It is estimated that the Medicare Part A trust fund will be exhausted by 2016 to 2019; also, the quality of care delivered in the United States is highly variable.13 Value is typically defined as the quality achieved for a given cost (ie, value = quality/cost). The focus on the 3 related concepts of value, quality, and cost of health care is likely to continue to increase. Previously, the U.S. Department of Health and Human Services (HHS) made value‐driven health care one of the Department's top priorities.4 Policymakers are in a period of transition but the publicly available plans of the President and Senate leadership indicate that the focus on value‐based initiatives will likely continue to increase as our nation strives to achieve better outcomes for our health care dollar.5, 6 Specifically, the federal government and other payers increasingly align payment incentives with value and quality, encourage public reporting on quality and Medicare payment costs, such as on the Hospital Compare website (http://www.hospitalcompare.hhs.gov), and implement and evaluate demonstrations to test mechanisms such as health information technology (HIT) to improve value‐based performance.

Since hospital care represented $648 billion in 2006, which is 37% of the total patient‐related U.S. health care expenditure, the trend to pay for value will likely have significant impact on hospitals and hospitalists.7 The Society of Hospital Medicine has a public policy committee that provides feedback to government on programs and policies related to value‐driven health care. The policies and programs need consideration and input from the broader community of hospitalists. This work outlines some of the major national initiatives and policies focused on value‐driven health care and their implications for hospitalists. Hospitalists will need to understand the policy landscape and trends, lead improvement in their individual hospitals to receive value‐based incentives, and assess the opportunities and challenges of current and potential payer programs and policies.

Policies and Initiatives: Implications for Hospitals and Hospitalists

Within the portfolio of value‐driven health care, there are at least 6 major government programs, initiatives, and policies with implications for hospitals and hospitalists: value‐based purchasing (VBP), quality and cost public reporting, Medicare demonstrations, hospital‐acquired conditions, incentives for use of effective HIT, and the physician quality reporting initiative (PQRI) (Table 1).

Future Considerations

The political leadership at the federal and state level is beginning a new transition; however, the focus on quality and value for our health care dollar will likely continue to increase.5, 6 The U.S. health care system has untenable cost estimates, significant quality gaps, and a fractured payment system that fails to reward effective care coordination.2, 21, 22 This increased focus on quality and value should be viewed as an opportunity for hospitalists and hospitals. Hospitalist groups that can achieve high‐quality performance will be increasingly valued, and hospitals should further recognize the critical role hospitalists play in achieving high performance and the associated financial rewards. Hospitalists often lead quality improvement and safety programs in hospitals, and these programs are likely to be seen as progressively more important as payment is linked to performance. The Society of Hospital Medicine engages with policymakers and this role is increasingly significant as more policy and payment decisions impact hospitalists. The Society has focused on collaborative work with payers, policymakers, and other providers to find joint shared solutions. Hospitalists can serve as a link between providers and a focal point of care coordination, especially for the hospitalized patient. Finally, as our system and its incentives continue to progress toward alignment with value‐based high quality care, hospitalists should be leading the change and be an essential part of the solution to transform our health care system to provide high‐quality, efficient care to all Americans.

It is estimated that the Medicare Part A trust fund will be exhausted by 2016 to 2019; also, the quality of care delivered in the United States is highly variable.13 Value is typically defined as the quality achieved for a given cost (ie, value = quality/cost). The focus on the 3 related concepts of value, quality, and cost of health care is likely to continue to increase. Previously, the U.S. Department of Health and Human Services (HHS) made value‐driven health care one of the Department's top priorities.4 Policymakers are in a period of transition but the publicly available plans of the President and Senate leadership indicate that the focus on value‐based initiatives will likely continue to increase as our nation strives to achieve better outcomes for our health care dollar.5, 6 Specifically, the federal government and other payers increasingly align payment incentives with value and quality, encourage public reporting on quality and Medicare payment costs, such as on the Hospital Compare website (http://www.hospitalcompare.hhs.gov), and implement and evaluate demonstrations to test mechanisms such as health information technology (HIT) to improve value‐based performance.

Since hospital care represented $648 billion in 2006, which is 37% of the total patient‐related U.S. health care expenditure, the trend to pay for value will likely have significant impact on hospitals and hospitalists.7 The Society of Hospital Medicine has a public policy committee that provides feedback to government on programs and policies related to value‐driven health care. The policies and programs need consideration and input from the broader community of hospitalists. This work outlines some of the major national initiatives and policies focused on value‐driven health care and their implications for hospitalists. Hospitalists will need to understand the policy landscape and trends, lead improvement in their individual hospitals to receive value‐based incentives, and assess the opportunities and challenges of current and potential payer programs and policies.

Policies and Initiatives: Implications for Hospitals and Hospitalists

Within the portfolio of value‐driven health care, there are at least 6 major government programs, initiatives, and policies with implications for hospitals and hospitalists: value‐based purchasing (VBP), quality and cost public reporting, Medicare demonstrations, hospital‐acquired conditions, incentives for use of effective HIT, and the physician quality reporting initiative (PQRI) (Table 1).

Future Considerations

The political leadership at the federal and state level is beginning a new transition; however, the focus on quality and value for our health care dollar will likely continue to increase.5, 6 The U.S. health care system has untenable cost estimates, significant quality gaps, and a fractured payment system that fails to reward effective care coordination.2, 21, 22 This increased focus on quality and value should be viewed as an opportunity for hospitalists and hospitals. Hospitalist groups that can achieve high‐quality performance will be increasingly valued, and hospitals should further recognize the critical role hospitalists play in achieving high performance and the associated financial rewards. Hospitalists often lead quality improvement and safety programs in hospitals, and these programs are likely to be seen as progressively more important as payment is linked to performance. The Society of Hospital Medicine engages with policymakers and this role is increasingly significant as more policy and payment decisions impact hospitalists. The Society has focused on collaborative work with payers, policymakers, and other providers to find joint shared solutions. Hospitalists can serve as a link between providers and a focal point of care coordination, especially for the hospitalized patient. Finally, as our system and its incentives continue to progress toward alignment with value‐based high quality care, hospitalists should be leading the change and be an essential part of the solution to transform our health care system to provide high‐quality, efficient care to all Americans.

It is estimated that the Medicare Part A trust fund will be exhausted by 2016 to 2019; also, the quality of care delivered in the United States is highly variable.13 Value is typically defined as the quality achieved for a given cost (ie, value = quality/cost). The focus on the 3 related concepts of value, quality, and cost of health care is likely to continue to increase. Previously, the U.S. Department of Health and Human Services (HHS) made value‐driven health care one of the Department's top priorities.4 Policymakers are in a period of transition but the publicly available plans of the President and Senate leadership indicate that the focus on value‐based initiatives will likely continue to increase as our nation strives to achieve better outcomes for our health care dollar.5, 6 Specifically, the federal government and other payers increasingly align payment incentives with value and quality, encourage public reporting on quality and Medicare payment costs, such as on the Hospital Compare website (http://www.hospitalcompare.hhs.gov), and implement and evaluate demonstrations to test mechanisms such as health information technology (HIT) to improve value‐based performance.

Since hospital care represented $648 billion in 2006, which is 37% of the total patient‐related U.S. health care expenditure, the trend to pay for value will likely have significant impact on hospitals and hospitalists.7 The Society of Hospital Medicine has a public policy committee that provides feedback to government on programs and policies related to value‐driven health care. The policies and programs need consideration and input from the broader community of hospitalists. This work outlines some of the major national initiatives and policies focused on value‐driven health care and their implications for hospitalists. Hospitalists will need to understand the policy landscape and trends, lead improvement in their individual hospitals to receive value‐based incentives, and assess the opportunities and challenges of current and potential payer programs and policies.

Policies and Initiatives: Implications for Hospitals and Hospitalists

Within the portfolio of value‐driven health care, there are at least 6 major government programs, initiatives, and policies with implications for hospitals and hospitalists: value‐based purchasing (VBP), quality and cost public reporting, Medicare demonstrations, hospital‐acquired conditions, incentives for use of effective HIT, and the physician quality reporting initiative (PQRI) (Table 1).

Future Considerations

The political leadership at the federal and state level is beginning a new transition; however, the focus on quality and value for our health care dollar will likely continue to increase.5, 6 The U.S. health care system has untenable cost estimates, significant quality gaps, and a fractured payment system that fails to reward effective care coordination.2, 21, 22 This increased focus on quality and value should be viewed as an opportunity for hospitalists and hospitals. Hospitalist groups that can achieve high‐quality performance will be increasingly valued, and hospitals should further recognize the critical role hospitalists play in achieving high performance and the associated financial rewards. Hospitalists often lead quality improvement and safety programs in hospitals, and these programs are likely to be seen as progressively more important as payment is linked to performance. The Society of Hospital Medicine engages with policymakers and this role is increasingly significant as more policy and payment decisions impact hospitalists. The Society has focused on collaborative work with payers, policymakers, and other providers to find joint shared solutions. Hospitalists can serve as a link between providers and a focal point of care coordination, especially for the hospitalized patient. Finally, as our system and its incentives continue to progress toward alignment with value‐based high quality care, hospitalists should be leading the change and be an essential part of the solution to transform our health care system to provide high‐quality, efficient care to all Americans.

By the year 2010, more than 20,000 hospitalists will be in practice, compared to 5000 rheumatologists and 8000 pulmonologists.13 The growth of this career option has been driven by an industry need to reduce healthcare costs, increase the emphasis on quality improvement of healthcare services, and improve the efficiency and delivery of care between that provided in the hospital and that provided by the primary care physician.49

While hospitalists' roots can be traced back to community hospitals in the late 1980s and early 1990s, this career option is now flourishing in academic centers, with the rise of hospitalist faculty and hospitalist faculty tracks.10, 11 One potential advantage of having faculty who are hospitalists is the availability and expertise of physicians who specialize in the care of hospitalized patients.8, 12 Additionally, hospitalist faculty have been reported to achieve high resident satisfaction scores, increase understanding of cost‐effective measures, and improve the supervision of hospital procedures.12 To our knowledge, the medical literature does not provide an estimate of the percentage of internal medicine residency programs utilizing hospitalist faculty.

Critics of hospitalist faculty point to the potential loss of teaching opportunities from shorter hospital stays, bemoan the decreased physician‐patient continuity between inpatient and outpatient arenas, and fear that the hospitalists' presence may decrease resident autonomy and decrease subspecialty consultations by fellows.5, 13, 14 Hospitalist faculty need development, education, and training to match the teaching activities they are expected to fulfill. The challenge is for hospitalist societies and national residency organizations to define, plan for, and meet their faculty development needs.

