Brian Harte, MD

In the not-too-distant future, a multiphysician hospitalist group is a participant in a pay-for-performance (P4P) program. Dr. Buchmann, the group’s lead hospitalist, is confronted by his hospital’s administration and informed that his doctors are performing below regional benchmarks for standards of care for community-acquired pneumonia, and, in fact, the hospital as a whole is below the mean performance levels.

The hospital threatens to break its contract with Dr. Buchmann’s group despite his response that his group sees a far more complex population than these standards can account for—and besides, his group has implemented a number of important quality initiatives in other diseases that are not part of the P4P program.

Several of the group’s hospitalists state that they will stop seeing indigent patients and will no longer take referrals for high-risk patients. Another partner feels it is unethical to continue treating pneumonia patients at the hospital without informing them of these quality findings and at least offering the option of transfer to a facility with better scores. Dr. Buchmann finds all these propositions unsettling.

While these physicians’ responses may sound extreme, the behavior of physicians caught between the hammer of financial survival and the anvil of professional ethics is unpredictable. Medicare and other payers have been implementing P4P plans as the latest attempt to stimulate quality reform. There are dozens of P4P-based programs operating in the United States, and the financial implications are daunting. Further, P4P is taking hold despite a relative paucity of research regarding its effectiveness in improving outcomes.

The underlying rationale of P4P is the use of economic incentives to stimulate changes in provider behavior. Recent work from the RAND Corporation suggests that as much as one-half of healthcare is not based on “accepted” best practices.1 And with increasing attention on the role of errors in medical practice, any effort to improve care seems, on its surface, laudable.

In general, key elements of P4P programs include a set of performance measures, the collection of data, comparison of provider data with benchmarks, and rewards for physicians who meet or exceed those targets. The interface between economic and financial incentives requires physicians to ensure that their behavior is in line with ethical and professional standards. While journals of medicine, law, and business contain many articles devoted to the policy and market implications of P4P, there is surprisingly little discussion in the literature regarding the potential ethical challenges that physicians may face in these programs.

For hospitalists (and other physicians), P4P may present several troubling ethical issues. Because the current scope of P4P is limited to a few diseases, widespread implementation might lead to relative neglect of patients with other illnesses. Higher-risk patients might be avoided, and individual patient concerns might become subjugated to population performance measures. Hospitalists could face the additional conflict of being accountable to (and/or dependent upon) hospitals, which feel P4P pressures of their own. A final issue is the question of whether shared decision-making and patient-centered care mandate disclosure of non-public quality data to patients.

The American Medical Association (AMA) has a policy that specifically addresses P4P.2 Its “Pay-for-Performance Principles and Guidelines” call for physician participation in P4P to be voluntary and to allow physicians to access their ratings for potential appeal prior to wider release. The policy insists that quality of care be paramount over cost savings and that the physician-patient relationship be preserved. Of course, P4P programs may not share the AMA’s ethical concerns and are not bound to consider them.

Of particular concern—for both inpatient and outpatient physicians—is the fate of high-risk and unassigned patients. According to SHM’s 2005-2006 “Bi-Annual Survey on the State of the Hospital Medicine Movement” (www.hospital medicine.org), 96% of hospitalists are involved in the care of unassigned patients, and, in general, one of the strengths of hospital-based physicians should be their relative familiarity with the acute problems of patients who are older and of those with concomitant morbidities. Yet these are precisely the patient groups that are not well served by typical P4P measurements.

The potential for P4P incentives to create disparities in patient care among different patient groups and diseases is one of the prime concerns in the Council on Ethical and Judicial Affairs’ recent opinion for the AMA on P4P programs.3 The care of older patients, for instance, because of their own choices and due to frequency of comorbidities, may well come up short in performance measures designed for individuals who have a single disease.

This is not just a policy problem because P4P is not only unlikely to adequately address the ethical concerns of equitable care to these groups, it could exacerbate the vulnerability of these populations by creating a disincentive to provide care.4 A recent publication describing reports of cardiac surgeons turning away high-risk patients after “CABG report cards” became publicly available suggests that when given the option at least some physicians may indeed change their behavior when quality information is being collected and reported.5 Ironically, a system that incentivizes doctors to avoid the highest-risk patients could worsen—rather than improve—the overall quality of care.

Hospitalists may not be as sensitive to these pressures as surgeons or outpatient physicians, especially given the hospitalist’s limited flexibility in “choosing” patients. Care of unassigned patients may be a contractual obligation for which a hospitalist is paid by the hospital (which may face its own pressures in this area). And lower-risk referrals from outpatient physicians may “compensate” for the occasional complex patient.

Hospitalists are generally “need-based” practitioners who legally and ethically may not have the option to refuse care without risking patient abandonment. Yet the fact that hospitalists take on such patients may make their performance scores inferior to even non-hospital-based doctors—a difficult position to be in if one’s group receives payments from the hospital with an expectation of superior performance. Hospitalists in particular must consider whether or not insurance companies and the Centers for Medicare and Medicaid Services (CMS) could really accommodate all possible confounders in a risk-adjustment model to offset the nature of their patients. While the ethical choice might be for hospitalists simply to refuse to participate in P4P, citing multiple conflicts of interest, there is no clear indication regarding how “optional” these programs will be as they become increasingly prevalent, presenting yet another ethical issue.

Further, Medicare’s current P4P system for hospitals is directed at just five conditions, only two of which (congestive heart failure and pneumonia) are likely to fall within a hospitalist’s realm. But the list of common diagnoses under the hospitalist’s umbrella is, of course, much larger, including thromboembolism, pyelonephritis, COPD, cirrhosis, and sepsis. The data that exists for compliance with recommended care for some of these conditions (e.g., COPD) suggests that there may be substantial variability.6

But if hospitals base their support for hospitalist programs on their performance within a few CMS diagnoses, the effect on care for and development of appropriate guidelines and resources toward many other conditions may suffer. Already, hospital discharge forms are pre-printed with checkboxes for an angiotensin-converting enzyme (ACE) inhibitor prescription for congestive heart failure and counseling for smoking cessation. The (unethical) implication is that some diagnoses are more valuable than others, and that physician energies may be inequitably distributed—whether consciously or not. It is difficult to see how P4P could encompass standards for every patient condition, or how hospitals and providers could avoid focusing resources on those conditions that are more closely scrutinized by their payers.

Another issue arises if patient autonomy dictates that a treatment plan has to deviate from established guidelines; in such a case, hospitalists and other physicians may be forced to provide a care plan that is entirely reasonable from a medical standpoint but counts against them when compared with a benchmark. Ethical principles dictate that patient care be given priority, but unless consideration is made within the scoring system, performance measures that do not accommodate the ethical mandates to respect patient wishes or physician judgment are substantial pitfalls in the pursuit of better quality.7,8

One last issue concerns the question of whether or not providers have an obligation to disclose quality data to patients in the context of shared decision-making. This is a murky subject that involves determining the boundaries between the best means of pursuing quality improvement and the ethics of patient advocacy. The AMA’s Code of Medical Ethics states, “Patients should receive guidance from their physicians as to the optimal course of action,” and the issue of competence and responsibility to the care of the individual patient is the focus of several of the Principles of Medical Ethics. However, there is practically nothing published regarding the ethics, implications, or results of such disclosure, presumably because the availability of large amounts of quality-based data is such a new phenomenon and the considerations of such disclosure are so uncomfortable for many physicians.

Of course, some information—“CABG report cards,” for instance—is publicly available, but the evidence that patients actually utilize this information to a significant extent or that quality has improved due to its use is mixed.5 The question of whether an obligation exists to disclose non-public information when a provider knows that there is a question about performance relative to a benchmark or comparative peer group is uncharted water, ethically speaking; the issue is further complicated by the fact that appraisal of quality is far from a perfect science. It may be that the benefits of P4P result primarily from transparency, rather than from financial incentives. If so, disclosure may be the major component of quality reform, giving further weight to this question.

The ethical problems raised by P4P are underappreciated and inadequately discussed in the literature, particularly for how rapidly and rampantly these programs are being piloted and implemented. Although the AMA has taken a fairly clear and reasonable stance on the appropriate considerations for P4P programs, it is not clear that payers are incorporating all these concerns. A substantial number of hospitalist groups receive payments from medical groups or hospitals, which in turn are already involved in P4P for some diagnoses.

All hospitalists should read and familiarize themselves with these guidelines and carefully assess the implications of forthcoming P4P proposals for their own practices and patients. On a larger scale, SHM and its membership should strongly consider taking the lead in defining appropriate processes and outcomes for hospital care that incorporate these ethical concerns and allow for meaningful conclusions regarding both quality of care and opportunities for improvement.

Dr. Harte works at the Cleveland Clinic, and Dr. Rajput works at the Robert Wood Johnson Medical School, Camden, N.J. The authors wish to thank Tom Baudendistel and Donald Krause for their review and suggestions.

References

1. Asch SM, Kerr EA, Keesey J, et al. Who is at greatest risk for receiving poor-quality health care? N Engl J Med. 2006 Mar 16;354:1147-1156.
2. American Medical Association. Pay-for-performance principles and guidelines. Accessible at: www.ama-assn.org/meetings/public/annual05/bot5a05.doc. Last accessed September 13, 2006.
3. American Medical Association. CEJA 3-1-05 Report, July 2006. Available at: www.ama-assn.org/ama/pub/category/4325.html. Last accessed September 13, 2006.
4. Morreim EH. Result-based compensation in health care: a good, but limited, idea. J Law Med Ethics. 2001 Summer;29(2):174-181.
5. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA. 2005 Mar 9;293:1239-1244.
6. Lindenauer PK, Pekow P, Gao S, et al. Quality of care for patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 2006;144(12):894-903.
7. Walter LC, Davidowitz NP, Heineken PA, et al. Pitfalls of converting practice guidelines into quality measures: lessons learned from a VA performance measure. JAMA. 2004 May 26;291(20):2466-2470.
8. Boyd CM, Darer J, Boult C, et al. Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance. JAMA. 2005 Aug 10;294(6):716-724.

In the not-too-distant future, a multiphysician hospitalist group is a participant in a pay-for-performance (P4P) program. Dr. Buchmann, the group’s lead hospitalist, is confronted by his hospital’s administration and informed that his doctors are performing below regional benchmarks for standards of care for community-acquired pneumonia, and, in fact, the hospital as a whole is below the mean performance levels.

The hospital threatens to break its contract with Dr. Buchmann’s group despite his response that his group sees a far more complex population than these standards can account for—and besides, his group has implemented a number of important quality initiatives in other diseases that are not part of the P4P program.

Several of the group’s hospitalists state that they will stop seeing indigent patients and will no longer take referrals for high-risk patients. Another partner feels it is unethical to continue treating pneumonia patients at the hospital without informing them of these quality findings and at least offering the option of transfer to a facility with better scores. Dr. Buchmann finds all these propositions unsettling.

While these physicians’ responses may sound extreme, the behavior of physicians caught between the hammer of financial survival and the anvil of professional ethics is unpredictable. Medicare and other payers have been implementing P4P plans as the latest attempt to stimulate quality reform. There are dozens of P4P-based programs operating in the United States, and the financial implications are daunting. Further, P4P is taking hold despite a relative paucity of research regarding its effectiveness in improving outcomes.

The underlying rationale of P4P is the use of economic incentives to stimulate changes in provider behavior. Recent work from the RAND Corporation suggests that as much as one-half of healthcare is not based on “accepted” best practices.1 And with increasing attention on the role of errors in medical practice, any effort to improve care seems, on its surface, laudable.

In general, key elements of P4P programs include a set of performance measures, the collection of data, comparison of provider data with benchmarks, and rewards for physicians who meet or exceed those targets. The interface between economic and financial incentives requires physicians to ensure that their behavior is in line with ethical and professional standards. While journals of medicine, law, and business contain many articles devoted to the policy and market implications of P4P, there is surprisingly little discussion in the literature regarding the potential ethical challenges that physicians may face in these programs.

For hospitalists (and other physicians), P4P may present several troubling ethical issues. Because the current scope of P4P is limited to a few diseases, widespread implementation might lead to relative neglect of patients with other illnesses. Higher-risk patients might be avoided, and individual patient concerns might become subjugated to population performance measures. Hospitalists could face the additional conflict of being accountable to (and/or dependent upon) hospitals, which feel P4P pressures of their own. A final issue is the question of whether shared decision-making and patient-centered care mandate disclosure of non-public quality data to patients.

The American Medical Association (AMA) has a policy that specifically addresses P4P.2 Its “Pay-for-Performance Principles and Guidelines” call for physician participation in P4P to be voluntary and to allow physicians to access their ratings for potential appeal prior to wider release. The policy insists that quality of care be paramount over cost savings and that the physician-patient relationship be preserved. Of course, P4P programs may not share the AMA’s ethical concerns and are not bound to consider them.

Of particular concern—for both inpatient and outpatient physicians—is the fate of high-risk and unassigned patients. According to SHM’s 2005-2006 “Bi-Annual Survey on the State of the Hospital Medicine Movement” (www.hospital medicine.org), 96% of hospitalists are involved in the care of unassigned patients, and, in general, one of the strengths of hospital-based physicians should be their relative familiarity with the acute problems of patients who are older and of those with concomitant morbidities. Yet these are precisely the patient groups that are not well served by typical P4P measurements.

The potential for P4P incentives to create disparities in patient care among different patient groups and diseases is one of the prime concerns in the Council on Ethical and Judicial Affairs’ recent opinion for the AMA on P4P programs.3 The care of older patients, for instance, because of their own choices and due to frequency of comorbidities, may well come up short in performance measures designed for individuals who have a single disease.

This is not just a policy problem because P4P is not only unlikely to adequately address the ethical concerns of equitable care to these groups, it could exacerbate the vulnerability of these populations by creating a disincentive to provide care.4 A recent publication describing reports of cardiac surgeons turning away high-risk patients after “CABG report cards” became publicly available suggests that when given the option at least some physicians may indeed change their behavior when quality information is being collected and reported.5 Ironically, a system that incentivizes doctors to avoid the highest-risk patients could worsen—rather than improve—the overall quality of care.

Hospitalists may not be as sensitive to these pressures as surgeons or outpatient physicians, especially given the hospitalist’s limited flexibility in “choosing” patients. Care of unassigned patients may be a contractual obligation for which a hospitalist is paid by the hospital (which may face its own pressures in this area). And lower-risk referrals from outpatient physicians may “compensate” for the occasional complex patient.

