- Email: [email protected]
Monitor Wizards Can Be Dangerous
stain has an 85% sensitivity and enables physicians to select appropriate antimicrobial therapy in more than 90% of cases. The Infectious Diseases Society of America has recently issued a set of guidelines for management of community-acquired pneumonia. 4 Blood cultures and evaluation of carefully collected and cytologically screened sputum samples by Gram's stain and culture should be carried out on patients requiring hospitalization . For a subset of patients , testing for Legionella species, preferably byculture and urine antigen assay, should be obtained. Other studies should be done on a more selective basis. Unless new classes of antimicrobial agents are discovered, drug-resistant pathogens will become a major problem for physicians of the 21st century. It is predicted that by mid-century human life expectancy will be less than it is now, and that common bacterial infections will become increasingly important as causes of death. Precise etiologic diagnosis of pneumonia will become an increasingly important issue. Breakthroughs in diagnosis will include not only technologies that we cannot yet imagine, but also refinement and new uses of existing technologies, such as the polymerase chain reaction. Physicians will appreciate the wisdom of the old Asian blessing and curse: "May you live in interesting times. " Meanwhile, we can thank the Okayama investigators for adding another piece to the global puzzle of pneumonia, a disease that continues to hold a place among the "Captains of the Men of Death." Charles S. Bryan, MD Columbia, SC Heyward Gibbes Distinguished Professor of Internal Medicine and Chair, Department of Medicine, University of South Carolina School of Medicine. Correspondence to: Charles S. Bryan, MD, 2 Medical Park, Suite 502, Columhia, SC 29203; e-mail: [email protected]. sc.edu REFERENCES Fang CD, Fine M, Orloff J, et a!. New and emerging etiologies for community-acquired pneumonia with implications for therapy: a prospective multicenter study of 359 cases. Medici ne 1990; 69:307-316 2 Marston BJ, Lipman HB, Breiman RF. Surveillance for Legionnaires' disease; risk factors for morbidity and mortality. Arch Intern Med 1994; 154:2417-2422 3 Gleckman R, Devita J, Hibe rt D, et a!. Sputum gram stain assessment in community-acquired bacteremic pneumonia. J Clio Microbiol 1988; 26:846-849 4 Bartlett JG, Breiman RF , Mandell LA, et al. Communityacqui red pneumonia in adults: guidelines for management. Clio Infect Dis 1998; 26:811-838 1
Monitor Wizards Can Be Dangerous A
ccumulating evidence on the extensive risks and lack of benefit of the use of pulmonary artery flow catheters (Swan-Ganz catheters [SGC] ) has sparked substantial efforts to replace or, more often, supplement their use. These efforts usually center on relatively noninvasive methods for monitoring cardiac output. This issue of CHEST (see page 1643) contains a report on the use of a new bioimpedance method to measure cardiac output. Superimposed on this approach is a host of monitoring measurements. The patients in the study were subjected to the new bioimpedance method for measming cardiac output combined with routine blood pressure measurements, blood gas measurements, pulse oximetry, and transcutaneous Po 2 measurements. The workers did not avoid the use of SGC. In this article, there is a conspicuous absence of acceptable evidence that this barrage of monitoring measurements improved patient outcome. As independently stated by a s enior author of the study (in a letter to the Editor of CHEST), ".. . we have been conducting a prospective study of trauma patients entering the ED for the past 6 months and more recently have begun a similar study intraoperatively in high-risk surgical patients. Both studies are ongoing bywe are allowed to review the status at 3-month intervals, the protocol patients have about half the incidence of ARDS and other organ failures, but this is not yet statistically significant" (emphasis our own). This admission raises anumber of important issues (discussed later), but two issues immediately come to mind-why publish a methodological approach, and why use it on patients when statistical analysis fails to show benefits to patients? Extensive preoccupation with intensive, numerous, and complex monitoring approaches is not unusual among critical care physicians. The practitioners of this approach to patient care might be called monitor wizards, physicians who pursue extensive, usually unproven monitoring modalities almost as a primary goal in patient care. Monitor wizards do not always use numerous independent monitoring approaches. A single method can be expanded to provide a variety of measurements. For example, in addition to the usual data provided by the SGC, bioimpedance measures right ventricular ejection fraction , right ventricular volumes, and continuous cardiac output; it also CHEST I 114 I 6 I DECEMBER, 1998
1511
Table 1-Survival in Critical Care Units* Days
Control group, % (No SGC used)
Experim ental group, % (SGC used )
30 60 180
69.4
62.8 53.7
62.0 54.5 46.3
*Data taken from reference 2.