Goals

The goals of this study were to describe the current involvement of hospitalists in internal medicine residencies. More specifically, we wanted to determine: (1) the percentage of programs with hospitalists as faculty, (2) the teaching activities of hospitalists, (3) regional differences in academic hospitalist activity, and (4) the number of programs with hospitalist training tracks.

Materials and Methods

Results

A total of 272 (response rate 70%) programs completed the 2005 survey, and 236 (response rate 62%) completed the 2007 survey. A total of 171 programs completed both surveys. In 2007, 15 (6%) program directors reported that they were hospitalists while 118 (50%) claimed to be traditional general internists.

For the program directors who answered both surveys, 57% indicated that their primary teaching hospital employed hospitalists before the residency work‐hour limits were implemented (before June 2002). At the time of the survey in 2005, 77% said that hospitalists were employed, a 20% increase in 3 years. When we surveyed these same programs again in 2007, the proportion had risen to 81% (Fisher's exact P = 0.02 compared to before work‐hour limits; Figure 1).

Figure 1

Discussion

This is the first study to document the national rise of hospitalist faculty in internal medicine residency programs. Program directors noted a 20% increase in teaching hospitals that employed hospitalists after the work‐hour regulations went into effecta trend that continued to rise. The tendency was seen first on the coasts, where managed care has higher penetration, and particularly in the Northeast, where New York's resident work‐hour reforms occurred by state mandate prior to the residency accreditation action that affected the rest of the country. Not only have hospitalists picked up the burden of service at these hospitals,19, 20 but the vast majority of programs (>80%) have utilized hospitalists as teachers in important areas of their residency. The magnitude of hospitalist involvement in residency training may have important implications.

Beyond the financial significance of hospitalists at academic teaching hospitals,21 only a few studies have addressed their impact on resident education. On the monthly evaluations at the University of California, San Francisco (San Francisco, CA), residents' satisfaction with their attendings was significantly higher when the physician was a hospitalist rather than a traditional faculty member.22 Residents believed hospitalists were more effective teachers, and provided more effective feedback. At Emory University (Atlanta, GA), a methodologically more rigorous study of postrotation assessment of faculty demonstrated that ratings of hospitalists were not different from traditional general internists; both scored higher than subspecialists.23 The hospitalists as a group had completed training more recently, which also was associated with higher scores.

Even community hospitals that sponsor residency programs have benefited from hospitalist faculty. At Norwalk Hospital (Norwalk, CT), the program had used resident teams led by a group of community physicians and a small group of employed internists. But time pressures and reimbursement concerns created tension between the workload and education balance. After hiring 2 hospitalist clinician‐educators, the length of stay and cost per case were substantially reduced, while resident evaluations indicated improved teaching rounds, conferences, and bedside teaching.24

The results of our study fit with the role of hospitalists as well as what individual programs have reported about hospitalist faculty in the past. Hospitalist faculty serve by and large (92%) as attendings on the hospital ward services. Theoretically, who better to have round with residents in the hospital than the specialists of hospital medicine. For a pulmonary curricular experience, residents work with pulmonologists. But beyond serving as attendings in the hospital, they perform the traditional functions of hospital attendings: providing teaching rounds (>80%), evaluating clinical skills (67%), and even lecturing to residents (>65%). The Southern region trends toward slower adoption of hospitalists as faculty, particularly compared with the Northeast. Overall, what is striking is how much hospitalist faculty already are filling the roles expected of all academic faculty.

We found that only 11% of programs have a hospitalist track through which internal medicine residents may develop the specialized skills and knowledge needed to function optimally in a hospitalist career.25 But given the rapid growth of this specialty, we might expect to see a similar rise in programs providing such specialized training.

Are there risks to having hospitalists teaching residents? One concern is the potential to model fragmented medical care to trainees when hospitals and ambulatory health systems neglect to ensure quality handoffs.26 In an era that heralds the demise of the primary care general internist,27 the impact of hospitalist faculty on general internal medicine nationally, the gravitation of residents toward or away from hospitalist and ambulatory careers, and the role of the traditional general internist in residency training programs in the future remain to be seen. These were not addressed by our data, but are ripe areas for study.

This study has several limitations. It relies on self‐reported data from program directors that, while knowing the intimate details of their educational program, may not have exact knowledge of the number of hospitalists employed by their hospitals. There is also the potential for recall bias by asking the group to remember the number of hospitalists before duty‐hour implementation. Both points in time of our 2 surveys (2005 and 2007) were after the incident growth of hospital medicine as evidenced by the high prevalence of hospitalists in both surveys. Yet, most program directors know that the service needs of the hospitals were acutely increased when the duty‐hours policies went into affect, and probably were fairly involved in hospital decisions to utilize hospitalist physicians to meet these needs. Finally, our study does not address hospitalism within the family medicine and pediatric specialties, both of which have a significant stake in hospital medicine.

In conclusion, our study documents the recent growth and current prevalence of hospitalists' activities in the teaching hospitals of internal medicine residencies in the United States, the duties they perform in resident education, and the magnitude of their penetration in the geographic regions of the country, both in community‐based and university‐based programs. The high degree of involvement of hospitalists in resident education may have important implications for the future of internal medicine as a discipline both with regard to the need for academic faculty development of this important sector of the education community as well as for the education and career development of the residents whom they train.

By the year 2010, more than 20,000 hospitalists will be in practice, compared to 5000 rheumatologists and 8000 pulmonologists.13 The growth of this career option has been driven by an industry need to reduce healthcare costs, increase the emphasis on quality improvement of healthcare services, and improve the efficiency and delivery of care between that provided in the hospital and that provided by the primary care physician.49

While hospitalists' roots can be traced back to community hospitals in the late 1980s and early 1990s, this career option is now flourishing in academic centers, with the rise of hospitalist faculty and hospitalist faculty tracks.10, 11 One potential advantage of having faculty who are hospitalists is the availability and expertise of physicians who specialize in the care of hospitalized patients.8, 12 Additionally, hospitalist faculty have been reported to achieve high resident satisfaction scores, increase understanding of cost‐effective measures, and improve the supervision of hospital procedures.12 To our knowledge, the medical literature does not provide an estimate of the percentage of internal medicine residency programs utilizing hospitalist faculty.

Critics of hospitalist faculty point to the potential loss of teaching opportunities from shorter hospital stays, bemoan the decreased physician‐patient continuity between inpatient and outpatient arenas, and fear that the hospitalists' presence may decrease resident autonomy and decrease subspecialty consultations by fellows.5, 13, 14 Hospitalist faculty need development, education, and training to match the teaching activities they are expected to fulfill. The challenge is for hospitalist societies and national residency organizations to define, plan for, and meet their faculty development needs.

Goals

The goals of this study were to describe the current involvement of hospitalists in internal medicine residencies. More specifically, we wanted to determine: (1) the percentage of programs with hospitalists as faculty, (2) the teaching activities of hospitalists, (3) regional differences in academic hospitalist activity, and (4) the number of programs with hospitalist training tracks.

Materials and Methods

Results

A total of 272 (response rate 70%) programs completed the 2005 survey, and 236 (response rate 62%) completed the 2007 survey. A total of 171 programs completed both surveys. In 2007, 15 (6%) program directors reported that they were hospitalists while 118 (50%) claimed to be traditional general internists.

For the program directors who answered both surveys, 57% indicated that their primary teaching hospital employed hospitalists before the residency work‐hour limits were implemented (before June 2002). At the time of the survey in 2005, 77% said that hospitalists were employed, a 20% increase in 3 years. When we surveyed these same programs again in 2007, the proportion had risen to 81% (Fisher's exact P = 0.02 compared to before work‐hour limits; Figure 1).

Figure 1

Discussion

This is the first study to document the national rise of hospitalist faculty in internal medicine residency programs. Program directors noted a 20% increase in teaching hospitals that employed hospitalists after the work‐hour regulations went into effecta trend that continued to rise. The tendency was seen first on the coasts, where managed care has higher penetration, and particularly in the Northeast, where New York's resident work‐hour reforms occurred by state mandate prior to the residency accreditation action that affected the rest of the country. Not only have hospitalists picked up the burden of service at these hospitals,19, 20 but the vast majority of programs (>80%) have utilized hospitalists as teachers in important areas of their residency. The magnitude of hospitalist involvement in residency training may have important implications.

Beyond the financial significance of hospitalists at academic teaching hospitals,21 only a few studies have addressed their impact on resident education. On the monthly evaluations at the University of California, San Francisco (San Francisco, CA), residents' satisfaction with their attendings was significantly higher when the physician was a hospitalist rather than a traditional faculty member.22 Residents believed hospitalists were more effective teachers, and provided more effective feedback. At Emory University (Atlanta, GA), a methodologically more rigorous study of postrotation assessment of faculty demonstrated that ratings of hospitalists were not different from traditional general internists; both scored higher than subspecialists.23 The hospitalists as a group had completed training more recently, which also was associated with higher scores.

Even community hospitals that sponsor residency programs have benefited from hospitalist faculty. At Norwalk Hospital (Norwalk, CT), the program had used resident teams led by a group of community physicians and a small group of employed internists. But time pressures and reimbursement concerns created tension between the workload and education balance. After hiring 2 hospitalist clinician‐educators, the length of stay and cost per case were substantially reduced, while resident evaluations indicated improved teaching rounds, conferences, and bedside teaching.24

The results of our study fit with the role of hospitalists as well as what individual programs have reported about hospitalist faculty in the past. Hospitalist faculty serve by and large (92%) as attendings on the hospital ward services. Theoretically, who better to have round with residents in the hospital than the specialists of hospital medicine. For a pulmonary curricular experience, residents work with pulmonologists. But beyond serving as attendings in the hospital, they perform the traditional functions of hospital attendings: providing teaching rounds (>80%), evaluating clinical skills (67%), and even lecturing to residents (>65%). The Southern region trends toward slower adoption of hospitalists as faculty, particularly compared with the Northeast. Overall, what is striking is how much hospitalist faculty already are filling the roles expected of all academic faculty.

We found that only 11% of programs have a hospitalist track through which internal medicine residents may develop the specialized skills and knowledge needed to function optimally in a hospitalist career.25 But given the rapid growth of this specialty, we might expect to see a similar rise in programs providing such specialized training.

Are there risks to having hospitalists teaching residents? One concern is the potential to model fragmented medical care to trainees when hospitals and ambulatory health systems neglect to ensure quality handoffs.26 In an era that heralds the demise of the primary care general internist,27 the impact of hospitalist faculty on general internal medicine nationally, the gravitation of residents toward or away from hospitalist and ambulatory careers, and the role of the traditional general internist in residency training programs in the future remain to be seen. These were not addressed by our data, but are ripe areas for study.