Hospitalists are generally “need-based” practitioners who legally and ethically may not have the option to refuse care without risking patient abandonment. Yet the fact that hospitalists take on such patients may make their performance scores inferior to even non-hospital-based doctors—a difficult position to be in if one’s group receives payments from the hospital with an expectation of superior performance. Hospitalists in particular must consider whether or not insurance companies and the Centers for Medicare and Medicaid Services (CMS) could really accommodate all possible confounders in a risk-adjustment model to offset the nature of their patients. While the ethical choice might be for hospitalists simply to refuse to participate in P4P, citing multiple conflicts of interest, there is no clear indication regarding how “optional” these programs will be as they become increasingly prevalent, presenting yet another ethical issue.

Further, Medicare’s current P4P system for hospitals is directed at just five conditions, only two of which (congestive heart failure and pneumonia) are likely to fall within a hospitalist’s realm. But the list of common diagnoses under the hospitalist’s umbrella is, of course, much larger, including thromboembolism, pyelonephritis, COPD, cirrhosis, and sepsis. The data that exists for compliance with recommended care for some of these conditions (e.g., COPD) suggests that there may be substantial variability.6

But if hospitals base their support for hospitalist programs on their performance within a few CMS diagnoses, the effect on care for and development of appropriate guidelines and resources toward many other conditions may suffer. Already, hospital discharge forms are pre-printed with checkboxes for an angiotensin-converting enzyme (ACE) inhibitor prescription for congestive heart failure and counseling for smoking cessation. The (unethical) implication is that some diagnoses are more valuable than others, and that physician energies may be inequitably distributed—whether consciously or not. It is difficult to see how P4P could encompass standards for every patient condition, or how hospitals and providers could avoid focusing resources on those conditions that are more closely scrutinized by their payers.

Another issue arises if patient autonomy dictates that a treatment plan has to deviate from established guidelines; in such a case, hospitalists and other physicians may be forced to provide a care plan that is entirely reasonable from a medical standpoint but counts against them when compared with a benchmark. Ethical principles dictate that patient care be given priority, but unless consideration is made within the scoring system, performance measures that do not accommodate the ethical mandates to respect patient wishes or physician judgment are substantial pitfalls in the pursuit of better quality.7,8

One last issue concerns the question of whether or not providers have an obligation to disclose quality data to patients in the context of shared decision-making. This is a murky subject that involves determining the boundaries between the best means of pursuing quality improvement and the ethics of patient advocacy. The AMA’s Code of Medical Ethics states, “Patients should receive guidance from their physicians as to the optimal course of action,” and the issue of competence and responsibility to the care of the individual patient is the focus of several of the Principles of Medical Ethics. However, there is practically nothing published regarding the ethics, implications, or results of such disclosure, presumably because the availability of large amounts of quality-based data is such a new phenomenon and the considerations of such disclosure are so uncomfortable for many physicians.

Of course, some information—“CABG report cards,” for instance—is publicly available, but the evidence that patients actually utilize this information to a significant extent or that quality has improved due to its use is mixed.5 The question of whether an obligation exists to disclose non-public information when a provider knows that there is a question about performance relative to a benchmark or comparative peer group is uncharted water, ethically speaking; the issue is further complicated by the fact that appraisal of quality is far from a perfect science. It may be that the benefits of P4P result primarily from transparency, rather than from financial incentives. If so, disclosure may be the major component of quality reform, giving further weight to this question.

The ethical problems raised by P4P are underappreciated and inadequately discussed in the literature, particularly for how rapidly and rampantly these programs are being piloted and implemented. Although the AMA has taken a fairly clear and reasonable stance on the appropriate considerations for P4P programs, it is not clear that payers are incorporating all these concerns. A substantial number of hospitalist groups receive payments from medical groups or hospitals, which in turn are already involved in P4P for some diagnoses.

All hospitalists should read and familiarize themselves with these guidelines and carefully assess the implications of forthcoming P4P proposals for their own practices and patients. On a larger scale, SHM and its membership should strongly consider taking the lead in defining appropriate processes and outcomes for hospital care that incorporate these ethical concerns and allow for meaningful conclusions regarding both quality of care and opportunities for improvement.

Dr. Harte works at the Cleveland Clinic, and Dr. Rajput works at the Robert Wood Johnson Medical School, Camden, N.J. The authors wish to thank Tom Baudendistel and Donald Krause for their review and suggestions.

References

1. Asch SM, Kerr EA, Keesey J, et al. Who is at greatest risk for receiving poor-quality health care? N Engl J Med. 2006 Mar 16;354:1147-1156.
2. American Medical Association. Pay-for-performance principles and guidelines. Accessible at: www.ama-assn.org/meetings/public/annual05/bot5a05.doc. Last accessed September 13, 2006.
3. American Medical Association. CEJA 3-1-05 Report, July 2006. Available at: www.ama-assn.org/ama/pub/category/4325.html. Last accessed September 13, 2006.
4. Morreim EH. Result-based compensation in health care: a good, but limited, idea. J Law Med Ethics. 2001 Summer;29(2):174-181.
5. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA. 2005 Mar 9;293:1239-1244.
6. Lindenauer PK, Pekow P, Gao S, et al. Quality of care for patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 2006;144(12):894-903.
7. Walter LC, Davidowitz NP, Heineken PA, et al. Pitfalls of converting practice guidelines into quality measures: lessons learned from a VA performance measure. JAMA. 2004 May 26;291(20):2466-2470.
8. Boyd CM, Darer J, Boult C, et al. Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance. JAMA. 2005 Aug 10;294(6):716-724.

In the not-too-distant future, a multiphysician hospitalist group is a participant in a pay-for-performance (P4P) program. Dr. Buchmann, the group’s lead hospitalist, is confronted by his hospital’s administration and informed that his doctors are performing below regional benchmarks for standards of care for community-acquired pneumonia, and, in fact, the hospital as a whole is below the mean performance levels.

The hospital threatens to break its contract with Dr. Buchmann’s group despite his response that his group sees a far more complex population than these standards can account for—and besides, his group has implemented a number of important quality initiatives in other diseases that are not part of the P4P program.

Several of the group’s hospitalists state that they will stop seeing indigent patients and will no longer take referrals for high-risk patients. Another partner feels it is unethical to continue treating pneumonia patients at the hospital without informing them of these quality findings and at least offering the option of transfer to a facility with better scores. Dr. Buchmann finds all these propositions unsettling.

While these physicians’ responses may sound extreme, the behavior of physicians caught between the hammer of financial survival and the anvil of professional ethics is unpredictable. Medicare and other payers have been implementing P4P plans as the latest attempt to stimulate quality reform. There are dozens of P4P-based programs operating in the United States, and the financial implications are daunting. Further, P4P is taking hold despite a relative paucity of research regarding its effectiveness in improving outcomes.

The underlying rationale of P4P is the use of economic incentives to stimulate changes in provider behavior. Recent work from the RAND Corporation suggests that as much as one-half of healthcare is not based on “accepted” best practices.1 And with increasing attention on the role of errors in medical practice, any effort to improve care seems, on its surface, laudable.

In general, key elements of P4P programs include a set of performance measures, the collection of data, comparison of provider data with benchmarks, and rewards for physicians who meet or exceed those targets. The interface between economic and financial incentives requires physicians to ensure that their behavior is in line with ethical and professional standards. While journals of medicine, law, and business contain many articles devoted to the policy and market implications of P4P, there is surprisingly little discussion in the literature regarding the potential ethical challenges that physicians may face in these programs.

For hospitalists (and other physicians), P4P may present several troubling ethical issues. Because the current scope of P4P is limited to a few diseases, widespread implementation might lead to relative neglect of patients with other illnesses. Higher-risk patients might be avoided, and individual patient concerns might become subjugated to population performance measures. Hospitalists could face the additional conflict of being accountable to (and/or dependent upon) hospitals, which feel P4P pressures of their own. A final issue is the question of whether shared decision-making and patient-centered care mandate disclosure of non-public quality data to patients.

The American Medical Association (AMA) has a policy that specifically addresses P4P.2 Its “Pay-for-Performance Principles and Guidelines” call for physician participation in P4P to be voluntary and to allow physicians to access their ratings for potential appeal prior to wider release. The policy insists that quality of care be paramount over cost savings and that the physician-patient relationship be preserved. Of course, P4P programs may not share the AMA’s ethical concerns and are not bound to consider them.

Of particular concern—for both inpatient and outpatient physicians—is the fate of high-risk and unassigned patients. According to SHM’s 2005-2006 “Bi-Annual Survey on the State of the Hospital Medicine Movement” (www.hospital medicine.org), 96% of hospitalists are involved in the care of unassigned patients, and, in general, one of the strengths of hospital-based physicians should be their relative familiarity with the acute problems of patients who are older and of those with concomitant morbidities. Yet these are precisely the patient groups that are not well served by typical P4P measurements.

The potential for P4P incentives to create disparities in patient care among different patient groups and diseases is one of the prime concerns in the Council on Ethical and Judicial Affairs’ recent opinion for the AMA on P4P programs.3 The care of older patients, for instance, because of their own choices and due to frequency of comorbidities, may well come up short in performance measures designed for individuals who have a single disease.

This is not just a policy problem because P4P is not only unlikely to adequately address the ethical concerns of equitable care to these groups, it could exacerbate the vulnerability of these populations by creating a disincentive to provide care.4 A recent publication describing reports of cardiac surgeons turning away high-risk patients after “CABG report cards” became publicly available suggests that when given the option at least some physicians may indeed change their behavior when quality information is being collected and reported.5 Ironically, a system that incentivizes doctors to avoid the highest-risk patients could worsen—rather than improve—the overall quality of care.

Hospitalists may not be as sensitive to these pressures as surgeons or outpatient physicians, especially given the hospitalist’s limited flexibility in “choosing” patients. Care of unassigned patients may be a contractual obligation for which a hospitalist is paid by the hospital (which may face its own pressures in this area). And lower-risk referrals from outpatient physicians may “compensate” for the occasional complex patient.

Hospitalists are generally “need-based” practitioners who legally and ethically may not have the option to refuse care without risking patient abandonment. Yet the fact that hospitalists take on such patients may make their performance scores inferior to even non-hospital-based doctors—a difficult position to be in if one’s group receives payments from the hospital with an expectation of superior performance. Hospitalists in particular must consider whether or not insurance companies and the Centers for Medicare and Medicaid Services (CMS) could really accommodate all possible confounders in a risk-adjustment model to offset the nature of their patients. While the ethical choice might be for hospitalists simply to refuse to participate in P4P, citing multiple conflicts of interest, there is no clear indication regarding how “optional” these programs will be as they become increasingly prevalent, presenting yet another ethical issue.

Further, Medicare’s current P4P system for hospitals is directed at just five conditions, only two of which (congestive heart failure and pneumonia) are likely to fall within a hospitalist’s realm. But the list of common diagnoses under the hospitalist’s umbrella is, of course, much larger, including thromboembolism, pyelonephritis, COPD, cirrhosis, and sepsis. The data that exists for compliance with recommended care for some of these conditions (e.g., COPD) suggests that there may be substantial variability.6

But if hospitals base their support for hospitalist programs on their performance within a few CMS diagnoses, the effect on care for and development of appropriate guidelines and resources toward many other conditions may suffer. Already, hospital discharge forms are pre-printed with checkboxes for an angiotensin-converting enzyme (ACE) inhibitor prescription for congestive heart failure and counseling for smoking cessation. The (unethical) implication is that some diagnoses are more valuable than others, and that physician energies may be inequitably distributed—whether consciously or not. It is difficult to see how P4P could encompass standards for every patient condition, or how hospitals and providers could avoid focusing resources on those conditions that are more closely scrutinized by their payers.

Another issue arises if patient autonomy dictates that a treatment plan has to deviate from established guidelines; in such a case, hospitalists and other physicians may be forced to provide a care plan that is entirely reasonable from a medical standpoint but counts against them when compared with a benchmark. Ethical principles dictate that patient care be given priority, but unless consideration is made within the scoring system, performance measures that do not accommodate the ethical mandates to respect patient wishes or physician judgment are substantial pitfalls in the pursuit of better quality.7,8

One last issue concerns the question of whether or not providers have an obligation to disclose quality data to patients in the context of shared decision-making. This is a murky subject that involves determining the boundaries between the best means of pursuing quality improvement and the ethics of patient advocacy. The AMA’s Code of Medical Ethics states, “Patients should receive guidance from their physicians as to the optimal course of action,” and the issue of competence and responsibility to the care of the individual patient is the focus of several of the Principles of Medical Ethics. However, there is practically nothing published regarding the ethics, implications, or results of such disclosure, presumably because the availability of large amounts of quality-based data is such a new phenomenon and the considerations of such disclosure are so uncomfortable for many physicians.

Of course, some information—“CABG report cards,” for instance—is publicly available, but the evidence that patients actually utilize this information to a significant extent or that quality has improved due to its use is mixed.5 The question of whether an obligation exists to disclose non-public information when a provider knows that there is a question about performance relative to a benchmark or comparative peer group is uncharted water, ethically speaking; the issue is further complicated by the fact that appraisal of quality is far from a perfect science. It may be that the benefits of P4P result primarily from transparency, rather than from financial incentives. If so, disclosure may be the major component of quality reform, giving further weight to this question.

The ethical problems raised by P4P are underappreciated and inadequately discussed in the literature, particularly for how rapidly and rampantly these programs are being piloted and implemented. Although the AMA has taken a fairly clear and reasonable stance on the appropriate considerations for P4P programs, it is not clear that payers are incorporating all these concerns. A substantial number of hospitalist groups receive payments from medical groups or hospitals, which in turn are already involved in P4P for some diagnoses.

All hospitalists should read and familiarize themselves with these guidelines and carefully assess the implications of forthcoming P4P proposals for their own practices and patients. On a larger scale, SHM and its membership should strongly consider taking the lead in defining appropriate processes and outcomes for hospital care that incorporate these ethical concerns and allow for meaningful conclusions regarding both quality of care and opportunities for improvement.

Dr. Harte works at the Cleveland Clinic, and Dr. Rajput works at the Robert Wood Johnson Medical School, Camden, N.J. The authors wish to thank Tom Baudendistel and Donald Krause for their review and suggestions.