computes oxygen delivery and oxygen consumption. 1 These measurements are said to advance understanding of pathophysiology and the effects of interventions. But in the true spirit of monitoring, nothing is said about the impact of these measurements on patient outcome. Nor are any data provided. It should be emphasized that the term monitor wizardry is not meant to describe all sequential measurements performed on patients. There are a number of such measurements, such as ECG or blood gases, that have been established, usually pragmatically; that are not particularly hazardous; and that can lead to important beneficial changes in therapy. Rather, monitor wizardry is the extensive use of undocumented and potentially dangerous serial measurements that might lead to disastrous changes in patient management. Monitor wizards frequently overlook 1:\vo important clinical imperatives. The first is growing recognition of the importance of (acceptable) evidencebased medicine. It is beginning to be recognized that specific types of evidence must be evaluated for adequacy with scientific and epidemiologic rigor. The second imperative is older, more important, and more frequently ignored. It is the requirement for evaluating therapeutic and diagnostic measures in terms of impact on patient outcome. The second imperative can be used to generate a definition of the role of c1itical care physicians and critical care units . This role is to increase the probability of an increased quantity of life and/or quality of life, as judged by the patient. We do not know how often modern critical care units are successful in this role, but currently available data provide a sobering picture. A study2 used data derived from critical care units in five academic hospitals; the patient survival figures are shown in Table l. These data indicate that only about half of critically ill patients survive up to 3 months following admission to presumed excellent critical care units. Undoubtedly the patients were quite ill. Nevertheless, these results require us to question seriously whether extensive monitoring increases survival of patients in critical care units. 1512
Incidentally, it would be of interest to have a prospective, controlled, clinical trial comparing survival rates of patients admitted to critical care units with the rates of those treated on general medical wards. One doubts that such a study will be done. It should be emphasized that monitoring modalities can increase morbidity and mortality by two general mechanisms. Invasive monitoring may cause direct injury or death. But both invasive and noninvasive monitoring can result in death or injury as a result of incorrect diagnostic or therapeutic decisions generated by monitoring data. These in turn may be caused by physician inadequacy (lack of understanding) or may be caused by the inadequacies or limitations of the monitoring approach. Given these problems, a number of issues become relevant. Because the monitoring modalities in this multi-institutional study have not been validated by acceptable epidemiologic methods, the patients in the study served as experimental subjects. Were they so informed? Was each patient provided a process of informed choice (a better term than informed consent) before submitting to the study? Was the project cleared by appropriate institutional review boards? Were the patients provided free care while serving as experimental subjects? Have the patients who participated in the study been informed that, to date, the benefits of their management have not been demonstrable? Several general areas in medicine have been relatively exempt from the probing epidemiologic approaches that are required for patient safety. Mass screening is one such area. Although the preconditions for safe and effective mass screening were developed by the World Health Organization some 35 years ago, these are seldom, if ever, used. The use of vaccines is seldom subjected to appropriate randomized, prospective, doubleblind clinical trials. Monitoring approaches to patient care should be added to the list of medical procedures and therapies that need close epidemiologic scrutiny. The final issue is the publication of the multiinstitutional study. If this editorial is accurate, should the paper by the multi-institutional groups have been published in CHEST? We believe so. Medical journals should be more than arbiters of scientific or clinical rigor. Publishing the multi-institutional study provides an opportunity for debate and a deeper analysis of the issues involved. It is hoped that the Editorials
readership of CHEST will respond with their evaluation and analysis of the issues involved. Eugene D. Robin, MD Stanford, CA Robert F. McCauley, MD Anaheim, CA Dr. Robin is an Active Professor Emeritus at Stanford University School of Medicine. Dr. McCauley is an internist at West Anaheim Medical Center. This work was supported in part by a grant from the Sandler Family Supported Foundation. Corresponaence to: E. D. Robin, MD, PO Box 1185, Trinidad, CA 95570-1185 REFERENCES 1 Wei! HW. The assault on the Swan-Ganz catheter. Chest 1998; 113:1379-1386 2 Connors AF, SperoffT, Dawson NV, et al. The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA 1996; 276:889-897
Insulin Without Injections??? What Have you Been Puffing?