This study has several limitations. It relies on self‐reported data from program directors that, while knowing the intimate details of their educational program, may not have exact knowledge of the number of hospitalists employed by their hospitals. There is also the potential for recall bias by asking the group to remember the number of hospitalists before duty‐hour implementation. Both points in time of our 2 surveys (2005 and 2007) were after the incident growth of hospital medicine as evidenced by the high prevalence of hospitalists in both surveys. Yet, most program directors know that the service needs of the hospitals were acutely increased when the duty‐hours policies went into affect, and probably were fairly involved in hospital decisions to utilize hospitalist physicians to meet these needs. Finally, our study does not address hospitalism within the family medicine and pediatric specialties, both of which have a significant stake in hospital medicine.

In conclusion, our study documents the recent growth and current prevalence of hospitalists' activities in the teaching hospitals of internal medicine residencies in the United States, the duties they perform in resident education, and the magnitude of their penetration in the geographic regions of the country, both in community‐based and university‐based programs. The high degree of involvement of hospitalists in resident education may have important implications for the future of internal medicine as a discipline both with regard to the need for academic faculty development of this important sector of the education community as well as for the education and career development of the residents whom they train.

By the year 2010, more than 20,000 hospitalists will be in practice, compared to 5000 rheumatologists and 8000 pulmonologists.13 The growth of this career option has been driven by an industry need to reduce healthcare costs, increase the emphasis on quality improvement of healthcare services, and improve the efficiency and delivery of care between that provided in the hospital and that provided by the primary care physician.49

While hospitalists' roots can be traced back to community hospitals in the late 1980s and early 1990s, this career option is now flourishing in academic centers, with the rise of hospitalist faculty and hospitalist faculty tracks.10, 11 One potential advantage of having faculty who are hospitalists is the availability and expertise of physicians who specialize in the care of hospitalized patients.8, 12 Additionally, hospitalist faculty have been reported to achieve high resident satisfaction scores, increase understanding of cost‐effective measures, and improve the supervision of hospital procedures.12 To our knowledge, the medical literature does not provide an estimate of the percentage of internal medicine residency programs utilizing hospitalist faculty.

Critics of hospitalist faculty point to the potential loss of teaching opportunities from shorter hospital stays, bemoan the decreased physician‐patient continuity between inpatient and outpatient arenas, and fear that the hospitalists' presence may decrease resident autonomy and decrease subspecialty consultations by fellows.5, 13, 14 Hospitalist faculty need development, education, and training to match the teaching activities they are expected to fulfill. The challenge is for hospitalist societies and national residency organizations to define, plan for, and meet their faculty development needs.

Goals

The goals of this study were to describe the current involvement of hospitalists in internal medicine residencies. More specifically, we wanted to determine: (1) the percentage of programs with hospitalists as faculty, (2) the teaching activities of hospitalists, (3) regional differences in academic hospitalist activity, and (4) the number of programs with hospitalist training tracks.

Materials and Methods

Results

A total of 272 (response rate 70%) programs completed the 2005 survey, and 236 (response rate 62%) completed the 2007 survey. A total of 171 programs completed both surveys. In 2007, 15 (6%) program directors reported that they were hospitalists while 118 (50%) claimed to be traditional general internists.

For the program directors who answered both surveys, 57% indicated that their primary teaching hospital employed hospitalists before the residency work‐hour limits were implemented (before June 2002). At the time of the survey in 2005, 77% said that hospitalists were employed, a 20% increase in 3 years. When we surveyed these same programs again in 2007, the proportion had risen to 81% (Fisher's exact P = 0.02 compared to before work‐hour limits; Figure 1).

Figure 1

Discussion

This is the first study to document the national rise of hospitalist faculty in internal medicine residency programs. Program directors noted a 20% increase in teaching hospitals that employed hospitalists after the work‐hour regulations went into effecta trend that continued to rise. The tendency was seen first on the coasts, where managed care has higher penetration, and particularly in the Northeast, where New York's resident work‐hour reforms occurred by state mandate prior to the residency accreditation action that affected the rest of the country. Not only have hospitalists picked up the burden of service at these hospitals,19, 20 but the vast majority of programs (>80%) have utilized hospitalists as teachers in important areas of their residency. The magnitude of hospitalist involvement in residency training may have important implications.

Beyond the financial significance of hospitalists at academic teaching hospitals,21 only a few studies have addressed their impact on resident education. On the monthly evaluations at the University of California, San Francisco (San Francisco, CA), residents' satisfaction with their attendings was significantly higher when the physician was a hospitalist rather than a traditional faculty member.22 Residents believed hospitalists were more effective teachers, and provided more effective feedback. At Emory University (Atlanta, GA), a methodologically more rigorous study of postrotation assessment of faculty demonstrated that ratings of hospitalists were not different from traditional general internists; both scored higher than subspecialists.23 The hospitalists as a group had completed training more recently, which also was associated with higher scores.

Even community hospitals that sponsor residency programs have benefited from hospitalist faculty. At Norwalk Hospital (Norwalk, CT), the program had used resident teams led by a group of community physicians and a small group of employed internists. But time pressures and reimbursement concerns created tension between the workload and education balance. After hiring 2 hospitalist clinician‐educators, the length of stay and cost per case were substantially reduced, while resident evaluations indicated improved teaching rounds, conferences, and bedside teaching.24

The results of our study fit with the role of hospitalists as well as what individual programs have reported about hospitalist faculty in the past. Hospitalist faculty serve by and large (92%) as attendings on the hospital ward services. Theoretically, who better to have round with residents in the hospital than the specialists of hospital medicine. For a pulmonary curricular experience, residents work with pulmonologists. But beyond serving as attendings in the hospital, they perform the traditional functions of hospital attendings: providing teaching rounds (>80%), evaluating clinical skills (67%), and even lecturing to residents (>65%). The Southern region trends toward slower adoption of hospitalists as faculty, particularly compared with the Northeast. Overall, what is striking is how much hospitalist faculty already are filling the roles expected of all academic faculty.

We found that only 11% of programs have a hospitalist track through which internal medicine residents may develop the specialized skills and knowledge needed to function optimally in a hospitalist career.25 But given the rapid growth of this specialty, we might expect to see a similar rise in programs providing such specialized training.

Are there risks to having hospitalists teaching residents? One concern is the potential to model fragmented medical care to trainees when hospitals and ambulatory health systems neglect to ensure quality handoffs.26 In an era that heralds the demise of the primary care general internist,27 the impact of hospitalist faculty on general internal medicine nationally, the gravitation of residents toward or away from hospitalist and ambulatory careers, and the role of the traditional general internist in residency training programs in the future remain to be seen. These were not addressed by our data, but are ripe areas for study.

This study has several limitations. It relies on self‐reported data from program directors that, while knowing the intimate details of their educational program, may not have exact knowledge of the number of hospitalists employed by their hospitals. There is also the potential for recall bias by asking the group to remember the number of hospitalists before duty‐hour implementation. Both points in time of our 2 surveys (2005 and 2007) were after the incident growth of hospital medicine as evidenced by the high prevalence of hospitalists in both surveys. Yet, most program directors know that the service needs of the hospitals were acutely increased when the duty‐hours policies went into affect, and probably were fairly involved in hospital decisions to utilize hospitalist physicians to meet these needs. Finally, our study does not address hospitalism within the family medicine and pediatric specialties, both of which have a significant stake in hospital medicine.

In conclusion, our study documents the recent growth and current prevalence of hospitalists' activities in the teaching hospitals of internal medicine residencies in the United States, the duties they perform in resident education, and the magnitude of their penetration in the geographic regions of the country, both in community‐based and university‐based programs. The high degree of involvement of hospitalists in resident education may have important implications for the future of internal medicine as a discipline both with regard to the need for academic faculty development of this important sector of the education community as well as for the education and career development of the residents whom they train.

Hospitalists are increasingly assuming a primary role in medical education in the hospital setting, as they also steadily care for a larger portion of hospitalized patients.1 This issue of the Journal of Hospital Medicine highlights the role of hospitalists as teachers in academic medical centers, confirming their expanding and positive role in resident and medical student education. A survey of academic medical centers, a systematic review, an evaluation of the implementation of an educational curriculum, and a survey of residents hint at the challenges hospitalists face in teaching, but also expose us to a more advanced yet facile approach to evaluating the effectiveness of a teaching intervention.25 These publications provoke interesting questions about clinical teaching that hospital medicine educators and researchers should pursue answering. I believe they will also encourage us to innovate in medical education and assessment of that teaching.

Traditionally, teaching attendings for resident teams on medicine or pediatric services rotated through these duties for 1 to 3 months each year, while spending the majority of their time in clinic or research activities. The increasing complexity of hospitalized patients and the pressure to reduce length of stay prompted closer oversight of trainees. With the advent of resident work‐hour restrictions, the need for greater clinical involvement by attending physicians made it increasingly difficult to maintain the traditional model of limited engagement by faculty attendings. Simply put, the dwindling pool of willing and able teaching attendings encouraged teaching hospitals to employ hospitalists to fill the gap in teaching and supervision, as well as clinical coverage.6

Beasley et al.2 report that resident work‐hour restrictions were associated with an increase in the number of teaching hospitals employing hospitalists to 79% of 193 surveyed hospitals in 2007. Of those hospitals with hospitalists, 92% reported that hospitalists serve as attendings on the teaching service. Hospitalists also teach in a number of other venues within these programs, including formal teaching rounds without direct care responsibility, along with delivering didactic lectures and clinical skills education.

How well are teaching hospitalists performing compared to traditional teaching attendings? Natarajan et al.4 provide an important summary of the evidence in a systematic review of studies comparing teaching efforts of hospitalist attendings to those of nonhospitalist attendings. Eight studies from a variety of institutions measured trainee (resident or medical student) attitudes. It is gratifying to learn that hospitalists were generally rated higher at overall teaching effectiveness, provision of feedback, knowledge base, and involvement of the learner in patient care. It seems likely that publication bias would overestimate the positive effect of hospitalists on learner attitudes. However, there are plausible reasons that the positive effect is accurate. Because their professional responsibilities are focused in the hospital, hospitalists should naturally be more available to learners for teaching and feedback. Hospitalists tend to be younger in their academic careers, placing them closer to the cutting edge of knowledge gained during residency and possibly fellowship. They may be more in tune with the needs and pressures faced by their learners, having dealt with these same challenges either during recent training or during nonteaching rotations.

