricular dimensions, cardiac valves, and great vessels.
Issue 7: What, then, are indications for PAC use? Despite the current lack of a prospective, clinical trial to document improvement in outcome in critically ill patients, most practitioners (including the authors) are convinced of a substantial benefit from the PAC in patient management and therapeutic decision-making in selected cases. Indeed, many of our cherished diagnostic and monitoring devices have not held up to the scrutiny of "outcome improvement."22 Thus, while not all CABG patients need a PAC14, the redo-CABG with severely reduced left ventricular function will almost certainly require aggressive inotropic and vasopressor support with frequent hemodynamic assessments-a difficult challenge for even the most seasoned physician without diagnostic data readily available from the PAC.
In summary, while the PAC remains an instructive means of interpreting bedside physiology, no evidence substantiates improved patient outcome with its use. Indeed, the cumulative weight of evidence suggests the PAC may increase morbidity and mortality in a variety of patient populations. 15, 16 With current understanding of fluid resuscitation, inotropes, vasopressors and vasodilators, the role of the PAC should now be reevaluated. Thus, we encourage each practitioner to reevaluate their current practice and unlearn the "SWAN song."
Drs. Prielipp and Morell are from the Department of Anesthesia, Bowman Gray School of Medicine, Winston-Salem, NC.