F
or patients with diabetes, 1998 should have been a diamond jubilee. It was 75 years ago, in 1923, that insulin first became available commercially, and Frederick Banting was awarded the Nobel Prize for his discovery of insulin 2 years earlier. But instead of celebrating the anniversary of this achievement, patient groups are lamenting what they perceive as so little progress in curing diabetes. Transplantation of insulin-secreting tissue is available to only a few patients and is largely reserved for those who are also receiving immunosuppression for kidney transplants. The need for multiple injections of insulin still continues to dictate the daily routine of patients with type 1 diabetes. Although recent data have shown that strict control of hyperglycemia can delay or prevent the development of microvascular complications, this goal is difficult if not impossible to achieve for most patients because of the severe hypoglycemic reactions that inescapably accompany attempts to reduce glucose to nearly normal levels. And worst of all, for many patients, is the need to inject themselves. But now, a change may be in the air. In this issue of CHEST, Laube and colleagues (see page 1734) have shown that inhalation of insulin by patients with type 2 diabetes can lead to high enough insulin blood levels to reduce endogenous insulin secretion and to bring glucose levels down to near
normal levels. Although only about one fifth of the inhaled insulin is actually absorbed, preliminary data from repeated injections suggest that this fraction is reproducible. If reproducibility can be demonstrated under conditions of ordinary use, inhaled insulin might provide an alternative to subcutaneous insulin for those patients whose erratic absorption of injected insulin leads to a pattern of sustained hyperglycemia punctuated by episodes of severe hypoglycemia. With a peak insulin concentration 30 to 60 min after treatment, inhaled insulin might provide an alternative to mealtime injections, although the need for a daily injection of neutral protamine Hagedorn or lente insulin to maintain basal blood levels would not be eliminated. Thus, even the most optimistic projection would not foresee that inhaled insulin could totally eliminate the need for injections in patients with type 1 diabetes. However, the situation for some patients with type 2 diabetes might be different. These patients maintain endogenous insulin secretion, but the amount secreted is inadequate to maintain normal glucose levels. As glucose levels rise, insulin secretion tends to fall (the so-called "glucose toxic effect"), leading to a vicious cycle of everworsening hyperglycemia and insulin deficiency. A puff of insulin before each meal might be enough to break this cycle, while allowing the patient's own beta cells to meet basal insulin requirements. This therapeutic approach may be especially appealing for women with gestational diabetes who often need insulin treatment only during the pregnancy itself. The use of inhaled insulin might save some of these women the need for insulin injections altogether. Despite these encouraging findings, it is important to remember that the results presented by Laube and colleagues are only preliminary and that it will probably take years before an inhaled insulin product is available commercially. One hopes that future clinical trials will demonstrate that inhaled insulin is reliable enough to be used outside a research setting, and that animal studies will show that the lungs are not damaged by chronic exposure to aerosolized insulin. In the meantime, the hope of insulin without injections will remain a pipe dream. Robert I. Misbin, MD Rockville, MD Endocrinology and Metabolism, Rockville, MD. These views represent those of the author and do not necessarily reflect the position of the Food and Drug Administration. Correspondence to: Robert I. Misbin, MD, Endocrinology and Metabolism, 5727 Crawford Drive, Rockville, MD 20851
CHEST I 114 I 6 I DECEMBER, 1998
1513
Monitor Wizards Can Be Dangerous A
ccumulating evidence on the extensive risks and lack of benefit of the use of pulmonary artery flow catheters (Swan-Ganz catheters [SGC] ) has sparked substantial efforts to replace or, more often, supplement their use. These efforts usually center on relatively noninvasive methods for monitoring cardiac output. This issue of CHEST (see page 1643) contains a report on the use of a new bioimpedance method to measure cardiac output. Superimposed on this approach is a host of monitoring measurements. The patients in the study were subjected to the new bioimpedance method for measming cardiac output combined with routine blood pressure measurements, blood gas measurements, pulse oximetry, and transcutaneous Po 2 measurements. The workers did not avoid the use of SGC. In this article, there is a conspicuous absence of acceptable evidence that this barrage of monitoring measurements improved patient outcome. As independently stated by a s enior author of the study (in a letter to the Editor of CHEST), ".. . we have been conducting a prospective study of trauma patients entering the ED for the past 6 months and more recently have begun a similar study intraoperatively in high-risk surgical patients. Both studies are ongoing bywe are allowed to review the status at 3-month intervals, the protocol patients have about half the incidence of ARDS and other organ failures, but this is not yet statistically significant" (emphasis our own). This admission raises anumber of important issues (discussed later), but two issues immediately come to mind-why publish a methodological approach, and why use it on patients when statistical analysis fails to show benefits to patients? Extensive preoccupation with intensive, numerous, and complex monitoring approaches is not unusual among critical care physicians. The practitioners of this approach to patient care might be called monitor wizards, physicians who pursue extensive, usually unproven monitoring modalities almost as a primary goal in patient care. Monitor wizards do not always use numerous independent monitoring approaches. A single method can be expanded to provide a variety of measurements. For example, in addition to the usual data provided by the SGC, bioimpedance measures right ventricular ejection fraction , right ventricular volumes, and continuous cardiac output; it also CHEST I 114 I 6 I DECEMBER, 1998
1511
Table 1-Survival in Critical Care Units* Days
Control group, % (No SGC used)
Experim ental group, % (SGC used )
30 60 180
69.4
62.8 53.7
62.0 54.5 46.3
*Data taken from reference 2.