As a relatively young specialty with young and developing academic hospitalists, will the advantage suggested by the Natarajan et al.4 systematic review be sustained over the long term as careers in hospital medicine mature? A 2005 systematic review studying this question among practicing clinicians found, somewhat paradoxically, that older, more experienced clinicians appeared to be at risk for providing lower‐quality care.7 To avoid this decline in clinical effectiveness, hospitalists should proactively seek innovative ways to refresh and update their knowledge and skills throughout their careers. This is particularly critical for teaching physicians. We should seize the opportunity to study the relationship between advancing clinical/teaching experience and educational quality within our teaching programs.

The review by Natarajan et al.4 should also challenge the hospitalist community to achieve even higher levels of proficiency as teachers of medicine. The review alludes to bedside teaching and attention to psychosocial aspects of care as opportunities for improvement by hospitalist teachers. A recent study suggested that physical examination instruction receives declining attention from inpatient teachers and that there are opportunities to increase the amount of bedside teaching.8 A provocative study of inpatients admitted to a teaching service found that physical examination could substantially impact patient care, but that trainees often failed to appreciate significant findings on initial examination.9 How do teaching hospitalists become proficient at physical examination and bedside teaching? Are there models around the country that are successfully developing outstanding clinician educators, incorporating teach‐the‐teacher models to improve physical examination and bedside teaching?

A practical limitation of attitude surveys and learner evaluation is the well‐known phenomenon of grade inflation that resulted in high ratings for all attending groups in the studies summarized by Natarajan et al.4 This limits the ability of surveys or evaluations to distinguish truly outstanding teachers and consequently makes it difficult to analyze the attributes of these teachers. We need better tools to detect and learn teaching techniques from great teachers in the clinical environment. We need studies evaluating the effect of teaching hospitalists on learner knowledge or, even more importantly, learner outcomes. Ultimately, we need studies of educational interventions that evaluate the impact of these interventions on patient outcome.

Wright et al.5 provide guidance as they describe the evaluation of a teaching intervention that moves beyond measurement of knowledge or attitudes. The Johns Hopkins Bayview hospitalist group sought to improve the quality of medical consultations performed by hospitalists and by residents rotating on the consultation service using a case‐based teaching module with audits of recent notes. The participants then audited their most recent consultation notes with feedback from the module teacher. The study employed pretests and posttests of knowledgea standard evaluation for educational interventions. This tells us little about the true impact of the teaching module. However, the study then assessed the quality of written consultations done by hospitalists before and after the educational interventions. Scores of consult notes improved significantly after the intervention, although the number of assessments for each physician was limited. Importantly, we need to know if interventions such as this are sustained over time. Wright's well‐established medical education research group's study design assessed the impact of an intervention on physician performance and moves us closer to assessment of the impact on actual patient outcomes. As clinical teachers, we would like to believe that our teaching and our educational innovations are having a positive impact on patient care. Can we demonstrate this?

As academic medical centers contend with further resident work‐hour restrictions proposed by the Institute of Medicine (IOM),10 how will this affect hospitalist teachers? The study by Mazotti et al.3 from the University of California at San Francisco residency program found that about one‐quarter of residents reported spending less time teaching after implementation of the Accreditation Council for Graduate Medical Education (ACGME) duty‐hour restrictions in 2003. Interestingly, those residents reporting less time spent teaching also reported less emotional exhaustion and perceived that they were delivering higher‐quality patient care. This raises a fascinating question for academic hospitalists. Would these findings be similar among teaching hospitalists and nonteaching hospitalists? What about hospitalists who rotate through months of teaching and nonteaching services? Is teaching emotionally exhausting for experienced teachers? A Mayo Clinic study suggested that the extent that faculty physicians are able to engage in work that is most meaningful to them as individuals is a strong determinant of faculty burnout.11 Is the hospitalist who finds teaching most rewarding at risk of burnout if they are assigned only 2 weeks a year as a teaching attending? The answers to these questions will be critical to hospitalist program leaders trying to assure sustainable careers for hospitalists in their programs.

Although the study by Mazotti et al.3 did not assess the impact of the reduction in resident teaching time on the teaching responsibilities for academic hospitalists, previous studies suggest that faculty are also teaching less since the introduction of work‐hour restrictions.12, 13 If the new IOM recommendations are enacted, who will teach? Although the reported experience following the 2003 work‐hour restrictions begs pessimism, the anticipated changes represent an opportunity for creative hospitalist teachers to demonstrate effective adaptations to the changing and compressed inpatient teaching environment.

In summary, this issue of the Journal presents studies that praise the role hospitalists play in teaching the next generation of physicians, but also gives a glimpse of future challenges and opportunities. We should take advantage of hospitalists' central position in clinical education in the hospital to innovate, study the effect on both learner outcomes and patient outcomes, and share our experiences with the hospitalist and medical education communities.

Hospitalists are increasingly assuming a primary role in medical education in the hospital setting, as they also steadily care for a larger portion of hospitalized patients.1 This issue of the Journal of Hospital Medicine highlights the role of hospitalists as teachers in academic medical centers, confirming their expanding and positive role in resident and medical student education. A survey of academic medical centers, a systematic review, an evaluation of the implementation of an educational curriculum, and a survey of residents hint at the challenges hospitalists face in teaching, but also expose us to a more advanced yet facile approach to evaluating the effectiveness of a teaching intervention.25 These publications provoke interesting questions about clinical teaching that hospital medicine educators and researchers should pursue answering. I believe they will also encourage us to innovate in medical education and assessment of that teaching.

Traditionally, teaching attendings for resident teams on medicine or pediatric services rotated through these duties for 1 to 3 months each year, while spending the majority of their time in clinic or research activities. The increasing complexity of hospitalized patients and the pressure to reduce length of stay prompted closer oversight of trainees. With the advent of resident work‐hour restrictions, the need for greater clinical involvement by attending physicians made it increasingly difficult to maintain the traditional model of limited engagement by faculty attendings. Simply put, the dwindling pool of willing and able teaching attendings encouraged teaching hospitals to employ hospitalists to fill the gap in teaching and supervision, as well as clinical coverage.6

Beasley et al.2 report that resident work‐hour restrictions were associated with an increase in the number of teaching hospitals employing hospitalists to 79% of 193 surveyed hospitals in 2007. Of those hospitals with hospitalists, 92% reported that hospitalists serve as attendings on the teaching service. Hospitalists also teach in a number of other venues within these programs, including formal teaching rounds without direct care responsibility, along with delivering didactic lectures and clinical skills education.

How well are teaching hospitalists performing compared to traditional teaching attendings? Natarajan et al.4 provide an important summary of the evidence in a systematic review of studies comparing teaching efforts of hospitalist attendings to those of nonhospitalist attendings. Eight studies from a variety of institutions measured trainee (resident or medical student) attitudes. It is gratifying to learn that hospitalists were generally rated higher at overall teaching effectiveness, provision of feedback, knowledge base, and involvement of the learner in patient care. It seems likely that publication bias would overestimate the positive effect of hospitalists on learner attitudes. However, there are plausible reasons that the positive effect is accurate. Because their professional responsibilities are focused in the hospital, hospitalists should naturally be more available to learners for teaching and feedback. Hospitalists tend to be younger in their academic careers, placing them closer to the cutting edge of knowledge gained during residency and possibly fellowship. They may be more in tune with the needs and pressures faced by their learners, having dealt with these same challenges either during recent training or during nonteaching rotations.

As a relatively young specialty with young and developing academic hospitalists, will the advantage suggested by the Natarajan et al.4 systematic review be sustained over the long term as careers in hospital medicine mature? A 2005 systematic review studying this question among practicing clinicians found, somewhat paradoxically, that older, more experienced clinicians appeared to be at risk for providing lower‐quality care.7 To avoid this decline in clinical effectiveness, hospitalists should proactively seek innovative ways to refresh and update their knowledge and skills throughout their careers. This is particularly critical for teaching physicians. We should seize the opportunity to study the relationship between advancing clinical/teaching experience and educational quality within our teaching programs.

The review by Natarajan et al.4 should also challenge the hospitalist community to achieve even higher levels of proficiency as teachers of medicine. The review alludes to bedside teaching and attention to psychosocial aspects of care as opportunities for improvement by hospitalist teachers. A recent study suggested that physical examination instruction receives declining attention from inpatient teachers and that there are opportunities to increase the amount of bedside teaching.8 A provocative study of inpatients admitted to a teaching service found that physical examination could substantially impact patient care, but that trainees often failed to appreciate significant findings on initial examination.9 How do teaching hospitalists become proficient at physical examination and bedside teaching? Are there models around the country that are successfully developing outstanding clinician educators, incorporating teach‐the‐teacher models to improve physical examination and bedside teaching?

A practical limitation of attitude surveys and learner evaluation is the well‐known phenomenon of grade inflation that resulted in high ratings for all attending groups in the studies summarized by Natarajan et al.4 This limits the ability of surveys or evaluations to distinguish truly outstanding teachers and consequently makes it difficult to analyze the attributes of these teachers. We need better tools to detect and learn teaching techniques from great teachers in the clinical environment. We need studies evaluating the effect of teaching hospitalists on learner knowledge or, even more importantly, learner outcomes. Ultimately, we need studies of educational interventions that evaluate the impact of these interventions on patient outcome.

Wright et al.5 provide guidance as they describe the evaluation of a teaching intervention that moves beyond measurement of knowledge or attitudes. The Johns Hopkins Bayview hospitalist group sought to improve the quality of medical consultations performed by hospitalists and by residents rotating on the consultation service using a case‐based teaching module with audits of recent notes. The participants then audited their most recent consultation notes with feedback from the module teacher. The study employed pretests and posttests of knowledgea standard evaluation for educational interventions. This tells us little about the true impact of the teaching module. However, the study then assessed the quality of written consultations done by hospitalists before and after the educational interventions. Scores of consult notes improved significantly after the intervention, although the number of assessments for each physician was limited. Importantly, we need to know if interventions such as this are sustained over time. Wright's well‐established medical education research group's study design assessed the impact of an intervention on physician performance and moves us closer to assessment of the impact on actual patient outcomes. As clinical teachers, we would like to believe that our teaching and our educational innovations are having a positive impact on patient care. Can we demonstrate this?

As academic medical centers contend with further resident work‐hour restrictions proposed by the Institute of Medicine (IOM),10 how will this affect hospitalist teachers? The study by Mazotti et al.3 from the University of California at San Francisco residency program found that about one‐quarter of residents reported spending less time teaching after implementation of the Accreditation Council for Graduate Medical Education (ACGME) duty‐hour restrictions in 2003. Interestingly, those residents reporting less time spent teaching also reported less emotional exhaustion and perceived that they were delivering higher‐quality patient care. This raises a fascinating question for academic hospitalists. Would these findings be similar among teaching hospitalists and nonteaching hospitalists? What about hospitalists who rotate through months of teaching and nonteaching services? Is teaching emotionally exhausting for experienced teachers? A Mayo Clinic study suggested that the extent that faculty physicians are able to engage in work that is most meaningful to them as individuals is a strong determinant of faculty burnout.11 Is the hospitalist who finds teaching most rewarding at risk of burnout if they are assigned only 2 weeks a year as a teaching attending? The answers to these questions will be critical to hospitalist program leaders trying to assure sustainable careers for hospitalists in their programs.