References

1. Asch SM, Kerr EA, Keesey J, et al. Who is at greatest risk for receiving poor-quality health care? N Engl J Med. 2006 Mar 16;354:1147-1156.
2. American Medical Association. Pay-for-performance principles and guidelines. Accessible at: www.ama-assn.org/meetings/public/annual05/bot5a05.doc. Last accessed September 13, 2006.
3. American Medical Association. CEJA 3-1-05 Report, July 2006. Available at: www.ama-assn.org/ama/pub/category/4325.html. Last accessed September 13, 2006.
4. Morreim EH. Result-based compensation in health care: a good, but limited, idea. J Law Med Ethics. 2001 Summer;29(2):174-181.
5. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA. 2005 Mar 9;293:1239-1244.
6. Lindenauer PK, Pekow P, Gao S, et al. Quality of care for patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 2006;144(12):894-903.
7. Walter LC, Davidowitz NP, Heineken PA, et al. Pitfalls of converting practice guidelines into quality measures: lessons learned from a VA performance measure. JAMA. 2004 May 26;291(20):2466-2470.
8. Boyd CM, Darer J, Boult C, et al. Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance. JAMA. 2005 Aug 10;294(6):716-724.

1. Dexter PR, Perkins SM, Mahany KS, Jones K, McDonald CJ. Inpatient computer-based standing orders vs. physician reminders to increase influenza and pneumococcal vaccination rates: a randomized trial. JAMA. 2004; 292: 2366-71.

Past studies have suggested that most patients admitted with severe pneumococcal infections have been hospitalized in the preceding 5 years, and simply being hospitalized is a potential risk factor for later pneumococcal infection. Likewise, hospitalization provides an opportunity to vaccinate high-risk patients against influenza, and raising pneumococcal and influenza immunization rates is a CMS quality improvement priority. Prior investigations have supported the use of labor-intensive manual standing orders as well as computerized reminders, but this prospective trial was conducted in 1998 and 1999 to assess the effectiveness of a computer-based system to screen for eligible patients and then generate orders to perform pneumonia and influenza vaccinations on inpatients at the time of discharge.

Over 13 months, a total of 3777 inpatients were entered into the study. The hospital computer identified patients eligible for vaccination based on common criteria and randomized them to one of two groups of physician teams. For one group of teams, the computer order-entry system would automatically generate vaccination orders at the time of discharge for vaccine-eligible patients; for the other group of teams, only computer reminders were provided to physicians. The outcome measure was administration of vaccine; long-term outcomes such as incidence of subsequent disease or mortality were not measured.

During the study period, 50% of all hospitalized patients were identified as eligible for influenza vaccination; 22% were eligible for pneumococcal vaccination. In each case, the “standing order” group received vaccine more often (influenza: 42% vs. 30%, p<.001; pneumococcal vaccine: 51% vs. 31%). The numbers were subsequently adjusted to allow for patients who had previously received vaccine, but the impressive differences persisted. Nurses reported reasons for non-administration in 98% of the eligible patients who were not vaccinated; the most common reason was patient refusal. It is not clear if the physicians knew that a study was being conducted. No adverse reactions were reported.

CMS has pushed for the development of institutional standing order sets as a tool to improve compliance with vaccination rate targets. Where the technology is available, computer systems that can screen eligible patients and generate automatic orders are an effective tool in implementing many quality-improvement initiatives, and hospitalists are in a crucial position to take an active role in their development and implementation.

2. Fang MC, Chang Y, Hylek EM, et al. Advanced age, anticoagulation intensity, and risk for intracranial hemorrhage among patients taking warfarin for a trial fibrillation. Ann Intern Med. 2004; 141: 745-52.

Warfarin has been shown to reduce risk of stroke in patients with chronic and paroxysmal atrial fibrillation. Intracranial hemorrhage remains one of the most feared complications of warfarin, especially among older patients, prompting suggestions to consider lower intensity anticoagulation among patients older than 75 years who have atrial fibrillation.

This study evaluated the relationship between the intensity of anticoagulation, risk of intracranial hemorrhage, and age of patients with atrial fibrillation.

This was a retrospective case control study conducted at a tertiary care medical center. One-hundred and seventy patients on warfarin and admitted with intracranial hemorrhage from 1993 to 2002 were matched with 1020 patients who were on warfarin but without intracranial bleed. After controlling for comorbid conditions and aspirin use, authors conducted multivariable logistic regression analysis to determine the odds of intracranial hemorrhage with regard to age and INR. The risk of intracranial hemorrhage increased at 85 years of age and at INR values of 3.5 or greater. The risk of intracranial hemorrhage at INR less than 2.0 did not differ statistically from the risk at INR of 2.0–3.0.

This study shows the risk of intracranial hemorrhage is not decreased by choosing lower intensity anticoagulation, and target INR should still be kept at 2.5 among elderly patients. However, patients older than 85 years should be counseled about their higher risk of intracranial hemorrhage.

3. Heeschen C, Hamm CW, Mitrovic V, et al. N-terminal pro-B-type natriuretic peptide levels for dynamic risk stratification of patients with acute coronary syndromes. Circulation. 2004;110: 3206-12.

Recent data demonstrate the prognostic value of assessment of neurohormonal activation in patients with acute coronary syndromes (ACS). B-type natriuretic peptide levels (BNP) and levels of the N-terminal fragment of the BNP prohormone (NT-proBNP) predict adverse long-term outcomes in patients with ACS. Investigators reviewed plasma samples of Troponin T (TnT) and NT-proBNP obtained from patients with ACS enrolled in the Platelet Receptor Inhibition in Ischemic Syndrome Management (PRISM) trial, which randomized patients to tirofiban or heparin for 48 hours and assessed mortality and myocardial infarction at 30 day follow-up. TnT and NT-proBNP determinations were available at baseline for 1791 patients, and at 48 and 72 hours from 1401 patients. Baseline NT-proBNP levels >250 ng/L were associated with significantly higher rates of death and myocardial infarction at 7 and 30 day follow-up. After adjustment for TnT and C-reactive protein levels, elevated NTproBNP levels maintained its predictive value (OR 2.7; p<.001). In patients with normal TnT levels, NT-proBNP levels identified a subgroup of patients at increased risk (OR 3.0; p=.004). However, in patients with high TnT levels (>0.1 mcg/L), NT-proBNP lost its predictive value (p=.58). More importantly, patients with normal levels of both TnT and NT-proBNP were at very low risk (0.6% event rate at 30 day follow-up).

Serial determinations of NT-proBNP levels at 48 and 72 hours were reviewed in patients without major adverse cardiac events (death or myocardial infarction); these patients were subdivided into groups with and without refractory ischemia. Patients without refractory ischemia showed a significant decline in NT-proBNP levels, whereas patients with refractory ischemia had no significant change. Persistently elevated NT-proBNP levels at 72 hours were associated with a 17.2% risk of death or MI at 30 days, compared with 0.6% risk if NT-proBNP returned to normal at 72 hours (p<.001). Neither TnT nor C-reactive protein demonstrated similar predictive value.

The study is limited by its retrospective nature, by potential selection bias by including only patients with direct evidence of coronary artery disease, and by limitations of the generalizability of its findings (e.g., to emergency department patients with chest pain).

As BNP and NT-proBNP are counter-regulatory hormones that play an active role in the response to ischemic injury, the authors suggest that NT-proBNP is a promising tool for dynamic risk assessment in patients with ACS. The authors also do not differentiate between BNP and NT-proBNP with regard to use in risk stratification, which might lead one to believe that these tests share similar predictive value. (Of note, the study was entirely funded by a company that produces an assay for NT-proBNP). Prospective trials to validate this tool are warranted

4. Heuschmann PU, Kolominsky-Rabas PL, Roether J, et al. Predictors of in-hospital mortality in patients with acute ischemic stroke treated with thrombolytic therapy. JAMA. 2004;292:1831-38.

The objective of this study was to identify factors associated with in-hospital mortality in ischemic stroke patients treated with recombinant tissue plasminogen activator (tPA). It was a prospective observational cohort study of 1658 patients conducted at 225 community and academic hospitals throughout Germany with main outcome of in-hospital mortality.

In this study 10% of patients who were treated with tPA died during their hospital stay, with 2/3 of deaths occurring in the first 7 days. Relative probability of in-hospital mortality increased with increasing patient age, with an odds ratio (OR) of 1.6 for each 10-year increment in age. Age was an independent predictor of in hospital mortality irrespective of tPA administration, with patients older than 75 years age having 4 fold higher mortality than the youngest cohort of less than 55 years age.

Other factors predicting in hospital mortality were altered level of consciousness and relative lack of experience with tPA treatment in the center. Altered level of consciousness was a predictor of stroke severity and an independent predictor of in-hospital mortality (OR 3.4). The increase in mortality in centers with limited experience with tPA administration (OR 0.97) reflected learning curve issues with these patients. The study was not designed to separate out the confounders of operator experience curve from institutional experience curves, or to derive the exact relationship between experience and outcomes.

5. McAlister FA, Bertsch K, Man J, et al. Incidence of and risk factors for pulmonary complications after non-thoracic surgery. Am J Respir Crit Care Med. 2004; published ahead of print on November 24, 2004 as doi:10.1164/rccm.200408-1069OC. Accessed January 27, 2005.

Postoperative pulmonary complications after nonthoracic surgery are a cause of significant morbidity and increased length of hospital stay. Previous studies of preoperative pulmonary assessment were limited by non-representative patient samples, conflicting results, and lack of explicit definitions of these complications. The authors conducted a prospective cohort study of 1055 patients seen in a Pre-Admission Clinic of a tertiary care university hospital. Mean age was 55 years, 50% male, and the cohort consisted of patients scheduled for intermediate risk elective surgery (upper abdominal, lower abdominal, orthopedic). They evaluated physical exam maneuvers (cough test, wheeze test, maximal laryngeal height, and forced expiratory time, all of which are described in an online data supplement) and preoperative spirometry values, and collected information on clinically significant postoperative pulmonary complications, including pneumonia, respiratory failure requiring mechanical ventilation, atelectasis requiring bronchoscopy, or pneumothorax or pleural effusion requiring percutaneous intervention. Twenty-eight patients (2.7%) suffered a pulmonary complication within 7 days of surgery, one of whom died. Length of stay was significantly prolonged in this group (mean 27.9 days vs 4.5 days, p=.006). Multivariate regression analysis revealed four variables that were independently associated with increased risk for postoperative pulmonary complications: age > 65 years, positive cough test (repeated coughing after asking the patient to inhale deeply and cough once), perioperative nasogastric tube, and anesthesia duration 2.5 hours or greater. Number of pack years smoked, FEV1, FEV1/FVC ratio, and upper abdominal surgery were associated with postoperative pulmonary complications but were not found to be independently associated by multivariate analysis.

While it is not surprising that the above risk factors are predictive of postoperative complications, this is the first study to incorporate specific exam maneuvers and spirometry into a risk prediction analysis. Limitations of this model are lack of independent validation and lack of generalizability to other populations, e.g., inpatients awaiting urgent surgery. Of note, the study further provides further support for not routinely obtaining pulmonary function testing for risk stratification prior to noncardiac surgery.

6. Mortenson, EM, Restrepo M, Anzeuto A, Pugh J. Effects of guideline-concordant antimicrobial therapy on mortality among patients with community-acquired pneumonia. Am J Med. 2004;117:726-31.

The American Thoracic Society (ATS) and the Infectious Diseases Society of America (IDSA) have published guidelines for the management of community acquired pneumonia that include recommendations for antibiotic selection. This retrospective cohort study attempted to measure the association between adherence to such guidelines and 30-day mortality in patients admitted with pneumonia to two Texas hospitals.

The characteristics of the patients studied reflect a reasonable cross-section of typical pneumonia patients, with the exception that the patients were mostly (85%) men. Of the study population, 78% were admitted through the ER, 20% admitted to the ICU, and 9% were nursing home residents. Antibiotics were considered “concordant” if they were consistent with either the most recent IDSA or ATS recommendations. The “nonconcordant” group was slightly older (66 vs. 61), generally sicker (higher rate of comorbid COPD and CVD), had more cerebrovascular disease, was more likely to present with altered mental status, and less frequently received antibiotics within 8 hours of presentation. The study did not comment on patients’ vaccination status. Thirty-day mortality was 6.2% in the guideline-concordant group, versus 21.7% in the other group (p < .001). The most common “non-concordant” regimen described was use of a beta-lactam alone, although specific antibiotic regimens were not evaluated.

While the results of this study are not surprising, they provide us with both the validation to continue practice according to existing recommendations, particularly the avoidance of monotherapy with beta-lactam antibiotics. The study also provides us with the imperative to take the lead in developing evidence-based pneumonia pathways at our own hospitals.

7. Pham MX, Whooley MA, Evans GT Jr, et al. Prognostic value of low-level cardiac troponin-I elevations in patients without definite acute coronary syndromes. Am Heart J. 2004;148: 776-82.

With the availability of rapid and highly sensitive and specific troponin testing, many patients admitted to the hospital with non-cardiac diagnoses have been recognized as having abnormal serum troponin-I or troponin-T levels, often just slightly above the reference cutoff for “normal.” While the clinical assumption is often that the elevated enzyme level does not reflect an acute coronary syndrome per se, its significance regarding the patient’s underlying cardiac health is often unclear.

Pham et al. retrospectively reviewed the 1-year mortality of 366 patients who were admitted to the San Francisco VA without evidence of acute MI or ACS either clinically or by EKG, but who had low-level troponin-I elevations (up to 3.0 ng/mL—a level that the authors state was reached by institutional consensus, and which was measured by a “first-generation” assay). These patients were admitted for a broad spectrum of diagnoses ranging from CHF to COPD to sepsis. Ninety-six percent of the patients were men; their average age was 69.

Follow-up was accomplished after a mean of 288 days and included 97% of patients. The primary endpoint was MI or death due to cardiac disease at one year; secondary endpoints were revascularization or admission for unstable angina. The primary endpoint was reached by 11% of patients with cTn-I between 1.0 and 3.0 ng/mL, and 4% of the patients with cTn-I up to 1.0 ng/mL (adjusted HR 3.4, 95% CI, 1.3 to 9.4), and the higher the cTn-I, the higher the risk. However, the authors did not test for the level of risk by specific diagnosis, so they caution against overgeneralizing their findings.