computes oxygen delivery and oxygen consumption. 1 These measurements are said to advance understanding of pathophysiology and the effects of interventions. But in the true spirit of monitoring, nothing is said about the impact of these measurements on patient outcome. Nor are any data provided. It should be emphasized that the term monitor wizardry is not meant to describe all sequential measurements performed on patients. There are a number of such measurements, such as ECG or blood gases, that have been established, usually pragmatically; that are not particularly hazardous; and that can lead to important beneficial changes in therapy. Rather, monitor wizardry is the extensive use of undocumented and potentially dangerous serial measurements that might lead to disastrous changes in patient management. Monitor wizards frequently overlook 1:\vo important clinical imperatives. The first is growing recognition of the importance of (acceptable) evidencebased medicine. It is beginning to be recognized that specific types of evidence must be evaluated for adequacy with scientific and epidemiologic rigor. The second imperative is older, more important, and more frequently ignored. It is the requirement for evaluating therapeutic and diagnostic measures in terms of impact on patient outcome. The second imperative can be used to generate a definition of the role of c1itical care physicians and critical care units . This role is to increase the probability of an increased quantity of life and/or quality of life, as judged by the patient. We do not know how often modern critical care units are successful in this role, but currently available data provide a sobering picture. A study2 used data derived from critical care units in five academic hospitals; the patient survival figures are shown in Table l. These data indicate that only about half of critically ill patients survive up to 3 months following admission to presumed excellent critical care units. Undoubtedly the patients were quite ill. Nevertheless, these results require us to question seriously whether extensive monitoring increases survival of patients in critical care units. 1512
Incidentally, it would be of interest to have a prospective, controlled, clinical trial comparing survival rates of patients admitted to critical care units with the rates of those treated on general medical wards. One doubts that such a study will be done. It should be emphasized that monitoring modalities can increase morbidity and mortality by two general mechanisms. Invasive monitoring may cause direct injury or death. But both invasive and noninvasive monitoring can result in death or injury as a result of incorrect diagnostic or therapeutic decisions generated by monitoring data. These in turn may be caused by physician inadequacy (lack of understanding) or may be caused by the inadequacies or limitations of the monitoring approach. Given these problems, a number of issues become relevant. Because the monitoring modalities in this multi-institutional study have not been validated by acceptable epidemiologic methods, the patients in the study served as experimental subjects. Were they so informed? Was each patient provided a process of informed choice (a better term than informed consent) before submitting to the study? Was the project cleared by appropriate institutional review boards? Were the patients provided free care while serving as experimental subjects? Have the patients who participated in the study been informed that, to date, the benefits of their management have not been demonstrable? Several general areas in medicine have been relatively exempt from the probing epidemiologic approaches that are required for patient safety. Mass screening is one such area. Although the preconditions for safe and effective mass screening were developed by the World Health Organization some 35 years ago, these are seldom, if ever, used. The use of vaccines is seldom subjected to appropriate randomized, prospective, doubleblind clinical trials. Monitoring approaches to patient care should be added to the list of medical procedures and therapies that need close epidemiologic scrutiny. The final issue is the publication of the multiinstitutional study. If this editorial is accurate, should the paper by the multi-institutional groups have been published in CHEST? We believe so. Medical journals should be more than arbiters of scientific or clinical rigor. Publishing the multi-institutional study provides an opportunity for debate and a deeper analysis of the issues involved. It is hoped that the Editorials
readership of CHEST will respond with their evaluation and analysis of the issues involved. Eugene D. Robin, MD Stanford, CA Robert F. McCauley, MD Anaheim, CA Dr. Robin is an Active Professor Emeritus at Stanford University School of Medicine. Dr. McCauley is an internist at West Anaheim Medical Center. This work was supported in part by a grant from the Sandler Family Supported Foundation. Corresponaence to: E. D. Robin, MD, PO Box 1185, Trinidad, CA 95570-1185 REFERENCES 1 Wei! HW. The assault on the Swan-Ganz catheter. Chest 1998; 113:1379-1386 2 Connors AF, SperoffT, Dawson NV, et al. The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA 1996; 276:889-897
Insulin Without Injections??? What Have you Been Puffing?