Although the study by Mazotti et al.3 did not assess the impact of the reduction in resident teaching time on the teaching responsibilities for academic hospitalists, previous studies suggest that faculty are also teaching less since the introduction of work‐hour restrictions.12, 13 If the new IOM recommendations are enacted, who will teach? Although the reported experience following the 2003 work‐hour restrictions begs pessimism, the anticipated changes represent an opportunity for creative hospitalist teachers to demonstrate effective adaptations to the changing and compressed inpatient teaching environment.

In summary, this issue of the Journal presents studies that praise the role hospitalists play in teaching the next generation of physicians, but also gives a glimpse of future challenges and opportunities. We should take advantage of hospitalists' central position in clinical education in the hospital to innovate, study the effect on both learner outcomes and patient outcomes, and share our experiences with the hospitalist and medical education communities.

Hospitalists are increasingly assuming a primary role in medical education in the hospital setting, as they also steadily care for a larger portion of hospitalized patients.1 This issue of the Journal of Hospital Medicine highlights the role of hospitalists as teachers in academic medical centers, confirming their expanding and positive role in resident and medical student education. A survey of academic medical centers, a systematic review, an evaluation of the implementation of an educational curriculum, and a survey of residents hint at the challenges hospitalists face in teaching, but also expose us to a more advanced yet facile approach to evaluating the effectiveness of a teaching intervention.25 These publications provoke interesting questions about clinical teaching that hospital medicine educators and researchers should pursue answering. I believe they will also encourage us to innovate in medical education and assessment of that teaching.

Traditionally, teaching attendings for resident teams on medicine or pediatric services rotated through these duties for 1 to 3 months each year, while spending the majority of their time in clinic or research activities. The increasing complexity of hospitalized patients and the pressure to reduce length of stay prompted closer oversight of trainees. With the advent of resident work‐hour restrictions, the need for greater clinical involvement by attending physicians made it increasingly difficult to maintain the traditional model of limited engagement by faculty attendings. Simply put, the dwindling pool of willing and able teaching attendings encouraged teaching hospitals to employ hospitalists to fill the gap in teaching and supervision, as well as clinical coverage.6

Beasley et al.2 report that resident work‐hour restrictions were associated with an increase in the number of teaching hospitals employing hospitalists to 79% of 193 surveyed hospitals in 2007. Of those hospitals with hospitalists, 92% reported that hospitalists serve as attendings on the teaching service. Hospitalists also teach in a number of other venues within these programs, including formal teaching rounds without direct care responsibility, along with delivering didactic lectures and clinical skills education.

How well are teaching hospitalists performing compared to traditional teaching attendings? Natarajan et al.4 provide an important summary of the evidence in a systematic review of studies comparing teaching efforts of hospitalist attendings to those of nonhospitalist attendings. Eight studies from a variety of institutions measured trainee (resident or medical student) attitudes. It is gratifying to learn that hospitalists were generally rated higher at overall teaching effectiveness, provision of feedback, knowledge base, and involvement of the learner in patient care. It seems likely that publication bias would overestimate the positive effect of hospitalists on learner attitudes. However, there are plausible reasons that the positive effect is accurate. Because their professional responsibilities are focused in the hospital, hospitalists should naturally be more available to learners for teaching and feedback. Hospitalists tend to be younger in their academic careers, placing them closer to the cutting edge of knowledge gained during residency and possibly fellowship. They may be more in tune with the needs and pressures faced by their learners, having dealt with these same challenges either during recent training or during nonteaching rotations.

As a relatively young specialty with young and developing academic hospitalists, will the advantage suggested by the Natarajan et al.4 systematic review be sustained over the long term as careers in hospital medicine mature? A 2005 systematic review studying this question among practicing clinicians found, somewhat paradoxically, that older, more experienced clinicians appeared to be at risk for providing lower‐quality care.7 To avoid this decline in clinical effectiveness, hospitalists should proactively seek innovative ways to refresh and update their knowledge and skills throughout their careers. This is particularly critical for teaching physicians. We should seize the opportunity to study the relationship between advancing clinical/teaching experience and educational quality within our teaching programs.

The review by Natarajan et al.4 should also challenge the hospitalist community to achieve even higher levels of proficiency as teachers of medicine. The review alludes to bedside teaching and attention to psychosocial aspects of care as opportunities for improvement by hospitalist teachers. A recent study suggested that physical examination instruction receives declining attention from inpatient teachers and that there are opportunities to increase the amount of bedside teaching.8 A provocative study of inpatients admitted to a teaching service found that physical examination could substantially impact patient care, but that trainees often failed to appreciate significant findings on initial examination.9 How do teaching hospitalists become proficient at physical examination and bedside teaching? Are there models around the country that are successfully developing outstanding clinician educators, incorporating teach‐the‐teacher models to improve physical examination and bedside teaching?

A practical limitation of attitude surveys and learner evaluation is the well‐known phenomenon of grade inflation that resulted in high ratings for all attending groups in the studies summarized by Natarajan et al.4 This limits the ability of surveys or evaluations to distinguish truly outstanding teachers and consequently makes it difficult to analyze the attributes of these teachers. We need better tools to detect and learn teaching techniques from great teachers in the clinical environment. We need studies evaluating the effect of teaching hospitalists on learner knowledge or, even more importantly, learner outcomes. Ultimately, we need studies of educational interventions that evaluate the impact of these interventions on patient outcome.

Wright et al.5 provide guidance as they describe the evaluation of a teaching intervention that moves beyond measurement of knowledge or attitudes. The Johns Hopkins Bayview hospitalist group sought to improve the quality of medical consultations performed by hospitalists and by residents rotating on the consultation service using a case‐based teaching module with audits of recent notes. The participants then audited their most recent consultation notes with feedback from the module teacher. The study employed pretests and posttests of knowledgea standard evaluation for educational interventions. This tells us little about the true impact of the teaching module. However, the study then assessed the quality of written consultations done by hospitalists before and after the educational interventions. Scores of consult notes improved significantly after the intervention, although the number of assessments for each physician was limited. Importantly, we need to know if interventions such as this are sustained over time. Wright's well‐established medical education research group's study design assessed the impact of an intervention on physician performance and moves us closer to assessment of the impact on actual patient outcomes. As clinical teachers, we would like to believe that our teaching and our educational innovations are having a positive impact on patient care. Can we demonstrate this?

As academic medical centers contend with further resident work‐hour restrictions proposed by the Institute of Medicine (IOM),10 how will this affect hospitalist teachers? The study by Mazotti et al.3 from the University of California at San Francisco residency program found that about one‐quarter of residents reported spending less time teaching after implementation of the Accreditation Council for Graduate Medical Education (ACGME) duty‐hour restrictions in 2003. Interestingly, those residents reporting less time spent teaching also reported less emotional exhaustion and perceived that they were delivering higher‐quality patient care. This raises a fascinating question for academic hospitalists. Would these findings be similar among teaching hospitalists and nonteaching hospitalists? What about hospitalists who rotate through months of teaching and nonteaching services? Is teaching emotionally exhausting for experienced teachers? A Mayo Clinic study suggested that the extent that faculty physicians are able to engage in work that is most meaningful to them as individuals is a strong determinant of faculty burnout.11 Is the hospitalist who finds teaching most rewarding at risk of burnout if they are assigned only 2 weeks a year as a teaching attending? The answers to these questions will be critical to hospitalist program leaders trying to assure sustainable careers for hospitalists in their programs.

Although the study by Mazotti et al.3 did not assess the impact of the reduction in resident teaching time on the teaching responsibilities for academic hospitalists, previous studies suggest that faculty are also teaching less since the introduction of work‐hour restrictions.12, 13 If the new IOM recommendations are enacted, who will teach? Although the reported experience following the 2003 work‐hour restrictions begs pessimism, the anticipated changes represent an opportunity for creative hospitalist teachers to demonstrate effective adaptations to the changing and compressed inpatient teaching environment.

In summary, this issue of the Journal presents studies that praise the role hospitalists play in teaching the next generation of physicians, but also gives a glimpse of future challenges and opportunities. We should take advantage of hospitalists' central position in clinical education in the hospital to innovate, study the effect on both learner outcomes and patient outcomes, and share our experiences with the hospitalist and medical education communities.

A 58‐year old woman was brought to the emergency department with confusion. Her husband stated that for several hours she had been drifting in and out at home, and that he had to shout to get her attention. He described no seizure activity, weakness, incontinence, or difficulty speaking, and had noted no complaints of headache, fevers, chest pain, shortness of breath, or gastrointestinal complaints.

Altered mental status in a middle‐aged woman can result from a diverse set of etiologies. A key distinction in the neurological examination will be to assure that the complaint of confusion is accurate as opposed to aphasia; the former is usually indicative of diffuse cerebral dysfunction while the latter suggests a focal lesion in the dominant hemisphere.

The acuity of the change in mental status is important, as are the fluctuations described by the husband. Unwitnessed or nonconvulsive seizure activity can present this way. Toxic/metabolic etiologies, infectious and inflammatory disorders of the central nervous system (CNS), and vascular diseases are also important considerations. Although stroke does not typically present with global encephalopathy, intermittent large vessel occlusion, especially in the posterior circulation, can disrupt cognition in this manner. Following a physical examination, initial workup should focus on toxic/metabolic etiologies, followed rapidly by head imaging if no cause is identified.

Her past medical history was notable for type 2 diabetes mellitus, coronary artery disease, hyperlipidemia, and an unspecified seizure disorder, which according to her husband was diagnosed during a recent hospitalization for a similar presentation. She also had a remote history of venous thromboembolism and antithrombin‐III deficiency. She was unemployed, lived with her husband, and spent most of her time at home. She never smoked, and rarely drank alcohol. Her family history was unobtainable, and her husband denied that she used any illicit drugs. Her medications included pioglitazone, aspirin, simvastatin, pregabalin, ferrous sulfate, levetiracetam, warfarin, and magnesium oxide, and she was allergic to sulfa.