The findings of this study add to the evidence that any evidence of myocardial injury implies an increased risk of underlying heart disease and accompanying long—term cardiac complications– even if such injury occurs in the absence of ACS or known heart disease. Hospitalists often see such injury in the setting of acute infection and pulmonary disease and may be the first to recognize the possibility of CHD in a given patient. To date, guidelines addressing optimal prospective risk stratification have not been developed. Until they are, hospitalists should be aware of the ramifications of “troponin leak” and be prepared to initiate necessary inpatient monitoring and treatments, and to coordinate appropriate follow-up for these patients.

8. Saposnik G, Young B, Silver B, et al. Lack of improvement in patients with acute stroke after treatment with thrombolytic therapy: predictors and association with outcome. JAMA. 2004; 292: 1839-44.

Recombinant tissue plasminogen activator (tPA) has been shown to be one of the most efficacious therapies for acute stroke treatment. This was a systematic evaluation of predictors for outcomes at 24 hours after tPA therapy and of the prognostic significance of lack of improvement at 24 hours for long-term outcomes at 3 months.

The trial was a prospective cohort study of 216 consecutive patients admitted with acute ischemic stroke to a university hospital. The decision to treat with tPA was based on the NINDS protocol with one difference: patients with involvement of more than one third of the middle cerebral artery on the baseline CT scan were excluded. A control CT scan was performed at 24 hours to determine the presence of new infarction, cortical involvement, and extension of the ischemic lesion.

Lack of improvement was defined as a difference between the NIHSS score at baseline and at 24 hours of 3 points or less. Poor outcome at 3 months was defined by a modified Rankin Scale score of 3 to 5 or death.

After adjusting for age, gender, and stroke severity, hyperglycemia at admission (glucose > 144 mg/dL), cortical involvement, and time to treatment were independent predictors of lack of improvement at 24 hours. After adjusting for age, gender, and stroke severity, lack of improvement at 24 hours was an independent predictor of poor outcome and death at 3 months. Patients with lack of improvement at 24 hours also had longer lengths of hospitalization.

9. Taylor AL, Ziesche S, Yancy C, et al. Combination of isosorbide dinitrate and hydralazine in blacks with heart failure. N Engl J Med. 2004;351:2049-57.

Neurohormonal changes, endothelial dysfunction, impaired nitric oxide availability, and oxidant stress all contribute to the structural remodeling of the left ventricle in congestive heart failure. The combination of isosorbide dinitrate, an organic nitrate that stimulates nitric oxide signaling, and the antioxidant and vasodilator hydralazine improves survival in heart failure. Based on more recent data that black patients have a clinically significant response to this combination therapy, the authors of the African-American Heart Failure Trial (A-HeFT) evaluated 1050 black patients with congestive heart failure in a randomized, double-blind, placebo controlled trial. Patients were randomized to fixed doses of isosorbide dinitrate and hydralazine plus background therapy (i.e., digoxin, ACE inhibitors, beta-blockers, diuretics, angiotensin receptor blockers) or to placebo plus background therapy. After 18 months, the trial was stopped due to a significantly higher mortality rate in the placebo group (10.2% in the placebo group vs. 6.2% with combination

therapy, p=.02); survival differences emerged at 180 days and increased progressively thereafter. The combination therapy group reported more headache and dizziness but suffered fewer exacerbations of congestive heart failure and reported improvement in subjective assessments of quality of life as measured by questionnaires. Accompanying editorials discuss the role of nitric oxide and prevention of oxidative stress in the treatment of heart failure, as well as the controversial issues surrounding race-based therapeutics.

1. Dexter PR, Perkins SM, Mahany KS, Jones K, McDonald CJ. Inpatient computer-based standing orders vs. physician reminders to increase influenza and pneumococcal vaccination rates: a randomized trial. JAMA. 2004; 292: 2366-71.

Past studies have suggested that most patients admitted with severe pneumococcal infections have been hospitalized in the preceding 5 years, and simply being hospitalized is a potential risk factor for later pneumococcal infection. Likewise, hospitalization provides an opportunity to vaccinate high-risk patients against influenza, and raising pneumococcal and influenza immunization rates is a CMS quality improvement priority. Prior investigations have supported the use of labor-intensive manual standing orders as well as computerized reminders, but this prospective trial was conducted in 1998 and 1999 to assess the effectiveness of a computer-based system to screen for eligible patients and then generate orders to perform pneumonia and influenza vaccinations on inpatients at the time of discharge.

Over 13 months, a total of 3777 inpatients were entered into the study. The hospital computer identified patients eligible for vaccination based on common criteria and randomized them to one of two groups of physician teams. For one group of teams, the computer order-entry system would automatically generate vaccination orders at the time of discharge for vaccine-eligible patients; for the other group of teams, only computer reminders were provided to physicians. The outcome measure was administration of vaccine; long-term outcomes such as incidence of subsequent disease or mortality were not measured.

During the study period, 50% of all hospitalized patients were identified as eligible for influenza vaccination; 22% were eligible for pneumococcal vaccination. In each case, the “standing order” group received vaccine more often (influenza: 42% vs. 30%, p<.001; pneumococcal vaccine: 51% vs. 31%). The numbers were subsequently adjusted to allow for patients who had previously received vaccine, but the impressive differences persisted. Nurses reported reasons for non-administration in 98% of the eligible patients who were not vaccinated; the most common reason was patient refusal. It is not clear if the physicians knew that a study was being conducted. No adverse reactions were reported.

CMS has pushed for the development of institutional standing order sets as a tool to improve compliance with vaccination rate targets. Where the technology is available, computer systems that can screen eligible patients and generate automatic orders are an effective tool in implementing many quality-improvement initiatives, and hospitalists are in a crucial position to take an active role in their development and implementation.

2. Fang MC, Chang Y, Hylek EM, et al. Advanced age, anticoagulation intensity, and risk for intracranial hemorrhage among patients taking warfarin for a trial fibrillation. Ann Intern Med. 2004; 141: 745-52.

Warfarin has been shown to reduce risk of stroke in patients with chronic and paroxysmal atrial fibrillation. Intracranial hemorrhage remains one of the most feared complications of warfarin, especially among older patients, prompting suggestions to consider lower intensity anticoagulation among patients older than 75 years who have atrial fibrillation.

This study evaluated the relationship between the intensity of anticoagulation, risk of intracranial hemorrhage, and age of patients with atrial fibrillation.

This was a retrospective case control study conducted at a tertiary care medical center. One-hundred and seventy patients on warfarin and admitted with intracranial hemorrhage from 1993 to 2002 were matched with 1020 patients who were on warfarin but without intracranial bleed. After controlling for comorbid conditions and aspirin use, authors conducted multivariable logistic regression analysis to determine the odds of intracranial hemorrhage with regard to age and INR. The risk of intracranial hemorrhage increased at 85 years of age and at INR values of 3.5 or greater. The risk of intracranial hemorrhage at INR less than 2.0 did not differ statistically from the risk at INR of 2.0–3.0.

This study shows the risk of intracranial hemorrhage is not decreased by choosing lower intensity anticoagulation, and target INR should still be kept at 2.5 among elderly patients. However, patients older than 85 years should be counseled about their higher risk of intracranial hemorrhage.

3. Heeschen C, Hamm CW, Mitrovic V, et al. N-terminal pro-B-type natriuretic peptide levels for dynamic risk stratification of patients with acute coronary syndromes. Circulation. 2004;110: 3206-12.

Recent data demonstrate the prognostic value of assessment of neurohormonal activation in patients with acute coronary syndromes (ACS). B-type natriuretic peptide levels (BNP) and levels of the N-terminal fragment of the BNP prohormone (NT-proBNP) predict adverse long-term outcomes in patients with ACS. Investigators reviewed plasma samples of Troponin T (TnT) and NT-proBNP obtained from patients with ACS enrolled in the Platelet Receptor Inhibition in Ischemic Syndrome Management (PRISM) trial, which randomized patients to tirofiban or heparin for 48 hours and assessed mortality and myocardial infarction at 30 day follow-up. TnT and NT-proBNP determinations were available at baseline for 1791 patients, and at 48 and 72 hours from 1401 patients. Baseline NT-proBNP levels >250 ng/L were associated with significantly higher rates of death and myocardial infarction at 7 and 30 day follow-up. After adjustment for TnT and C-reactive protein levels, elevated NTproBNP levels maintained its predictive value (OR 2.7; p<.001). In patients with normal TnT levels, NT-proBNP levels identified a subgroup of patients at increased risk (OR 3.0; p=.004). However, in patients with high TnT levels (>0.1 mcg/L), NT-proBNP lost its predictive value (p=.58). More importantly, patients with normal levels of both TnT and NT-proBNP were at very low risk (0.6% event rate at 30 day follow-up).

Serial determinations of NT-proBNP levels at 48 and 72 hours were reviewed in patients without major adverse cardiac events (death or myocardial infarction); these patients were subdivided into groups with and without refractory ischemia. Patients without refractory ischemia showed a significant decline in NT-proBNP levels, whereas patients with refractory ischemia had no significant change. Persistently elevated NT-proBNP levels at 72 hours were associated with a 17.2% risk of death or MI at 30 days, compared with 0.6% risk if NT-proBNP returned to normal at 72 hours (p<.001). Neither TnT nor C-reactive protein demonstrated similar predictive value.

The study is limited by its retrospective nature, by potential selection bias by including only patients with direct evidence of coronary artery disease, and by limitations of the generalizability of its findings (e.g., to emergency department patients with chest pain).

As BNP and NT-proBNP are counter-regulatory hormones that play an active role in the response to ischemic injury, the authors suggest that NT-proBNP is a promising tool for dynamic risk assessment in patients with ACS. The authors also do not differentiate between BNP and NT-proBNP with regard to use in risk stratification, which might lead one to believe that these tests share similar predictive value. (Of note, the study was entirely funded by a company that produces an assay for NT-proBNP). Prospective trials to validate this tool are warranted

4. Heuschmann PU, Kolominsky-Rabas PL, Roether J, et al. Predictors of in-hospital mortality in patients with acute ischemic stroke treated with thrombolytic therapy. JAMA. 2004;292:1831-38.

The objective of this study was to identify factors associated with in-hospital mortality in ischemic stroke patients treated with recombinant tissue plasminogen activator (tPA). It was a prospective observational cohort study of 1658 patients conducted at 225 community and academic hospitals throughout Germany with main outcome of in-hospital mortality.

In this study 10% of patients who were treated with tPA died during their hospital stay, with 2/3 of deaths occurring in the first 7 days. Relative probability of in-hospital mortality increased with increasing patient age, with an odds ratio (OR) of 1.6 for each 10-year increment in age. Age was an independent predictor of in hospital mortality irrespective of tPA administration, with patients older than 75 years age having 4 fold higher mortality than the youngest cohort of less than 55 years age.

Other factors predicting in hospital mortality were altered level of consciousness and relative lack of experience with tPA treatment in the center. Altered level of consciousness was a predictor of stroke severity and an independent predictor of in-hospital mortality (OR 3.4). The increase in mortality in centers with limited experience with tPA administration (OR 0.97) reflected learning curve issues with these patients. The study was not designed to separate out the confounders of operator experience curve from institutional experience curves, or to derive the exact relationship between experience and outcomes.

5. McAlister FA, Bertsch K, Man J, et al. Incidence of and risk factors for pulmonary complications after non-thoracic surgery. Am J Respir Crit Care Med. 2004; published ahead of print on November 24, 2004 as doi:10.1164/rccm.200408-1069OC. Accessed January 27, 2005.

Postoperative pulmonary complications after nonthoracic surgery are a cause of significant morbidity and increased length of hospital stay. Previous studies of preoperative pulmonary assessment were limited by non-representative patient samples, conflicting results, and lack of explicit definitions of these complications. The authors conducted a prospective cohort study of 1055 patients seen in a Pre-Admission Clinic of a tertiary care university hospital. Mean age was 55 years, 50% male, and the cohort consisted of patients scheduled for intermediate risk elective surgery (upper abdominal, lower abdominal, orthopedic). They evaluated physical exam maneuvers (cough test, wheeze test, maximal laryngeal height, and forced expiratory time, all of which are described in an online data supplement) and preoperative spirometry values, and collected information on clinically significant postoperative pulmonary complications, including pneumonia, respiratory failure requiring mechanical ventilation, atelectasis requiring bronchoscopy, or pneumothorax or pleural effusion requiring percutaneous intervention. Twenty-eight patients (2.7%) suffered a pulmonary complication within 7 days of surgery, one of whom died. Length of stay was significantly prolonged in this group (mean 27.9 days vs 4.5 days, p=.006). Multivariate regression analysis revealed four variables that were independently associated with increased risk for postoperative pulmonary complications: age > 65 years, positive cough test (repeated coughing after asking the patient to inhale deeply and cough once), perioperative nasogastric tube, and anesthesia duration 2.5 hours or greater. Number of pack years smoked, FEV1, FEV1/FVC ratio, and upper abdominal surgery were associated with postoperative pulmonary complications but were not found to be independently associated by multivariate analysis.

While it is not surprising that the above risk factors are predictive of postoperative complications, this is the first study to incorporate specific exam maneuvers and spirometry into a risk prediction analysis. Limitations of this model are lack of independent validation and lack of generalizability to other populations, e.g., inpatients awaiting urgent surgery. Of note, the study further provides further support for not routinely obtaining pulmonary function testing for risk stratification prior to noncardiac surgery.

6. Mortenson, EM, Restrepo M, Anzeuto A, Pugh J. Effects of guideline-concordant antimicrobial therapy on mortality among patients with community-acquired pneumonia. Am J Med. 2004;117:726-31.

The American Thoracic Society (ATS) and the Infectious Diseases Society of America (IDSA) have published guidelines for the management of community acquired pneumonia that include recommendations for antibiotic selection. This retrospective cohort study attempted to measure the association between adherence to such guidelines and 30-day mortality in patients admitted with pneumonia to two Texas hospitals.

The characteristics of the patients studied reflect a reasonable cross-section of typical pneumonia patients, with the exception that the patients were mostly (85%) men. Of the study population, 78% were admitted through the ER, 20% admitted to the ICU, and 9% were nursing home residents. Antibiotics were considered “concordant” if they were consistent with either the most recent IDSA or ATS recommendations. The “nonconcordant” group was slightly older (66 vs. 61), generally sicker (higher rate of comorbid COPD and CVD), had more cerebrovascular disease, was more likely to present with altered mental status, and less frequently received antibiotics within 8 hours of presentation. The study did not comment on patients’ vaccination status. Thirty-day mortality was 6.2% in the guideline-concordant group, versus 21.7% in the other group (p < .001). The most common “non-concordant” regimen described was use of a beta-lactam alone, although specific antibiotic regimens were not evaluated.