F
or patients with diabetes, 1998 should have been a diamond jubilee. It was 75 years ago, in 1923, that insulin first became available commercially, and Frederick Banting was awarded the Nobel Prize for his discovery of insulin 2 years earlier. But instead of celebrating the anniversary of this achievement, patient groups are lamenting what they perceive as so little progress in curing diabetes. Transplantation of insulin-secreting tissue is available to only a few patients and is largely reserved for those who are also receiving immunosuppression for kidney transplants. The need for multiple injections of insulin still continues to dictate the daily routine of patients with type 1 diabetes. Although recent data have shown that strict control of hyperglycemia can delay or prevent the development of microvascular complications, this goal is difficult if not impossible to achieve for most patients because of the severe hypoglycemic reactions that inescapably accompany attempts to reduce glucose to nearly normal levels. And worst of all, for many patients, is the need to inject themselves. But now, a change may be in the air. In this issue of CHEST, Laube and colleagues (see page 1734) have shown that inhalation of insulin by patients with type 2 diabetes can lead to high enough insulin blood levels to reduce endogenous insulin secretion and to bring glucose levels down to near
normal levels. Although only about one fifth of the inhaled insulin is actually absorbed, preliminary data from repeated injections suggest that this fraction is reproducible. If reproducibility can be demonstrated under conditions of ordinary use, inhaled insulin might provide an alternative to subcutaneous insulin for those patients whose erratic absorption of injected insulin leads to a pattern of sustained hyperglycemia punctuated by episodes of severe hypoglycemia. With a peak insulin concentration 30 to 60 min after treatment, inhaled insulin might provide an alternative to mealtime injections, although the need for a daily injection of neutral protamine Hagedorn or lente insulin to maintain basal blood levels would not be eliminated. Thus, even the most optimistic projection would not foresee that inhaled insulin could totally eliminate the need for injections in patients with type 1 diabetes. However, the situation for some patients with type 2 diabetes might be different. These patients maintain endogenous insulin secretion, but the amount secreted is inadequate to maintain normal glucose levels. As glucose levels rise, insulin secretion tends to fall (the so-called "glucose toxic effect"), leading to a vicious cycle of everworsening hyperglycemia and insulin deficiency. A puff of insulin before each meal might be enough to break this cycle, while allowing the patient's own beta cells to meet basal insulin requirements. This therapeutic approach may be especially appealing for women with gestational diabetes who often need insulin treatment only during the pregnancy itself. The use of inhaled insulin might save some of these women the need for insulin injections altogether. Despite these encouraging findings, it is important to remember that the results presented by Laube and colleagues are only preliminary and that it will probably take years before an inhaled insulin product is available commercially. One hopes that future clinical trials will demonstrate that inhaled insulin is reliable enough to be used outside a research setting, and that animal studies will show that the lungs are not damaged by chronic exposure to aerosolized insulin. In the meantime, the hope of insulin without injections will remain a pipe dream. Robert I. Misbin, MD Rockville, MD Endocrinology and Metabolism, Rockville, MD. These views represent those of the author and do not necessarily reflect the position of the Food and Drug Administration. Correspondence to: Robert I. Misbin, MD, Endocrinology and Metabolism, 5727 Crawford Drive, Rockville, MD 20851
CHEST I 114 I 6 I DECEMBER, 1998
1513