While the differential diagnosis remains broad, 3 elements of the history are potentially relevant. The history of epilepsy based on a similar prior presentation increases the likelihood that the current spell is ictal in nature; examination of previous records would be important in order to document whether these spells have indeed been proven to be epileptic, as many conditions can mimic seizures. Given the history of venous thromboembolism and hypercoagulability, one must consider cerebral venous sinus thrombosis, which can present with global neurologic dysfunction and seizures. Prompt identification, usually via computed tomography (CT) or magnetic resonance angiography, is vital, because anticoagulation can mitigate this potentially life‐threatening illness. Finally, although many medications can cause encephalopathy in overdose, levetiracetam has well‐described cognitive side effects even at usual doses, including encephalopathy, irritability, and depression.

The records from that recent hospitalization remarked that she had presented confused and stuporous. Her potassium had been 2.7 mmol/L, international normalized ration (INR) 3.4, and hemoglobin 8 g/dL; other routine laboratory studies were normal. CT and magnetic resonance imaging (MRI) of the brain had been negative, and electroencephalogram (EEG) reportedly was performed but specific results were unknown. She was discharged alert and oriented 1 week prior to the current presentation on the above medications, including levetiracetam for this newly‐diagnosed seizure disorder.

Previous records confirm that the current presentation is that of a relapsing acute alteration in mental status. Regardless of the EEG findings or response to antiepileptic medications, a seizure disorder should remain a primary consideration, although relapsing inflammatory, toxic/metabolic conditions, and, rarely, vascular disorders can also present in this manner.

The neurologic manifestations of hypokalemia are usually peripheral in nature, including periodic paralysis; confusion accompanying hypokalemia is usually not a result of the low potassium itself but rather due to an underlying toxic or endocrinologic cause. Various causes of anemia can lead to mental status changes; the mean corpuscular volume (MCV) will be particularly helpful given known associations between megaloblastic anemia and confusional states.

On examination, she appeared to be in good health and in no distress. She was afebrile. Her blood pressure was 93/57, pulse 90 beats per minute, respiratory rate 16 per minute, and room air oxygen saturation 100%. She was oriented to her surroundings, but slow in her responses to questioning. There were no cranial nerve, motor, or sensory deficits, or abnormal reflexes or movements. Examination of the head, skin, chest, cardiovascular system, abdomen, and extremities was normal. Serum sodium was 136 mmol/L, creatinine 1.2 mg/dL, calcium 9.3 mg/dL, and glucose 81 mg/dL; other routine blood chemistries were normal. Her white blood cell (WBC) count was 7100/L, hemoglobin 9.2 g/dL with normal MCV, and platelet count 275,000/L. INR was 3.4, and liver function tests were normal. CT of the brain demonstrated no evidence of acute pathology.

Given that her laboratory results (aside from the hemoglobin) and CT were essentially normal, the most common etiology of a recurrent encephalopathy would be a toxic exposure including drugs, alcohol, and environmental toxins or poisons. A comprehensive serum drug screen, including heavy metals, could follow a basic urinary screen for drugs of abuse; specific etiologies may be suggested by patterns of injury seen on MRI such as those seen with carbon monoxide or methanol exposure. Other recurrent metabolic processes include the porphyrias and relapsing inflammatory disorders, which could be entertained if further diagnostics are unrevealing.

An EEG is warranted at this point and is a test that is underutilized in the workup of altered mental status. Patients who have a spell and do not quickly awaken should be considered to be in nonconvulsive status epilepticus until proven otherwise. This can be easily identified on the EEG and is an important entity to recognize quickly. Additional findings on EEG may suggest focal cerebral dysfunction (such as that following a seizure or acute unilateral injury), diffuse encephalopathy (eg, triphasic waves), or fairly specific diagnoses (eg, periodic lateralized epileptiform discharges from the temporal lobes in suspected herpes simplex meningoencephalitis). While the CT of the brain is a reasonable initial screen, MRI is more sensitive for structural disease and should be obtained if no etiology is rapidly identified.

Finally, acute infectious etiologies such as abscess, encephalitis, or meningoencephalitis need to be excluded via lumbar puncture. Spinal fluid examination also can be helpful in the consideration of inflammatory and autoimmune disorders.

Over the next several hours, while still in the emergency department, she became increasingly obtunded, to the point that she was unresponsive to all stimuli. No seizure activity was witnessed, her vital signs were unchanged, and no medications had been administered. She was urgently transferred to a tertiary care center, where, at the time of arrival, she was obtunded and nonverbal, and opened her eyes only to noxious stimuli. She would withdraw all 4 extremities in response to pain. Pupils were 2 mm and symmetrically reactive. Corneal reflexes were normal, and her gag reflex was diminished. Motor tone was decreased in all 4 extremities. No fasciculations were noted. Deep tendon reflexes were present but symmetrically diminished throughout, and Babinski testing demonstrated a withdrawal response bilaterally.

Coma is a state of profound unconsciousness where the patient is unarousable and unaware of her surroundings. Coma can result either from bihemispheric dysfunction or diffuse injury to the reticular activating system in the brainstem, and the physical examination should focus on distinguishing between these 2 sites. Because the nuclei of cranial nerves III through XII (excepting XI) reside in the brainstem, the coma examination emphasizes testing the cranial nerves; although all cranial nerves are not tested in this patient, the ones that are appear to be normal, making bihemispheric dysfunction most likely. Bihemispheric coma most commonly results from diffuse toxic or metabolic etiologies such as intoxication or hepatic encephalopathy, but it can also be caused by bilateral structural lesions (including the bilateral thalami) or ongoing seizure activity.

Although an EEG remains the key test in this patient given her past history and an MRI would prove extremely useful, her deterioration warrants a workup for CNS infection. Since the head CT was negative, it would be prudent to proceed with urgent lumbar puncture (although it should never be performed in a patient with significant coagulopathy due to risks of hemorrhage leading to spinal cord injury). She should be covered empirically with broad spectrum meningeal‐dose antibiotics, including acyclovir, until the results of the spinal fluid examination are known, given that bacterial meningitis and herpes meningoencephalitis carry a high morbidity and mortality if not treated promptly.

Routine blood tests were similar to her labs at the referring emergency room. Ammonia level was 10 mol/L. Urine toxicology screen was negative, and blood tests for ethanol, salicylates, lithium, and acetaminophen were negative. Chest X‐ray and urinalysis were normal, and electrocardiogram was notable only for a sinus tachycardia. Cultures of the blood were obtained and the patient was admitted to the intensive care unit.

Levetiracetam, vancomycin, piperacillin‐tazobactam, and acyclovir were initiated. A lumbar puncture was performed without reversing the anticoagulation, and the procedure was traumatic. The cerebrospinal fluid was bloody, with a clear supernatant. Cell count demonstrated a red blood cell (RBC) count of 1250/L and a WBC count of 9/L, with a WBC differential of 42% neutrophils, 48% lymphocytes, and 8% monocytes. The cerebrospinal fluid (CSF) glucose was 62 mg/dL (with a serum glucose of 74 mg/dL) and protein 41 mg/dL. The CSF Gram stain demonstrated no organisms, and fluid was sent for routine culture and polymerase chain reaction (PCR) to detect herpes simplex virus (HSV). A neurology consultation was urgently requested.

As mentioned, it would have been more appropriate to reverse the patient's anticoagulation prior to lumbar puncture. The absence of xanthochromia suggests that the RBCs seen in the sample were introduced at the time of the lumbar puncture, arguing against a hemorrhagic disorder of the CNS (occasionally seen with herpes simplex encephalitis) or spinal fluid (eg, subarachnoid hemorrhage).

A reasonable rule of thumb to correct for the number of RBCs in a traumatic lumbar puncture is to allow 1 WBC for every 700 RBCs/L. Given this conversion, there are still too many WBCs in this sample, indicating a mild pleocytosis that is approximately one‐half neutrophilic and one‐half lymphocytic. This profile is nonspecific and can occur with a variety of conditions including stroke, seizure, inflammatory disorders, and infections, including viruses such as West Nile virus.

While coverage with acyclovir and broad‐spectrum antibacterials is appropriate, it should be noted that piperacillin‐tazobactam has poor CSF penetration and therefore is not a good choice for empiric coverage of CNS infections.

The neurologist's examination additionally noted multifocal myoclonus with noxious stimuli, most prominent in the face and toes. An urgent EEG demonstrated continuous, slow, generalized triphasic wave activity (Figures 1 and 2); no epileptiform discharges were seen. The erythrocyte sedimentation rate (ESR) was 66 mm/hour (normal, 0‐30), and tests for antinuclear antibodies, serum levetiracetam level, and thyroid function studies were ordered.

Figure 1

Figure 2

Stimulus‐evoked multifocal myoclonus is a general marker of encephalopathy found in many conditions, including hepatic and renal failure, drug intoxication (eg, opiates), neurodegenerative disorders (eg, Creutzfeldt‐Jakob disease [CJD]), and postanoxic injury, the latter of which is termed the Lance‐Adams syndrome.

Triphasic waves on EEG, while commonly associated with hepatic encephalopathy, have a similarly broad differential diagnosis, although in a comatose patient, they must first and foremost be distinguished from the repetitive discharges characteristic of nonconvulsive status epilepticus. In addition to hepatic and renal failure, triphasic waves have also been described in medication toxicity (especially with anticonvulsants, lithium, and cephalosporins), CNS infections (including Lyme disease and West Nile virus), strokes involving the bilateral thalami (usually from deep venous thrombosis), inflammatory disorders (such as Hashimoto's encephalopathy [HE]), and neurodegenerative diseases. It is important to remember that a single EEG does not exclude the possibility of an episodic ictal disorder and longer‐term monitoring would be required to definitively exclude seizures.

At this point, although the myoclonus and triphasic waves most commonly would indicate a toxic/metabolic process, the elevated ESR and CSF pleocytosis argue for an inflammatory or infectious condition. An MRI remains the next most useful test to guide further workup because many such conditions have distinct signatures on MRI.

The following day, she was noted to have periods of alertnessopening her eyes and following some commandsbut at other times she was difficult to arouse or obtunded. Tremulous movements and sporadic myoclonic jerks continued but no focal neurologic signs were found. Although there was increased muscle tone throughout, she was intermittently seen moving her limbs spontaneously, but not to command. No new findings were appreciated on routine laboratory tests. Antinuclear antibody testing was negative. Serum levetiracetam level was 23.5 g/mL (reference range, 545). Serum thyroid‐stimulating hormone was less than _0.005 U/mL, but free T3 was 3.5 pg/mL (normal, 1.8‐4.6) and free T4 was 2.0 ng/dL (normal, 0.71.8). An MRI of the brain was compromised by motion artifact but no significant abnormalities were appreciated.