While the results of this study are not surprising, they provide us with both the validation to continue practice according to existing recommendations, particularly the avoidance of monotherapy with beta-lactam antibiotics. The study also provides us with the imperative to take the lead in developing evidence-based pneumonia pathways at our own hospitals.

7. Pham MX, Whooley MA, Evans GT Jr, et al. Prognostic value of low-level cardiac troponin-I elevations in patients without definite acute coronary syndromes. Am Heart J. 2004;148: 776-82.

With the availability of rapid and highly sensitive and specific troponin testing, many patients admitted to the hospital with non-cardiac diagnoses have been recognized as having abnormal serum troponin-I or troponin-T levels, often just slightly above the reference cutoff for “normal.” While the clinical assumption is often that the elevated enzyme level does not reflect an acute coronary syndrome per se, its significance regarding the patient’s underlying cardiac health is often unclear.

Pham et al. retrospectively reviewed the 1-year mortality of 366 patients who were admitted to the San Francisco VA without evidence of acute MI or ACS either clinically or by EKG, but who had low-level troponin-I elevations (up to 3.0 ng/mL—a level that the authors state was reached by institutional consensus, and which was measured by a “first-generation” assay). These patients were admitted for a broad spectrum of diagnoses ranging from CHF to COPD to sepsis. Ninety-six percent of the patients were men; their average age was 69.

Follow-up was accomplished after a mean of 288 days and included 97% of patients. The primary endpoint was MI or death due to cardiac disease at one year; secondary endpoints were revascularization or admission for unstable angina. The primary endpoint was reached by 11% of patients with cTn-I between 1.0 and 3.0 ng/mL, and 4% of the patients with cTn-I up to 1.0 ng/mL (adjusted HR 3.4, 95% CI, 1.3 to 9.4), and the higher the cTn-I, the higher the risk. However, the authors did not test for the level of risk by specific diagnosis, so they caution against overgeneralizing their findings.

The findings of this study add to the evidence that any evidence of myocardial injury implies an increased risk of underlying heart disease and accompanying long—term cardiac complications– even if such injury occurs in the absence of ACS or known heart disease. Hospitalists often see such injury in the setting of acute infection and pulmonary disease and may be the first to recognize the possibility of CHD in a given patient. To date, guidelines addressing optimal prospective risk stratification have not been developed. Until they are, hospitalists should be aware of the ramifications of “troponin leak” and be prepared to initiate necessary inpatient monitoring and treatments, and to coordinate appropriate follow-up for these patients.

8. Saposnik G, Young B, Silver B, et al. Lack of improvement in patients with acute stroke after treatment with thrombolytic therapy: predictors and association with outcome. JAMA. 2004; 292: 1839-44.

Recombinant tissue plasminogen activator (tPA) has been shown to be one of the most efficacious therapies for acute stroke treatment. This was a systematic evaluation of predictors for outcomes at 24 hours after tPA therapy and of the prognostic significance of lack of improvement at 24 hours for long-term outcomes at 3 months.

The trial was a prospective cohort study of 216 consecutive patients admitted with acute ischemic stroke to a university hospital. The decision to treat with tPA was based on the NINDS protocol with one difference: patients with involvement of more than one third of the middle cerebral artery on the baseline CT scan were excluded. A control CT scan was performed at 24 hours to determine the presence of new infarction, cortical involvement, and extension of the ischemic lesion.

Lack of improvement was defined as a difference between the NIHSS score at baseline and at 24 hours of 3 points or less. Poor outcome at 3 months was defined by a modified Rankin Scale score of 3 to 5 or death.

After adjusting for age, gender, and stroke severity, hyperglycemia at admission (glucose > 144 mg/dL), cortical involvement, and time to treatment were independent predictors of lack of improvement at 24 hours. After adjusting for age, gender, and stroke severity, lack of improvement at 24 hours was an independent predictor of poor outcome and death at 3 months. Patients with lack of improvement at 24 hours also had longer lengths of hospitalization.

9. Taylor AL, Ziesche S, Yancy C, et al. Combination of isosorbide dinitrate and hydralazine in blacks with heart failure. N Engl J Med. 2004;351:2049-57.

Neurohormonal changes, endothelial dysfunction, impaired nitric oxide availability, and oxidant stress all contribute to the structural remodeling of the left ventricle in congestive heart failure. The combination of isosorbide dinitrate, an organic nitrate that stimulates nitric oxide signaling, and the antioxidant and vasodilator hydralazine improves survival in heart failure. Based on more recent data that black patients have a clinically significant response to this combination therapy, the authors of the African-American Heart Failure Trial (A-HeFT) evaluated 1050 black patients with congestive heart failure in a randomized, double-blind, placebo controlled trial. Patients were randomized to fixed doses of isosorbide dinitrate and hydralazine plus background therapy (i.e., digoxin, ACE inhibitors, beta-blockers, diuretics, angiotensin receptor blockers) or to placebo plus background therapy. After 18 months, the trial was stopped due to a significantly higher mortality rate in the placebo group (10.2% in the placebo group vs. 6.2% with combination

therapy, p=.02); survival differences emerged at 180 days and increased progressively thereafter. The combination therapy group reported more headache and dizziness but suffered fewer exacerbations of congestive heart failure and reported improvement in subjective assessments of quality of life as measured by questionnaires. Accompanying editorials discuss the role of nitric oxide and prevention of oxidative stress in the treatment of heart failure, as well as the controversial issues surrounding race-based therapeutics.

1. Dexter PR, Perkins SM, Mahany KS, Jones K, McDonald CJ. Inpatient computer-based standing orders vs. physician reminders to increase influenza and pneumococcal vaccination rates: a randomized trial. JAMA. 2004; 292: 2366-71.

Past studies have suggested that most patients admitted with severe pneumococcal infections have been hospitalized in the preceding 5 years, and simply being hospitalized is a potential risk factor for later pneumococcal infection. Likewise, hospitalization provides an opportunity to vaccinate high-risk patients against influenza, and raising pneumococcal and influenza immunization rates is a CMS quality improvement priority. Prior investigations have supported the use of labor-intensive manual standing orders as well as computerized reminders, but this prospective trial was conducted in 1998 and 1999 to assess the effectiveness of a computer-based system to screen for eligible patients and then generate orders to perform pneumonia and influenza vaccinations on inpatients at the time of discharge.

Over 13 months, a total of 3777 inpatients were entered into the study. The hospital computer identified patients eligible for vaccination based on common criteria and randomized them to one of two groups of physician teams. For one group of teams, the computer order-entry system would automatically generate vaccination orders at the time of discharge for vaccine-eligible patients; for the other group of teams, only computer reminders were provided to physicians. The outcome measure was administration of vaccine; long-term outcomes such as incidence of subsequent disease or mortality were not measured.

During the study period, 50% of all hospitalized patients were identified as eligible for influenza vaccination; 22% were eligible for pneumococcal vaccination. In each case, the “standing order” group received vaccine more often (influenza: 42% vs. 30%, p<.001; pneumococcal vaccine: 51% vs. 31%). The numbers were subsequently adjusted to allow for patients who had previously received vaccine, but the impressive differences persisted. Nurses reported reasons for non-administration in 98% of the eligible patients who were not vaccinated; the most common reason was patient refusal. It is not clear if the physicians knew that a study was being conducted. No adverse reactions were reported.

CMS has pushed for the development of institutional standing order sets as a tool to improve compliance with vaccination rate targets. Where the technology is available, computer systems that can screen eligible patients and generate automatic orders are an effective tool in implementing many quality-improvement initiatives, and hospitalists are in a crucial position to take an active role in their development and implementation.

2. Fang MC, Chang Y, Hylek EM, et al. Advanced age, anticoagulation intensity, and risk for intracranial hemorrhage among patients taking warfarin for a trial fibrillation. Ann Intern Med. 2004; 141: 745-52.

Warfarin has been shown to reduce risk of stroke in patients with chronic and paroxysmal atrial fibrillation. Intracranial hemorrhage remains one of the most feared complications of warfarin, especially among older patients, prompting suggestions to consider lower intensity anticoagulation among patients older than 75 years who have atrial fibrillation.

This study evaluated the relationship between the intensity of anticoagulation, risk of intracranial hemorrhage, and age of patients with atrial fibrillation.

This was a retrospective case control study conducted at a tertiary care medical center. One-hundred and seventy patients on warfarin and admitted with intracranial hemorrhage from 1993 to 2002 were matched with 1020 patients who were on warfarin but without intracranial bleed. After controlling for comorbid conditions and aspirin use, authors conducted multivariable logistic regression analysis to determine the odds of intracranial hemorrhage with regard to age and INR. The risk of intracranial hemorrhage increased at 85 years of age and at INR values of 3.5 or greater. The risk of intracranial hemorrhage at INR less than 2.0 did not differ statistically from the risk at INR of 2.0–3.0.

This study shows the risk of intracranial hemorrhage is not decreased by choosing lower intensity anticoagulation, and target INR should still be kept at 2.5 among elderly patients. However, patients older than 85 years should be counseled about their higher risk of intracranial hemorrhage.

3. Heeschen C, Hamm CW, Mitrovic V, et al. N-terminal pro-B-type natriuretic peptide levels for dynamic risk stratification of patients with acute coronary syndromes. Circulation. 2004;110: 3206-12.

Recent data demonstrate the prognostic value of assessment of neurohormonal activation in patients with acute coronary syndromes (ACS). B-type natriuretic peptide levels (BNP) and levels of the N-terminal fragment of the BNP prohormone (NT-proBNP) predict adverse long-term outcomes in patients with ACS. Investigators reviewed plasma samples of Troponin T (TnT) and NT-proBNP obtained from patients with ACS enrolled in the Platelet Receptor Inhibition in Ischemic Syndrome Management (PRISM) trial, which randomized patients to tirofiban or heparin for 48 hours and assessed mortality and myocardial infarction at 30 day follow-up. TnT and NT-proBNP determinations were available at baseline for 1791 patients, and at 48 and 72 hours from 1401 patients. Baseline NT-proBNP levels >250 ng/L were associated with significantly higher rates of death and myocardial infarction at 7 and 30 day follow-up. After adjustment for TnT and C-reactive protein levels, elevated NTproBNP levels maintained its predictive value (OR 2.7; p<.001). In patients with normal TnT levels, NT-proBNP levels identified a subgroup of patients at increased risk (OR 3.0; p=.004). However, in patients with high TnT levels (>0.1 mcg/L), NT-proBNP lost its predictive value (p=.58). More importantly, patients with normal levels of both TnT and NT-proBNP were at very low risk (0.6% event rate at 30 day follow-up).

Serial determinations of NT-proBNP levels at 48 and 72 hours were reviewed in patients without major adverse cardiac events (death or myocardial infarction); these patients were subdivided into groups with and without refractory ischemia. Patients without refractory ischemia showed a significant decline in NT-proBNP levels, whereas patients with refractory ischemia had no significant change. Persistently elevated NT-proBNP levels at 72 hours were associated with a 17.2% risk of death or MI at 30 days, compared with 0.6% risk if NT-proBNP returned to normal at 72 hours (p<.001). Neither TnT nor C-reactive protein demonstrated similar predictive value.

The study is limited by its retrospective nature, by potential selection bias by including only patients with direct evidence of coronary artery disease, and by limitations of the generalizability of its findings (e.g., to emergency department patients with chest pain).

As BNP and NT-proBNP are counter-regulatory hormones that play an active role in the response to ischemic injury, the authors suggest that NT-proBNP is a promising tool for dynamic risk assessment in patients with ACS. The authors also do not differentiate between BNP and NT-proBNP with regard to use in risk stratification, which might lead one to believe that these tests share similar predictive value. (Of note, the study was entirely funded by a company that produces an assay for NT-proBNP). Prospective trials to validate this tool are warranted

4. Heuschmann PU, Kolominsky-Rabas PL, Roether J, et al. Predictors of in-hospital mortality in patients with acute ischemic stroke treated with thrombolytic therapy. JAMA. 2004;292:1831-38.

The objective of this study was to identify factors associated with in-hospital mortality in ischemic stroke patients treated with recombinant tissue plasminogen activator (tPA). It was a prospective observational cohort study of 1658 patients conducted at 225 community and academic hospitals throughout Germany with main outcome of in-hospital mortality.

In this study 10% of patients who were treated with tPA died during their hospital stay, with 2/3 of deaths occurring in the first 7 days. Relative probability of in-hospital mortality increased with increasing patient age, with an odds ratio (OR) of 1.6 for each 10-year increment in age. Age was an independent predictor of in hospital mortality irrespective of tPA administration, with patients older than 75 years age having 4 fold higher mortality than the youngest cohort of less than 55 years age.

Other factors predicting in hospital mortality were altered level of consciousness and relative lack of experience with tPA treatment in the center. Altered level of consciousness was a predictor of stroke severity and an independent predictor of in-hospital mortality (OR 3.4). The increase in mortality in centers with limited experience with tPA administration (OR 0.97) reflected learning curve issues with these patients. The study was not designed to separate out the confounders of operator experience curve from institutional experience curves, or to derive the exact relationship between experience and outcomes.

5. McAlister FA, Bertsch K, Man J, et al. Incidence of and risk factors for pulmonary complications after non-thoracic surgery. Am J Respir Crit Care Med. 2004; published ahead of print on November 24, 2004 as doi:10.1164/rccm.200408-1069OC. Accessed January 27, 2005.

Postoperative pulmonary complications after nonthoracic surgery are a cause of significant morbidity and increased length of hospital stay. Previous studies of preoperative pulmonary assessment were limited by non-representative patient samples, conflicting results, and lack of explicit definitions of these complications. The authors conducted a prospective cohort study of 1055 patients seen in a Pre-Admission Clinic of a tertiary care university hospital. Mean age was 55 years, 50% male, and the cohort consisted of patients scheduled for intermediate risk elective surgery (upper abdominal, lower abdominal, orthopedic). They evaluated physical exam maneuvers (cough test, wheeze test, maximal laryngeal height, and forced expiratory time, all of which are described in an online data supplement) and preoperative spirometry values, and collected information on clinically significant postoperative pulmonary complications, including pneumonia, respiratory failure requiring mechanical ventilation, atelectasis requiring bronchoscopy, or pneumothorax or pleural effusion requiring percutaneous intervention. Twenty-eight patients (2.7%) suffered a pulmonary complication within 7 days of surgery, one of whom died. Length of stay was significantly prolonged in this group (mean 27.9 days vs 4.5 days, p=.006). Multivariate regression analysis revealed four variables that were independently associated with increased risk for postoperative pulmonary complications: age > 65 years, positive cough test (repeated coughing after asking the patient to inhale deeply and cough once), perioperative nasogastric tube, and anesthesia duration 2.5 hours or greater. Number of pack years smoked, FEV1, FEV1/FVC ratio, and upper abdominal surgery were associated with postoperative pulmonary complications but were not found to be independently associated by multivariate analysis.