At this point, a family member in another state disclosed that the patient had also been hospitalized 2 months previously while visiting him. Her chief complaint had been shortness of breath. The records were obtained; a cardiac catheterization had revealed nonobstructive coronary disease, and medical management was recommended. The notes mentioned that during the hospitalization she developed altered mental status with disorientation and shaking. CT and MRI of the brain had been unremarkable. The confusion was not explained, but she was discharged in good condition, alert and fully‐oriented.

The additional history confirms a relapsing encephalopathy, now with at least 3 occurrences. The most common etiologies in the face of a normal MRI and basic labs would be recurrent intoxication or exposures, but the inflammatory CSF profile and elevated ESR are not consistent with this. A variety of inflammatory disorders can present with recurrent encephalopathy, including demyelinating diseases and neurosarcoidosis. Some systemic rheumatologic conditions, such as systemic lupus erythematosus, can present with relapsing neurologic symptoms due to seizures, vasculitis, or cerebritis. Vasculitis would fit this picture as well, except for the normal findings on 2 MRIs. In a patient with such dramatic symptoms of neurologic dysfunction, one would expect to see changes on the MRI of cerebral inflammation with probable ischemia.

Therefore, given the CSF, ESR, clinical course, and unrevealing MRI and EEG, the most likely group of disorders responsible would be the nonvasculitic autoimmune meningoencephalitides, which present with recurrent encephalopathy and feature spontaneous remissions and/or often‐dramatic responses to corticosteroids. Key disorders in this category include Sjogren's disease, lupus, and steroid responsive encephalopathy associated with autoimmune thyroiditis (Hashimoto's encephalopathy). The latter condition is the most common of the group and is suggested by the abnormal thyroid‐stimulating hormone testing, although it may occur in the setting of normal thyroid function. The diagnosis can be confirmed with thyroperoxidase and thyroglobulin antibody testing.

Three days into the hospitalization, her mental status had gradually improved such that she was more consistently awake and oriented to person and place, and she was transferred to a regular nursing unit. Final results from the CSF and blood cultures were negative, as was PCR for HSV. The antimicrobials were discontinued. Routine serum chemistries continued to be unremarkable. Additional studies recommended by the neurologist demonstrated an antithyroperoxidase antibody concentration of 587.1 IU/mL (normal, <5), and antithyroglobulin antibody level of 52.2 IU/mL (normal, <10).

These results confirm the diagnosis of HE which, in addition to its presentation as a recurrent illness, is an important treatable cause of dementia and should be considered in young patients, those with autoimmune and thyroid disorders, and those whose dementia is rapidly progressive. Most cases are thought to be steroid‐responsive, but some studies have defined the disorder based on this responsiveness, resulting in some nonresponders likely being overlooked.

A trial of corticosteroids should be considered if the patient does not quickly return to baseline given the potential morbidities associated with prolonged altered mental status to this degree. Whether initiation of chronic immunosuppression could prevent these attacks in the future is unclear from the literature but should be considered given the recurrent, dramatic presentation in this patient.

A diagnosis of HE was made, and she was prescribed corticosteroids. Twenty‐four hours later, she was alert and fully‐oriented. She was discharged to home on prednisone and seen in follow‐up in neurology clinic 1 month later. She had had no further episodes of confusion or stupor, but because of steroid‐induced hyperglycemia, her corticosteroids were decreased and mycophenolate mofetil added for chronic immunosuppression. Four months after discharge she was neurologically stable but continued to struggle with the adverse effects of chronic corticosteroid treatment.

COMMENTARY

HE is an uncommon condition that can present with a rapidly progressive decline and should be considered in patients who present with recurrent mental status change in the setting of normal imaging studies and routine laboratory results. The entity was initially described by Lord William Russell Brain in 1966, and in the most recent terminology is known as steroid‐responsive encephalopathy associated with autoimmune thyroiditis (SREAT).1 It is characterized by an acute or subacute encephalopathy associated with thyroid autoimmunity. Patients typically present with fluctuating symptoms, episodes of confusion, alterations of consciousness, and rapid cognitive decline.2 Common features include myoclonus, tremor, ataxia, speech disturbance, stroke‐like episodes, increased muscle tone, neuropsychiatric manifestations, and seizures, that in some cases may progress to status epilepticus.3, 4

Although serum antithyroglobulin and antithyroperoxidase antibodies are elevated in HE, their presence is thought to be an epiphenomenon of the condition rather than the direct cause. Supporting this are the facts that the incidence of encephalopathy is not increased in patients with established autoimmune thyroiditis, and the presence of antithyroid antibodies ranges from 5% to 20% in the general population.2, 5 There is also no evidence that thyroid antibodies directly react with brain tissue, and the levels of these antibodies do not correlate with either neurologic manifestations or clinical improvement.2, 4, 5 As HE has been reported in patients with euthyroidism, hypothyroidism, and hyperthyroidism (with hypothyroidismeither subclinical or activemost common), it is also unlikely that the level of thyroid hormones play a role in the etiology of this disease.2, 4, 6

The etiology and pathogenesis of HE are unclear, although an immune‐mediated process is generally implicated, either from an inflammatory vasculitis or as a form of acute disseminated encephalomyelitis.7‐9 Global hypoperfusion on single‐photon emission computed tomography (SPECT) studies has also been reported.10, 11 Patients with HE may have nonspecific evidence of inflammation, including an elevated ESR, CRP, and CSF protein.12 Other laboratory abnormalities may include a mild elevation of liver aminotransferase levels; renal impairment has also been reported in a few cases of HE in the form of glomerulonephritis, and may be related to deposition of immune complexes containing thyroglobulin antigen.6, 12‐14 MRI of the brain is normal or nonspecific in most cases, and the EEG most commonly shows diffuse slowing.

The differential for a rapidly progressive cognitive decline includes CJD, CNS vasculitis, paraneoplastic syndromes, and autoimmune and subacute infectious encephalopathies. In patients with CJD, T2‐weighted imaging may show hyperintense signals in the basal ganglia, while diffusion‐weighted sequences may reveal changes in the cortical ribbon and bilateral thalami.15 In CNS vasculitis, the imaging findings are variable and range from discrete areas of vascular infarcts to hemorrhagic lesions.16 In paraneoplastic and autoimmune encephalopathies (excluding HE), MRI often shows nonenhancing signal intensity changes in the mesial temporal lobes.12 This patient had repeatedly normal MRI studies of the brain, which in combination with the history of tremor, myoclonus, seizures, and interval return to baseline status, helped point to the diagnosis of HE.

Different approaches to treatment of HE have been recommended. As the acronym SREAT suggests, patients typically respond dramatically to high‐dose steroid therapy. Although a number of patients also improve spontaneously, up to 60% of patients experience a relapsing course and require chronic immunosuppressive agents for maintenance therapy, including long‐term steroids and azathioprine.2, 17 Treatment with plasma exchange and intravenous immune globulin have also been reported, but with mixed results.18, 19 Due to her history of multiple relapses, the patient was placed on mycophenolate mofetil for additional maintenance immunosuppression, as her corticosteroid dose was reduced due to adverse effects.

Acute mental status change is a potentially emergent situation that must be evaluated with careful history and studies to exclude life‐threatening metabolic, infectious, and vascular conditions. This patient presented similarly on 2 prior occasions, and each time her physician team evaluated what appeared to be a new onset of altered consciousness, reaching a plausible but ultimately incorrect diagnosis. The patient's third presentation was finally the charm, as her physicians learned of the repeated history of a confusional state, and in particular the return to baseline status, allowing them to create a differential that focused on etiologies of relapsing encephalopathy and make the correct diagnosis.

A 58‐year old woman was brought to the emergency department with confusion. Her husband stated that for several hours she had been drifting in and out at home, and that he had to shout to get her attention. He described no seizure activity, weakness, incontinence, or difficulty speaking, and had noted no complaints of headache, fevers, chest pain, shortness of breath, or gastrointestinal complaints.

Altered mental status in a middle‐aged woman can result from a diverse set of etiologies. A key distinction in the neurological examination will be to assure that the complaint of confusion is accurate as opposed to aphasia; the former is usually indicative of diffuse cerebral dysfunction while the latter suggests a focal lesion in the dominant hemisphere.

The acuity of the change in mental status is important, as are the fluctuations described by the husband. Unwitnessed or nonconvulsive seizure activity can present this way. Toxic/metabolic etiologies, infectious and inflammatory disorders of the central nervous system (CNS), and vascular diseases are also important considerations. Although stroke does not typically present with global encephalopathy, intermittent large vessel occlusion, especially in the posterior circulation, can disrupt cognition in this manner. Following a physical examination, initial workup should focus on toxic/metabolic etiologies, followed rapidly by head imaging if no cause is identified.

Her past medical history was notable for type 2 diabetes mellitus, coronary artery disease, hyperlipidemia, and an unspecified seizure disorder, which according to her husband was diagnosed during a recent hospitalization for a similar presentation. She also had a remote history of venous thromboembolism and antithrombin‐III deficiency. She was unemployed, lived with her husband, and spent most of her time at home. She never smoked, and rarely drank alcohol. Her family history was unobtainable, and her husband denied that she used any illicit drugs. Her medications included pioglitazone, aspirin, simvastatin, pregabalin, ferrous sulfate, levetiracetam, warfarin, and magnesium oxide, and she was allergic to sulfa.

While the differential diagnosis remains broad, 3 elements of the history are potentially relevant. The history of epilepsy based on a similar prior presentation increases the likelihood that the current spell is ictal in nature; examination of previous records would be important in order to document whether these spells have indeed been proven to be epileptic, as many conditions can mimic seizures. Given the history of venous thromboembolism and hypercoagulability, one must consider cerebral venous sinus thrombosis, which can present with global neurologic dysfunction and seizures. Prompt identification, usually via computed tomography (CT) or magnetic resonance angiography, is vital, because anticoagulation can mitigate this potentially life‐threatening illness. Finally, although many medications can cause encephalopathy in overdose, levetiracetam has well‐described cognitive side effects even at usual doses, including encephalopathy, irritability, and depression.

The records from that recent hospitalization remarked that she had presented confused and stuporous. Her potassium had been 2.7 mmol/L, international normalized ration (INR) 3.4, and hemoglobin 8 g/dL; other routine laboratory studies were normal. CT and magnetic resonance imaging (MRI) of the brain had been negative, and electroencephalogram (EEG) reportedly was performed but specific results were unknown. She was discharged alert and oriented 1 week prior to the current presentation on the above medications, including levetiracetam for this newly‐diagnosed seizure disorder.

Previous records confirm that the current presentation is that of a relapsing acute alteration in mental status. Regardless of the EEG findings or response to antiepileptic medications, a seizure disorder should remain a primary consideration, although relapsing inflammatory, toxic/metabolic conditions, and, rarely, vascular disorders can also present in this manner.