While it is not surprising that the above risk factors are predictive of postoperative complications, this is the first study to incorporate specific exam maneuvers and spirometry into a risk prediction analysis. Limitations of this model are lack of independent validation and lack of generalizability to other populations, e.g., inpatients awaiting urgent surgery. Of note, the study further provides further support for not routinely obtaining pulmonary function testing for risk stratification prior to noncardiac surgery.

6. Mortenson, EM, Restrepo M, Anzeuto A, Pugh J. Effects of guideline-concordant antimicrobial therapy on mortality among patients with community-acquired pneumonia. Am J Med. 2004;117:726-31.

The American Thoracic Society (ATS) and the Infectious Diseases Society of America (IDSA) have published guidelines for the management of community acquired pneumonia that include recommendations for antibiotic selection. This retrospective cohort study attempted to measure the association between adherence to such guidelines and 30-day mortality in patients admitted with pneumonia to two Texas hospitals.

The characteristics of the patients studied reflect a reasonable cross-section of typical pneumonia patients, with the exception that the patients were mostly (85%) men. Of the study population, 78% were admitted through the ER, 20% admitted to the ICU, and 9% were nursing home residents. Antibiotics were considered “concordant” if they were consistent with either the most recent IDSA or ATS recommendations. The “nonconcordant” group was slightly older (66 vs. 61), generally sicker (higher rate of comorbid COPD and CVD), had more cerebrovascular disease, was more likely to present with altered mental status, and less frequently received antibiotics within 8 hours of presentation. The study did not comment on patients’ vaccination status. Thirty-day mortality was 6.2% in the guideline-concordant group, versus 21.7% in the other group (p < .001). The most common “non-concordant” regimen described was use of a beta-lactam alone, although specific antibiotic regimens were not evaluated.

While the results of this study are not surprising, they provide us with both the validation to continue practice according to existing recommendations, particularly the avoidance of monotherapy with beta-lactam antibiotics. The study also provides us with the imperative to take the lead in developing evidence-based pneumonia pathways at our own hospitals.

7. Pham MX, Whooley MA, Evans GT Jr, et al. Prognostic value of low-level cardiac troponin-I elevations in patients without definite acute coronary syndromes. Am Heart J. 2004;148: 776-82.

With the availability of rapid and highly sensitive and specific troponin testing, many patients admitted to the hospital with non-cardiac diagnoses have been recognized as having abnormal serum troponin-I or troponin-T levels, often just slightly above the reference cutoff for “normal.” While the clinical assumption is often that the elevated enzyme level does not reflect an acute coronary syndrome per se, its significance regarding the patient’s underlying cardiac health is often unclear.

Pham et al. retrospectively reviewed the 1-year mortality of 366 patients who were admitted to the San Francisco VA without evidence of acute MI or ACS either clinically or by EKG, but who had low-level troponin-I elevations (up to 3.0 ng/mL—a level that the authors state was reached by institutional consensus, and which was measured by a “first-generation” assay). These patients were admitted for a broad spectrum of diagnoses ranging from CHF to COPD to sepsis. Ninety-six percent of the patients were men; their average age was 69.

Follow-up was accomplished after a mean of 288 days and included 97% of patients. The primary endpoint was MI or death due to cardiac disease at one year; secondary endpoints were revascularization or admission for unstable angina. The primary endpoint was reached by 11% of patients with cTn-I between 1.0 and 3.0 ng/mL, and 4% of the patients with cTn-I up to 1.0 ng/mL (adjusted HR 3.4, 95% CI, 1.3 to 9.4), and the higher the cTn-I, the higher the risk. However, the authors did not test for the level of risk by specific diagnosis, so they caution against overgeneralizing their findings.

The findings of this study add to the evidence that any evidence of myocardial injury implies an increased risk of underlying heart disease and accompanying long—term cardiac complications– even if such injury occurs in the absence of ACS or known heart disease. Hospitalists often see such injury in the setting of acute infection and pulmonary disease and may be the first to recognize the possibility of CHD in a given patient. To date, guidelines addressing optimal prospective risk stratification have not been developed. Until they are, hospitalists should be aware of the ramifications of “troponin leak” and be prepared to initiate necessary inpatient monitoring and treatments, and to coordinate appropriate follow-up for these patients.

8. Saposnik G, Young B, Silver B, et al. Lack of improvement in patients with acute stroke after treatment with thrombolytic therapy: predictors and association with outcome. JAMA. 2004; 292: 1839-44.

Recombinant tissue plasminogen activator (tPA) has been shown to be one of the most efficacious therapies for acute stroke treatment. This was a systematic evaluation of predictors for outcomes at 24 hours after tPA therapy and of the prognostic significance of lack of improvement at 24 hours for long-term outcomes at 3 months.

The trial was a prospective cohort study of 216 consecutive patients admitted with acute ischemic stroke to a university hospital. The decision to treat with tPA was based on the NINDS protocol with one difference: patients with involvement of more than one third of the middle cerebral artery on the baseline CT scan were excluded. A control CT scan was performed at 24 hours to determine the presence of new infarction, cortical involvement, and extension of the ischemic lesion.

Lack of improvement was defined as a difference between the NIHSS score at baseline and at 24 hours of 3 points or less. Poor outcome at 3 months was defined by a modified Rankin Scale score of 3 to 5 or death.

After adjusting for age, gender, and stroke severity, hyperglycemia at admission (glucose > 144 mg/dL), cortical involvement, and time to treatment were independent predictors of lack of improvement at 24 hours. After adjusting for age, gender, and stroke severity, lack of improvement at 24 hours was an independent predictor of poor outcome and death at 3 months. Patients with lack of improvement at 24 hours also had longer lengths of hospitalization.

9. Taylor AL, Ziesche S, Yancy C, et al. Combination of isosorbide dinitrate and hydralazine in blacks with heart failure. N Engl J Med. 2004;351:2049-57.

Neurohormonal changes, endothelial dysfunction, impaired nitric oxide availability, and oxidant stress all contribute to the structural remodeling of the left ventricle in congestive heart failure. The combination of isosorbide dinitrate, an organic nitrate that stimulates nitric oxide signaling, and the antioxidant and vasodilator hydralazine improves survival in heart failure. Based on more recent data that black patients have a clinically significant response to this combination therapy, the authors of the African-American Heart Failure Trial (A-HeFT) evaluated 1050 black patients with congestive heart failure in a randomized, double-blind, placebo controlled trial. Patients were randomized to fixed doses of isosorbide dinitrate and hydralazine plus background therapy (i.e., digoxin, ACE inhibitors, beta-blockers, diuretics, angiotensin receptor blockers) or to placebo plus background therapy. After 18 months, the trial was stopped due to a significantly higher mortality rate in the placebo group (10.2% in the placebo group vs. 6.2% with combination

therapy, p=.02); survival differences emerged at 180 days and increased progressively thereafter. The combination therapy group reported more headache and dizziness but suffered fewer exacerbations of congestive heart failure and reported improvement in subjective assessments of quality of life as measured by questionnaires. Accompanying editorials discuss the role of nitric oxide and prevention of oxidative stress in the treatment of heart failure, as well as the controversial issues surrounding race-based therapeutics.

CARP Trial Suggests No Benefit to Revascularization Before Vascular Surgery

McFalls, EO, Ward HB, Mortiz TE, et al. Coronary-artery revascularization before elective major vascular surgery. N Engl J Med. 2004;351:2795-2804.

Recent studies have presented evidence that treatment with beta-blockers for patients with CAD could reduce the risk of perioperative cardiac complications. Beta-blockers have since become a critical part of the management plan for the perioperative patient. Evidence-based practice guidelines for cardiac risk assessment have been published by both the American College of Physicians and the American College of Cardiology/American Heart Association. However, practice patterns continue to vary between physicians and cardiologists, particularly for patients clinically stratified into the intermediate-risk category. Some physicians feel comfortable with a conservative approach of medical optimization even in the setting of established CAD, while others favor more aggressive treatment, even though the prospective data supporting cardiac revascularization before major surgery has been lacking. The study investigators sought to clarify this uncertainty.

The prospective trial enrolled 510 patients at 18 VA centers. Patients scheduled for major vascular operations were eligible, and were preoperatively assessed via clinical criteria, stress imaging, and angiography when appropriate. Eligible patients had significant (at least 70%) stenosis of at least one coronary artery. High-risk patients (i.e., those with left main disease, severe aortic stenosis, and LVEF <20%) were excluded. Patients were then randomized to one

of two groups. The first group underwent revascularization with PTCA or CABG plus medical optimization; the second group received only medical optimization. Most patients in both groups received beta-blockers, and more than half in each received statins. The patient populations were appropriately randomized, although overwhelmingly male (98%). Most patients had one- or two-vessel CAD. The primary endpoint was long-term mortality. Secondary endpoints included MI, stroke, renal failure requiring dialysis, and limb loss. Follow-up rates were similar in both groups (86% and 85%).

The major finding of the study was the lack of difference in mortality between the two groups at an average follow-up of 2.7 years (22% vs. 23%, RR= 0.98, 95% CI 0.70 to 1.37, p = 0.92). Analyzing by “treatment-received” instead of “intention-to-treat” did not significantly change this result. Of note, ten patients in the revascularization arm died between the revascularization procedure and the vascular surgery. Not surprisingly, revascularization also delayed the time to surgery for patients in that arm of the study. In the authors’ analysis, the patients were also divided into subgroups based on high-risk variables (prior CABG, category of Revised Cardiac Risk Index, etc.), but the study was not powered to detect mortality differences between the two arms within these subgroups. The authors concluded that there was no benefit to revascularization in patients with stable coronary syndromes prior to elective vascular surgeries.

The results of this study validate the conservative practice recommended by the existing guidelines— that is, to perform revascularization procedures in the preoperative setting only when indicated by clinical criteria such as unstable ischemic symptoms, and if likely to improve long-term survival. Beta-blockers, and based on recent studies probably “statins,” should continue to be the mainstay of perioperative risk optimization for patients with stable coronary disease.

There were, however, several important considerations: first, the study group was exclusively male, although there is little reason to believe that women would have better outcomes from revascularization. And second, the highest-risk patients were excluded, and therefore the results should not be extrapolated to that population. Prospective identification of the group of patients who may benefit from aggressive intervention should remain a target of risk assessment and further research. (BH)

Blood Transfusion May Increase Mortality in Acute Coronary Syndrome

Rao SV, Jollis JG, Harrington RA, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA. 2004;292:1555-62.

The increased use of invasive procedures and anticoagulant and fibrinolytic drugs in patients with ischemic heart disease in recent years predictably increases the potential for bleeding and perceived need for transfusion. Studies evaluating the association between transfusion and mortality have produced mixed results. A more pertinent clinical question is whether transfusion is beneficial or harmful in patients with acute coronary syndromes who acutely develop anemia during their hospitalization.

The authors used clinical data from three large international trials of patients with acute coronary syndromes (GUSTO IIb, PURSUIT, and PARAGON B) to determine the association between blood transfusion and outcomes among patients who developed moderate to severe bleeding, anemia, or both during their hospitalization.

Assessment of clinically significant bleeding complications was based on the GUSTO definition of severe (intracranial hemorrhage or hemodynamic compromise and requiring intervention) or moderate (hemodynamically stable but requiring blood transfusion) bleeding. The GUSTO IIb and PURSUIT trials used the above definition; PARAGON B categorized bleeding as “major or life threatening” (intracranial hemorrhage or bleeding leading to hemodynamic compromise requiring intervention) or “intermediate” (requiring transfusion or a decrease in hemoglobin of 5 g/dL or more, or a decrease in hematocrit ( 15%). Major or life-threatening bleeding episodes and intermediate bleeding episodes in PARAGON B were deemed equivalent to severe and moderate bleeding episodes in GUSTO.

Data were collected on the date, time, severity, and location of each bleeding event, and on the date and number of units of packed red blood cells and whole blood transfused. The primary end-point was 30-day all-cause mortality. Secondary end-points were occurrence of the composite of 30-day death or MI.

The unadjusted rates of 30-day death, MI, and composite death/MI were significantly higher among patients who received a transfusion (30-day death, 8.00% vs. 3.08%; p<.001; 30-day MI, 25.16% vs. 8.16%; p<.001; 30-day composite death/MI, 29.24% vs. 10.02%; p<.001).

After adjustment for baseline characteristics, bleeding and transfusion propensity, and nadir hematocrit, blood transfusion was associated with a hazard ratio for death of 3.94 (95% confidence interval, 3.26–4.75).

No significant association was found between transfusion and 30-day mortality at a nadir hematocrit of 25% or less (adjusted OR 1.13; 95% CI 0.70-1.82). However, at a nadir hematocrit higher than 25%, transfusion was associated with significantly higher odds of 30-day death, even after excluding patients who underwent CABG or those who died within the first 5 days of follow-up.

These findings differ from the findings of Wu et al. (1) who noted that blood transfusion was associated with lower 30-day mortality among elderly patients with MI if the admission hematocrit was 30% or lower. The current authors propose that their data is more robust due to meticulous collection through clinical trial records, and that their analysis accounts for timing of transfusion and indications for transfusion.

Many clinicians logically believe that augmentation of oxygen carrying capacity via transfusion would be beneficial to patients with active ischemia. However, the authors note that red blood cells in stored blood may be depleted of both 2,3-diphosphoglyceric acid and nitric oxide, both of which are critical components to oxygen delivery and exchange. These cells then function as nitric oxide “sinks,” promoting vasoconstriction, platelet aggregation, and impaired oxygen delivery to tissues. In addition, inflammatory mediators associated with exacerbation of myocardial ischemia may remain in transfused blood, potentially contributing to adverse outcomes.

As this is a nonrandomized, post hoc observational study, further prescriptive conclusions regarding transfusion cannot be made. However, the authors, along with an accompanying editorial, call for prospective randomized trials of transfusion in anemic patients with acute coronary syndromes to better define the role of this commonly used therapy. (CW)

1. Wu WC, Rathore SS, Wang Y, Radford MJ, Krumholz HM. Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med. 2001;345:1230-6.

Cost-effectiveness of Rhythm Versus Rate Control in Atrial Fibrillation

Marshall DA, Levy AR, Vidaillet H, et al. Cost-effectiveness of rhythm versus rate control in atrial fibrillation. Ann Intern Med. 2004;141:653-61.

Atrial fibrillation is the most common arrhythmia treated by physicians. It afflicts nearly 10% of patients age 80 years or older. There are two primary modalities of managing patients with atrial fibrillation; rate control versus cardioversion into sinus rhythm.

AFFIRM was a multicenter randomized controlled trial involving 213 centers in USA and Canada that compared rate versus rhythm control in 4060 patients. These patients had a mean age of 70 years. Sixty-one percent of the enrolled patients were men, and 66% had recurrent atrial fibrillation. Seventy-one percent of patients had hypertension, 39% had coronary artery disease, and 9% had congestive heart failure. Patients were block randomized by center to either rate control or rhythm control and followed for an average of 3.5 years. AFFIRM results showed no significant mortality difference between the two groups (hazard ratio for rate versus rhythm control, 0.87 with 95% CI, 0.75 to 1.01).

Primary data on survival and resource utilization were used to conduct the economic analysis from a third party payer perspective. Authors used intention to treat data for the economic analysis.

For resource utilization estimates, US healthcare cost figures for the year 2002 were used. All earlier costs were appropriately adjusted using Consumer Price Index, Medical Care component to estimate their nominal values in year 2002. Hospital costs were taken as the mean charges per day from Healthcare Cost and Utilization Project statistics for Diseases of the Circulatory System for patients older than 18 years age. Low and high end of these costs were assumed to be equivalent to 25th and 75th percentiles of the mean costs, respectively. Physician costs were assumed to be equivalent to be the average of all carriers’ payments for the relative value units of the services rendered based on a generic current procedural terminology code. Sensitivity analysis was conducted on these physician costs using minimum payment among these carriers as the low cost estimate and the standard charges for Marshfield Clinic for the high end. Costs of pacemaker and implantable cardioverter defibrillators were based on manufacturers’ list prices. For sensitivity analysis, hardware costs were excluded for low cost estimates and the maximum manufacturers’ list price for the high estimate was used.

At each follow-up visit during the AFFIRM trial, the number of cardioversion attempts since the prior visit was recorded. Costs of cardioversion were based on average payment to Marshfield Clinic for outpatient electrical cardioversion for the year 2002. Authors assigned no costs for low cost estimate and used billed charges for high costs for sensitivity analysis.

At each follow-up visit, the number of short stay and emergency department visits since the prior visit was recorded. Weighted average Medicare costs for level I and II facilities were used as the baseline estimate for these visits. Estimates for sensitivity analysis were the minimum and maximum Medicare payments for these visits. Physician fees were based on level III emergency department visit with low and high cost estimates assigned as described above.

Medication costs were based on the least average wholesale price (AWP) for a generic medication. Low and high cost estimates were taken from lowest quoted US Internet pharmacy price and highest AWP for the most expensive drug in the class respectively. Only medications used for atrial fibrillation and anticoagulation were considered for analysis.

The authors calculated the mean cost per patient In the Literature (continued) and the mean survival time between the two interventions. Future costs were discounted by 3%. For the base estimate, rhythm control was more expensive and less effective than rate control, i.e., dominated by the rate control. Rate control dominated rhythm control even for high and low estimates of the sensitivity analysis demonstrating stability of the results. Authors used 10,000 simulations to perform non-parametric bootstrapping analysis to find the 95% credible intervals around the base estimate. The bootstrap results showed that for 95% of the results rate control had higher survival time and was less costly than rhythm control. These simulation results clearly showed rate control is more cost-effective for patient population resembling that of AFFIRM trial.

The study has some limitations. These results are robust for patients similar to those in AFFIRM trial, i.e., older patients with cardiovascular defects that are at risk of cerebrovascular embolism. However these results may not be applicable to younger patients and those with “lone atrial fibrillation.” The study had a follow-up period of 3.5 years, and the cost-effectiveness analysis is confined to this period. It is difficult to determine mortality advantage of one treatment over another within the limited duration of the AFFIRM study. Most of the patients were on multiple pharmacologic agents for rhythm control and had a high incidence of cross-over from rhythm control to rate control reflecting modest benefits of the current agents. These results may not be applicable to patients whose atrial fibrillation is well controlled by a single agent or by non-pharmacological treatment. Patients on rhythm control agents had, as expected, more hospitalization days from the side effects and treatment protocols of the agents (especially pharmacologic) used to control the rhythm. With advances in both pharmacologic as well as nonpharmacologic methods for rhythm control generating safer and more efficacious technologies, the results of this analysis may become less valid in the future. The analysis was conducted from a third-party payer perspective, without accounting for the quality of life. Thus patients who have symptomatic atrial fibrillation and those with diastolic dysfunction may have improved quality of life from rhythm control over just rate control. The results may not be applicable to these patients. (SS)

CARP Trial Suggests No Benefit to Revascularization Before Vascular Surgery

McFalls, EO, Ward HB, Mortiz TE, et al. Coronary-artery revascularization before elective major vascular surgery. N Engl J Med. 2004;351:2795-2804.

Recent studies have presented evidence that treatment with beta-blockers for patients with CAD could reduce the risk of perioperative cardiac complications. Beta-blockers have since become a critical part of the management plan for the perioperative patient. Evidence-based practice guidelines for cardiac risk assessment have been published by both the American College of Physicians and the American College of Cardiology/American Heart Association. However, practice patterns continue to vary between physicians and cardiologists, particularly for patients clinically stratified into the intermediate-risk category. Some physicians feel comfortable with a conservative approach of medical optimization even in the setting of established CAD, while others favor more aggressive treatment, even though the prospective data supporting cardiac revascularization before major surgery has been lacking. The study investigators sought to clarify this uncertainty.

The prospective trial enrolled 510 patients at 18 VA centers. Patients scheduled for major vascular operations were eligible, and were preoperatively assessed via clinical criteria, stress imaging, and angiography when appropriate. Eligible patients had significant (at least 70%) stenosis of at least one coronary artery. High-risk patients (i.e., those with left main disease, severe aortic stenosis, and LVEF <20%) were excluded. Patients were then randomized to one

of two groups. The first group underwent revascularization with PTCA or CABG plus medical optimization; the second group received only medical optimization. Most patients in both groups received beta-blockers, and more than half in each received statins. The patient populations were appropriately randomized, although overwhelmingly male (98%). Most patients had one- or two-vessel CAD. The primary endpoint was long-term mortality. Secondary endpoints included MI, stroke, renal failure requiring dialysis, and limb loss. Follow-up rates were similar in both groups (86% and 85%).

The major finding of the study was the lack of difference in mortality between the two groups at an average follow-up of 2.7 years (22% vs. 23%, RR= 0.98, 95% CI 0.70 to 1.37, p = 0.92). Analyzing by “treatment-received” instead of “intention-to-treat” did not significantly change this result. Of note, ten patients in the revascularization arm died between the revascularization procedure and the vascular surgery. Not surprisingly, revascularization also delayed the time to surgery for patients in that arm of the study. In the authors’ analysis, the patients were also divided into subgroups based on high-risk variables (prior CABG, category of Revised Cardiac Risk Index, etc.), but the study was not powered to detect mortality differences between the two arms within these subgroups. The authors concluded that there was no benefit to revascularization in patients with stable coronary syndromes prior to elective vascular surgeries.

The results of this study validate the conservative practice recommended by the existing guidelines— that is, to perform revascularization procedures in the preoperative setting only when indicated by clinical criteria such as unstable ischemic symptoms, and if likely to improve long-term survival. Beta-blockers, and based on recent studies probably “statins,” should continue to be the mainstay of perioperative risk optimization for patients with stable coronary disease.

There were, however, several important considerations: first, the study group was exclusively male, although there is little reason to believe that women would have better outcomes from revascularization. And second, the highest-risk patients were excluded, and therefore the results should not be extrapolated to that population. Prospective identification of the group of patients who may benefit from aggressive intervention should remain a target of risk assessment and further research. (BH)

Blood Transfusion May Increase Mortality in Acute Coronary Syndrome

Rao SV, Jollis JG, Harrington RA, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA. 2004;292:1555-62.

The increased use of invasive procedures and anticoagulant and fibrinolytic drugs in patients with ischemic heart disease in recent years predictably increases the potential for bleeding and perceived need for transfusion. Studies evaluating the association between transfusion and mortality have produced mixed results. A more pertinent clinical question is whether transfusion is beneficial or harmful in patients with acute coronary syndromes who acutely develop anemia during their hospitalization.

The authors used clinical data from three large international trials of patients with acute coronary syndromes (GUSTO IIb, PURSUIT, and PARAGON B) to determine the association between blood transfusion and outcomes among patients who developed moderate to severe bleeding, anemia, or both during their hospitalization.

Assessment of clinically significant bleeding complications was based on the GUSTO definition of severe (intracranial hemorrhage or hemodynamic compromise and requiring intervention) or moderate (hemodynamically stable but requiring blood transfusion) bleeding. The GUSTO IIb and PURSUIT trials used the above definition; PARAGON B categorized bleeding as “major or life threatening” (intracranial hemorrhage or bleeding leading to hemodynamic compromise requiring intervention) or “intermediate” (requiring transfusion or a decrease in hemoglobin of 5 g/dL or more, or a decrease in hematocrit ( 15%). Major or life-threatening bleeding episodes and intermediate bleeding episodes in PARAGON B were deemed equivalent to severe and moderate bleeding episodes in GUSTO.

Data were collected on the date, time, severity, and location of each bleeding event, and on the date and number of units of packed red blood cells and whole blood transfused. The primary end-point was 30-day all-cause mortality. Secondary end-points were occurrence of the composite of 30-day death or MI.

The unadjusted rates of 30-day death, MI, and composite death/MI were significantly higher among patients who received a transfusion (30-day death, 8.00% vs. 3.08%; p<.001; 30-day MI, 25.16% vs. 8.16%; p<.001; 30-day composite death/MI, 29.24% vs. 10.02%; p<.001).

After adjustment for baseline characteristics, bleeding and transfusion propensity, and nadir hematocrit, blood transfusion was associated with a hazard ratio for death of 3.94 (95% confidence interval, 3.26–4.75).

No significant association was found between transfusion and 30-day mortality at a nadir hematocrit of 25% or less (adjusted OR 1.13; 95% CI 0.70-1.82). However, at a nadir hematocrit higher than 25%, transfusion was associated with significantly higher odds of 30-day death, even after excluding patients who underwent CABG or those who died within the first 5 days of follow-up.

These findings differ from the findings of Wu et al. (1) who noted that blood transfusion was associated with lower 30-day mortality among elderly patients with MI if the admission hematocrit was 30% or lower. The current authors propose that their data is more robust due to meticulous collection through clinical trial records, and that their analysis accounts for timing of transfusion and indications for transfusion.

Many clinicians logically believe that augmentation of oxygen carrying capacity via transfusion would be beneficial to patients with active ischemia. However, the authors note that red blood cells in stored blood may be depleted of both 2,3-diphosphoglyceric acid and nitric oxide, both of which are critical components to oxygen delivery and exchange. These cells then function as nitric oxide “sinks,” promoting vasoconstriction, platelet aggregation, and impaired oxygen delivery to tissues. In addition, inflammatory mediators associated with exacerbation of myocardial ischemia may remain in transfused blood, potentially contributing to adverse outcomes.

As this is a nonrandomized, post hoc observational study, further prescriptive conclusions regarding transfusion cannot be made. However, the authors, along with an accompanying editorial, call for prospective randomized trials of transfusion in anemic patients with acute coronary syndromes to better define the role of this commonly used therapy. (CW)

1. Wu WC, Rathore SS, Wang Y, Radford MJ, Krumholz HM. Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med. 2001;345:1230-6.

Cost-effectiveness of Rhythm Versus Rate Control in Atrial Fibrillation

Marshall DA, Levy AR, Vidaillet H, et al. Cost-effectiveness of rhythm versus rate control in atrial fibrillation. Ann Intern Med. 2004;141:653-61.

Atrial fibrillation is the most common arrhythmia treated by physicians. It afflicts nearly 10% of patients age 80 years or older. There are two primary modalities of managing patients with atrial fibrillation; rate control versus cardioversion into sinus rhythm.

AFFIRM was a multicenter randomized controlled trial involving 213 centers in USA and Canada that compared rate versus rhythm control in 4060 patients. These patients had a mean age of 70 years. Sixty-one percent of the enrolled patients were men, and 66% had recurrent atrial fibrillation. Seventy-one percent of patients had hypertension, 39% had coronary artery disease, and 9% had congestive heart failure. Patients were block randomized by center to either rate control or rhythm control and followed for an average of 3.5 years. AFFIRM results showed no significant mortality difference between the two groups (hazard ratio for rate versus rhythm control, 0.87 with 95% CI, 0.75 to 1.01).

Primary data on survival and resource utilization were used to conduct the economic analysis from a third party payer perspective. Authors used intention to treat data for the economic analysis.

For resource utilization estimates, US healthcare cost figures for the year 2002 were used. All earlier costs were appropriately adjusted using Consumer Price Index, Medical Care component to estimate their nominal values in year 2002. Hospital costs were taken as the mean charges per day from Healthcare Cost and Utilization Project statistics for Diseases of the Circulatory System for patients older than 18 years age. Low and high end of these costs were assumed to be equivalent to 25th and 75th percentiles of the mean costs, respectively. Physician costs were assumed to be equivalent to be the average of all carriers’ payments for the relative value units of the services rendered based on a generic current procedural terminology code. Sensitivity analysis was conducted on these physician costs using minimum payment among these carriers as the low cost estimate and the standard charges for Marshfield Clinic for the high end. Costs of pacemaker and implantable cardioverter defibrillators were based on manufacturers’ list prices. For sensitivity analysis, hardware costs were excluded for low cost estimates and the maximum manufacturers’ list price for the high estimate was used.

At each follow-up visit during the AFFIRM trial, the number of cardioversion attempts since the prior visit was recorded. Costs of cardioversion were based on average payment to Marshfield Clinic for outpatient electrical cardioversion for the year 2002. Authors assigned no costs for low cost estimate and used billed charges for high costs for sensitivity analysis.

At each follow-up visit, the number of short stay and emergency department visits since the prior visit was recorded. Weighted average Medicare costs for level I and II facilities were used as the baseline estimate for these visits. Estimates for sensitivity analysis were the minimum and maximum Medicare payments for these visits. Physician fees were based on level III emergency department visit with low and high cost estimates assigned as described above.

Medication costs were based on the least average wholesale price (AWP) for a generic medication. Low and high cost estimates were taken from lowest quoted US Internet pharmacy price and highest AWP for the most expensive drug in the class respectively. Only medications used for atrial fibrillation and anticoagulation were considered for analysis.

The authors calculated the mean cost per patient In the Literature (continued) and the mean survival time between the two interventions. Future costs were discounted by 3%. For the base estimate, rhythm control was more expensive and less effective than rate control, i.e., dominated by the rate control. Rate control dominated rhythm control even for high and low estimates of the sensitivity analysis demonstrating stability of the results. Authors used 10,000 simulations to perform non-parametric bootstrapping analysis to find the 95% credible intervals around the base estimate. The bootstrap results showed that for 95% of the results rate control had higher survival time and was less costly than rhythm control. These simulation results clearly showed rate control is more cost-effective for patient population resembling that of AFFIRM trial.

The study has some limitations. These results are robust for patients similar to those in AFFIRM trial, i.e., older patients with cardiovascular defects that are at risk of cerebrovascular embolism. However these results may not be applicable to younger patients and those with “lone atrial fibrillation.” The study had a follow-up period of 3.5 years, and the cost-effectiveness analysis is confined to this period. It is difficult to determine mortality advantage of one treatment over another within the limited duration of the AFFIRM study. Most of the patients were on multiple pharmacologic agents for rhythm control and had a high incidence of cross-over from rhythm control to rate control reflecting modest benefits of the current agents. These results may not be applicable to patients whose atrial fibrillation is well controlled by a single agent or by non-pharmacological treatment. Patients on rhythm control agents had, as expected, more hospitalization days from the side effects and treatment protocols of the agents (especially pharmacologic) used to control the rhythm. With advances in both pharmacologic as well as nonpharmacologic methods for rhythm control generating safer and more efficacious technologies, the results of this analysis may become less valid in the future. The analysis was conducted from a third-party payer perspective, without accounting for the quality of life. Thus patients who have symptomatic atrial fibrillation and those with diastolic dysfunction may have improved quality of life from rhythm control over just rate control. The results may not be applicable to these patients. (SS)

CARP Trial Suggests No Benefit to Revascularization Before Vascular Surgery

McFalls, EO, Ward HB, Mortiz TE, et al. Coronary-artery revascularization before elective major vascular surgery. N Engl J Med. 2004;351:2795-2804.

Recent studies have presented evidence that treatment with beta-blockers for patients with CAD could reduce the risk of perioperative cardiac complications. Beta-blockers have since become a critical part of the management plan for the perioperative patient. Evidence-based practice guidelines for cardiac risk assessment have been published by both the American College of Physicians and the American College of Cardiology/American Heart Association. However, practice patterns continue to vary between physicians and cardiologists, particularly for patients clinically stratified into the intermediate-risk category. Some physicians feel comfortable with a conservative approach of medical optimization even in the setting of established CAD, while others favor more aggressive treatment, even though the prospective data supporting cardiac revascularization before major surgery has been lacking. The study investigators sought to clarify this uncertainty.

The prospective trial enrolled 510 patients at 18 VA centers. Patients scheduled for major vascular operations were eligible, and were preoperatively assessed via clinical criteria, stress imaging, and angiography when appropriate. Eligible patients had significant (at least 70%) stenosis of at least one coronary artery. High-risk patients (i.e., those with left main disease, severe aortic stenosis, and LVEF <20%) were excluded. Patients were then randomized to one

of two groups. The first group underwent revascularization with PTCA or CABG plus medical optimization; the second group received only medical optimization. Most patients in both groups received beta-blockers, and more than half in each received statins. The patient populations were appropriately randomized, although overwhelmingly male (98%). Most patients had one- or two-vessel CAD. The primary endpoint was long-term mortality. Secondary endpoints included MI, stroke, renal failure requiring dialysis, and limb loss. Follow-up rates were similar in both groups (86% and 85%).

The major finding of the study was the lack of difference in mortality between the two groups at an average follow-up of 2.7 years (22% vs. 23%, RR= 0.98, 95% CI 0.70 to 1.37, p = 0.92). Analyzing by “treatment-received” instead of “intention-to-treat” did not significantly change this result. Of note, ten patients in the revascularization arm died between the revascularization procedure and the vascular surgery. Not surprisingly, revascularization also delayed the time to surgery for patients in that arm of the study. In the authors’ analysis, the patients were also divided into subgroups based on high-risk variables (prior CABG, category of Revised Cardiac Risk Index, etc.), but the study was not powered to detect mortality differences between the two arms within these subgroups. The authors concluded that there was no benefit to revascularization in patients with stable coronary syndromes prior to elective vascular surgeries.

The results of this study validate the conservative practice recommended by the existing guidelines— that is, to perform revascularization procedures in the preoperative setting only when indicated by clinical criteria such as unstable ischemic symptoms, and if likely to improve long-term survival. Beta-blockers, and based on recent studies probably “statins,” should continue to be the mainstay of perioperative risk optimization for patients with stable coronary disease.

There were, however, several important considerations: first, the study group was exclusively male, although there is little reason to believe that women would have better outcomes from revascularization. And second, the highest-risk patients were excluded, and therefore the results should not be extrapolated to that population. Prospective identification of the group of patients who may benefit from aggressive intervention should remain a target of risk assessment and further research. (BH)

Blood Transfusion May Increase Mortality in Acute Coronary Syndrome

Rao SV, Jollis JG, Harrington RA, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA. 2004;292:1555-62.

The increased use of invasive procedures and anticoagulant and fibrinolytic drugs in patients with ischemic heart disease in recent years predictably increases the potential for bleeding and perceived need for transfusion. Studies evaluating the association between transfusion and mortality have produced mixed results. A more pertinent clinical question is whether transfusion is beneficial or harmful in patients with acute coronary syndromes who acutely develop anemia during their hospitalization.

The authors used clinical data from three large international trials of patients with acute coronary syndromes (GUSTO IIb, PURSUIT, and PARAGON B) to determine the association between blood transfusion and outcomes among patients who developed moderate to severe bleeding, anemia, or both during their hospitalization.

Assessment of clinically significant bleeding complications was based on the GUSTO definition of severe (intracranial hemorrhage or hemodynamic compromise and requiring intervention) or moderate (hemodynamically stable but requiring blood transfusion) bleeding. The GUSTO IIb and PURSUIT trials used the above definition; PARAGON B categorized bleeding as “major or life threatening” (intracranial hemorrhage or bleeding leading to hemodynamic compromise requiring intervention) or “intermediate” (requiring transfusion or a decrease in hemoglobin of 5 g/dL or more, or a decrease in hematocrit ( 15%). Major or life-threatening bleeding episodes and intermediate bleeding episodes in PARAGON B were deemed equivalent to severe and moderate bleeding episodes in GUSTO.

Data were collected on the date, time, severity, and location of each bleeding event, and on the date and number of units of packed red blood cells and whole blood transfused. The primary end-point was 30-day all-cause mortality. Secondary end-points were occurrence of the composite of 30-day death or MI.

The unadjusted rates of 30-day death, MI, and composite death/MI were significantly higher among patients who received a transfusion (30-day death, 8.00% vs. 3.08%; p<.001; 30-day MI, 25.16% vs. 8.16%; p<.001; 30-day composite death/MI, 29.24% vs. 10.02%; p<.001).

After adjustment for baseline characteristics, bleeding and transfusion propensity, and nadir hematocrit, blood transfusion was associated with a hazard ratio for death of 3.94 (95% confidence interval, 3.26–4.75).

No significant association was found between transfusion and 30-day mortality at a nadir hematocrit of 25% or less (adjusted OR 1.13; 95% CI 0.70-1.82). However, at a nadir hematocrit higher than 25%, transfusion was associated with significantly higher odds of 30-day death, even after excluding patients who underwent CABG or those who died within the first 5 days of follow-up.

These findings differ from the findings of Wu et al. (1) who noted that blood transfusion was associated with lower 30-day mortality among elderly patients with MI if the admission hematocrit was 30% or lower. The current authors propose that their data is more robust due to meticulous collection through clinical trial records, and that their analysis accounts for timing of transfusion and indications for transfusion.

Many clinicians logically believe that augmentation of oxygen carrying capacity via transfusion would be beneficial to patients with active ischemia. However, the authors note that red blood cells in stored blood may be depleted of both 2,3-diphosphoglyceric acid and nitric oxide, both of which are critical components to oxygen delivery and exchange. These cells then function as nitric oxide “sinks,” promoting vasoconstriction, platelet aggregation, and impaired oxygen delivery to tissues. In addition, inflammatory mediators associated with exacerbation of myocardial ischemia may remain in transfused blood, potentially contributing to adverse outcomes.

As this is a nonrandomized, post hoc observational study, further prescriptive conclusions regarding transfusion cannot be made. However, the authors, along with an accompanying editorial, call for prospective randomized trials of transfusion in anemic patients with acute coronary syndromes to better define the role of this commonly used therapy. (CW)

1. Wu WC, Rathore SS, Wang Y, Radford MJ, Krumholz HM. Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med. 2001;345:1230-6.

Cost-effectiveness of Rhythm Versus Rate Control in Atrial Fibrillation

Marshall DA, Levy AR, Vidaillet H, et al. Cost-effectiveness of rhythm versus rate control in atrial fibrillation. Ann Intern Med. 2004;141:653-61.

Atrial fibrillation is the most common arrhythmia treated by physicians. It afflicts nearly 10% of patients age 80 years or older. There are two primary modalities of managing patients with atrial fibrillation; rate control versus cardioversion into sinus rhythm.

AFFIRM was a multicenter randomized controlled trial involving 213 centers in USA and Canada that compared rate versus rhythm control in 4060 patients. These patients had a mean age of 70 years. Sixty-one percent of the enrolled patients were men, and 66% had recurrent atrial fibrillation. Seventy-one percent of patients had hypertension, 39% had coronary artery disease, and 9% had congestive heart failure. Patients were block randomized by center to either rate control or rhythm control and followed for an average of 3.5 years. AFFIRM results showed no significant mortality difference between the two groups (hazard ratio for rate versus rhythm control, 0.87 with 95% CI, 0.75 to 1.01).

Primary data on survival and resource utilization were used to conduct the economic analysis from a third party payer perspective. Authors used intention to treat data for the economic analysis.

For resource utilization estimates, US healthcare cost figures for the year 2002 were used. All earlier costs were appropriately adjusted using Consumer Price Index, Medical Care component to estimate their nominal values in year 2002. Hospital costs were taken as the mean charges per day from Healthcare Cost and Utilization Project statistics for Diseases of the Circulatory System for patients older than 18 years age. Low and high end of these costs were assumed to be equivalent to 25th and 75th percentiles of the mean costs, respectively. Physician costs were assumed to be equivalent to be the average of all carriers’ payments for the relative value units of the services rendered based on a generic current procedural terminology code. Sensitivity analysis was conducted on these physician costs using minimum payment among these carriers as the low cost estimate and the standard charges for Marshfield Clinic for the high end. Costs of pacemaker and implantable cardioverter defibrillators were based on manufacturers’ list prices. For sensitivity analysis, hardware costs were excluded for low cost estimates and the maximum manufacturers’ list price for the high estimate was used.

At each follow-up visit during the AFFIRM trial, the number of cardioversion attempts since the prior visit was recorded. Costs of cardioversion were based on average payment to Marshfield Clinic for outpatient electrical cardioversion for the year 2002. Authors assigned no costs for low cost estimate and used billed charges for high costs for sensitivity analysis.

At each follow-up visit, the number of short stay and emergency department visits since the prior visit was recorded. Weighted average Medicare costs for level I and II facilities were used as the baseline estimate for these visits. Estimates for sensitivity analysis were the minimum and maximum Medicare payments for these visits. Physician fees were based on level III emergency department visit with low and high cost estimates assigned as described above.

Medication costs were based on the least average wholesale price (AWP) for a generic medication. Low and high cost estimates were taken from lowest quoted US Internet pharmacy price and highest AWP for the most expensive drug in the class respectively. Only medications used for atrial fibrillation and anticoagulation were considered for analysis.

The authors calculated the mean cost per patient In the Literature (continued) and the mean survival time between the two interventions. Future costs were discounted by 3%. For the base estimate, rhythm control was more expensive and less effective than rate control, i.e., dominated by the rate control. Rate control dominated rhythm control even for high and low estimates of the sensitivity analysis demonstrating stability of the results. Authors used 10,000 simulations to perform non-parametric bootstrapping analysis to find the 95% credible intervals around the base estimate. The bootstrap results showed that for 95% of the results rate control had higher survival time and was less costly than rhythm control. These simulation results clearly showed rate control is more cost-effective for patient population resembling that of AFFIRM trial.

The study has some limitations. These results are robust for patients similar to those in AFFIRM trial, i.e., older patients with cardiovascular defects that are at risk of cerebrovascular embolism. However these results may not be applicable to younger patients and those with “lone atrial fibrillation.” The study had a follow-up period of 3.5 years, and the cost-effectiveness analysis is confined to this period. It is difficult to determine mortality advantage of one treatment over another within the limited duration of the AFFIRM study. Most of the patients were on multiple pharmacologic agents for rhythm control and had a high incidence of cross-over from rhythm control to rate control reflecting modest benefits of the current agents. These results may not be applicable to patients whose atrial fibrillation is well controlled by a single agent or by non-pharmacological treatment. Patients on rhythm control agents had, as expected, more hospitalization days from the side effects and treatment protocols of the agents (especially pharmacologic) used to control the rhythm. With advances in both pharmacologic as well as nonpharmacologic methods for rhythm control generating safer and more efficacious technologies, the results of this analysis may become less valid in the future. The analysis was conducted from a third-party payer perspective, without accounting for the quality of life. Thus patients who have symptomatic atrial fibrillation and those with diastolic dysfunction may have improved quality of life from rhythm control over just rate control. The results may not be applicable to these patients. (SS)