The neurologic manifestations of hypokalemia are usually peripheral in nature, including periodic paralysis; confusion accompanying hypokalemia is usually not a result of the low potassium itself but rather due to an underlying toxic or endocrinologic cause. Various causes of anemia can lead to mental status changes; the mean corpuscular volume (MCV) will be particularly helpful given known associations between megaloblastic anemia and confusional states.

On examination, she appeared to be in good health and in no distress. She was afebrile. Her blood pressure was 93/57, pulse 90 beats per minute, respiratory rate 16 per minute, and room air oxygen saturation 100%. She was oriented to her surroundings, but slow in her responses to questioning. There were no cranial nerve, motor, or sensory deficits, or abnormal reflexes or movements. Examination of the head, skin, chest, cardiovascular system, abdomen, and extremities was normal. Serum sodium was 136 mmol/L, creatinine 1.2 mg/dL, calcium 9.3 mg/dL, and glucose 81 mg/dL; other routine blood chemistries were normal. Her white blood cell (WBC) count was 7100/L, hemoglobin 9.2 g/dL with normal MCV, and platelet count 275,000/L. INR was 3.4, and liver function tests were normal. CT of the brain demonstrated no evidence of acute pathology.

Given that her laboratory results (aside from the hemoglobin) and CT were essentially normal, the most common etiology of a recurrent encephalopathy would be a toxic exposure including drugs, alcohol, and environmental toxins or poisons. A comprehensive serum drug screen, including heavy metals, could follow a basic urinary screen for drugs of abuse; specific etiologies may be suggested by patterns of injury seen on MRI such as those seen with carbon monoxide or methanol exposure. Other recurrent metabolic processes include the porphyrias and relapsing inflammatory disorders, which could be entertained if further diagnostics are unrevealing.

An EEG is warranted at this point and is a test that is underutilized in the workup of altered mental status. Patients who have a spell and do not quickly awaken should be considered to be in nonconvulsive status epilepticus until proven otherwise. This can be easily identified on the EEG and is an important entity to recognize quickly. Additional findings on EEG may suggest focal cerebral dysfunction (such as that following a seizure or acute unilateral injury), diffuse encephalopathy (eg, triphasic waves), or fairly specific diagnoses (eg, periodic lateralized epileptiform discharges from the temporal lobes in suspected herpes simplex meningoencephalitis). While the CT of the brain is a reasonable initial screen, MRI is more sensitive for structural disease and should be obtained if no etiology is rapidly identified.

Finally, acute infectious etiologies such as abscess, encephalitis, or meningoencephalitis need to be excluded via lumbar puncture. Spinal fluid examination also can be helpful in the consideration of inflammatory and autoimmune disorders.

Over the next several hours, while still in the emergency department, she became increasingly obtunded, to the point that she was unresponsive to all stimuli. No seizure activity was witnessed, her vital signs were unchanged, and no medications had been administered. She was urgently transferred to a tertiary care center, where, at the time of arrival, she was obtunded and nonverbal, and opened her eyes only to noxious stimuli. She would withdraw all 4 extremities in response to pain. Pupils were 2 mm and symmetrically reactive. Corneal reflexes were normal, and her gag reflex was diminished. Motor tone was decreased in all 4 extremities. No fasciculations were noted. Deep tendon reflexes were present but symmetrically diminished throughout, and Babinski testing demonstrated a withdrawal response bilaterally.

Coma is a state of profound unconsciousness where the patient is unarousable and unaware of her surroundings. Coma can result either from bihemispheric dysfunction or diffuse injury to the reticular activating system in the brainstem, and the physical examination should focus on distinguishing between these 2 sites. Because the nuclei of cranial nerves III through XII (excepting XI) reside in the brainstem, the coma examination emphasizes testing the cranial nerves; although all cranial nerves are not tested in this patient, the ones that are appear to be normal, making bihemispheric dysfunction most likely. Bihemispheric coma most commonly results from diffuse toxic or metabolic etiologies such as intoxication or hepatic encephalopathy, but it can also be caused by bilateral structural lesions (including the bilateral thalami) or ongoing seizure activity.

Although an EEG remains the key test in this patient given her past history and an MRI would prove extremely useful, her deterioration warrants a workup for CNS infection. Since the head CT was negative, it would be prudent to proceed with urgent lumbar puncture (although it should never be performed in a patient with significant coagulopathy due to risks of hemorrhage leading to spinal cord injury). She should be covered empirically with broad spectrum meningeal‐dose antibiotics, including acyclovir, until the results of the spinal fluid examination are known, given that bacterial meningitis and herpes meningoencephalitis carry a high morbidity and mortality if not treated promptly.

Routine blood tests were similar to her labs at the referring emergency room. Ammonia level was 10 mol/L. Urine toxicology screen was negative, and blood tests for ethanol, salicylates, lithium, and acetaminophen were negative. Chest X‐ray and urinalysis were normal, and electrocardiogram was notable only for a sinus tachycardia. Cultures of the blood were obtained and the patient was admitted to the intensive care unit.

Levetiracetam, vancomycin, piperacillin‐tazobactam, and acyclovir were initiated. A lumbar puncture was performed without reversing the anticoagulation, and the procedure was traumatic. The cerebrospinal fluid was bloody, with a clear supernatant. Cell count demonstrated a red blood cell (RBC) count of 1250/L and a WBC count of 9/L, with a WBC differential of 42% neutrophils, 48% lymphocytes, and 8% monocytes. The cerebrospinal fluid (CSF) glucose was 62 mg/dL (with a serum glucose of 74 mg/dL) and protein 41 mg/dL. The CSF Gram stain demonstrated no organisms, and fluid was sent for routine culture and polymerase chain reaction (PCR) to detect herpes simplex virus (HSV). A neurology consultation was urgently requested.

As mentioned, it would have been more appropriate to reverse the patient's anticoagulation prior to lumbar puncture. The absence of xanthochromia suggests that the RBCs seen in the sample were introduced at the time of the lumbar puncture, arguing against a hemorrhagic disorder of the CNS (occasionally seen with herpes simplex encephalitis) or spinal fluid (eg, subarachnoid hemorrhage).

A reasonable rule of thumb to correct for the number of RBCs in a traumatic lumbar puncture is to allow 1 WBC for every 700 RBCs/L. Given this conversion, there are still too many WBCs in this sample, indicating a mild pleocytosis that is approximately one‐half neutrophilic and one‐half lymphocytic. This profile is nonspecific and can occur with a variety of conditions including stroke, seizure, inflammatory disorders, and infections, including viruses such as West Nile virus.

While coverage with acyclovir and broad‐spectrum antibacterials is appropriate, it should be noted that piperacillin‐tazobactam has poor CSF penetration and therefore is not a good choice for empiric coverage of CNS infections.

The neurologist's examination additionally noted multifocal myoclonus with noxious stimuli, most prominent in the face and toes. An urgent EEG demonstrated continuous, slow, generalized triphasic wave activity (Figures 1 and 2); no epileptiform discharges were seen. The erythrocyte sedimentation rate (ESR) was 66 mm/hour (normal, 0‐30), and tests for antinuclear antibodies, serum levetiracetam level, and thyroid function studies were ordered.

Figure 1

Figure 2

Stimulus‐evoked multifocal myoclonus is a general marker of encephalopathy found in many conditions, including hepatic and renal failure, drug intoxication (eg, opiates), neurodegenerative disorders (eg, Creutzfeldt‐Jakob disease [CJD]), and postanoxic injury, the latter of which is termed the Lance‐Adams syndrome.

Triphasic waves on EEG, while commonly associated with hepatic encephalopathy, have a similarly broad differential diagnosis, although in a comatose patient, they must first and foremost be distinguished from the repetitive discharges characteristic of nonconvulsive status epilepticus. In addition to hepatic and renal failure, triphasic waves have also been described in medication toxicity (especially with anticonvulsants, lithium, and cephalosporins), CNS infections (including Lyme disease and West Nile virus), strokes involving the bilateral thalami (usually from deep venous thrombosis), inflammatory disorders (such as Hashimoto's encephalopathy [HE]), and neurodegenerative diseases. It is important to remember that a single EEG does not exclude the possibility of an episodic ictal disorder and longer‐term monitoring would be required to definitively exclude seizures.

At this point, although the myoclonus and triphasic waves most commonly would indicate a toxic/metabolic process, the elevated ESR and CSF pleocytosis argue for an inflammatory or infectious condition. An MRI remains the next most useful test to guide further workup because many such conditions have distinct signatures on MRI.

The following day, she was noted to have periods of alertnessopening her eyes and following some commandsbut at other times she was difficult to arouse or obtunded. Tremulous movements and sporadic myoclonic jerks continued but no focal neurologic signs were found. Although there was increased muscle tone throughout, she was intermittently seen moving her limbs spontaneously, but not to command. No new findings were appreciated on routine laboratory tests. Antinuclear antibody testing was negative. Serum levetiracetam level was 23.5 g/mL (reference range, 545). Serum thyroid‐stimulating hormone was less than _0.005 U/mL, but free T3 was 3.5 pg/mL (normal, 1.8‐4.6) and free T4 was 2.0 ng/dL (normal, 0.71.8). An MRI of the brain was compromised by motion artifact but no significant abnormalities were appreciated.

At this point, a family member in another state disclosed that the patient had also been hospitalized 2 months previously while visiting him. Her chief complaint had been shortness of breath. The records were obtained; a cardiac catheterization had revealed nonobstructive coronary disease, and medical management was recommended. The notes mentioned that during the hospitalization she developed altered mental status with disorientation and shaking. CT and MRI of the brain had been unremarkable. The confusion was not explained, but she was discharged in good condition, alert and fully‐oriented.

The additional history confirms a relapsing encephalopathy, now with at least 3 occurrences. The most common etiologies in the face of a normal MRI and basic labs would be recurrent intoxication or exposures, but the inflammatory CSF profile and elevated ESR are not consistent with this. A variety of inflammatory disorders can present with recurrent encephalopathy, including demyelinating diseases and neurosarcoidosis. Some systemic rheumatologic conditions, such as systemic lupus erythematosus, can present with relapsing neurologic symptoms due to seizures, vasculitis, or cerebritis. Vasculitis would fit this picture as well, except for the normal findings on 2 MRIs. In a patient with such dramatic symptoms of neurologic dysfunction, one would expect to see changes on the MRI of cerebral inflammation with probable ischemia.

Therefore, given the CSF, ESR, clinical course, and unrevealing MRI and EEG, the most likely group of disorders responsible would be the nonvasculitic autoimmune meningoencephalitides, which present with recurrent encephalopathy and feature spontaneous remissions and/or often‐dramatic responses to corticosteroids. Key disorders in this category include Sjogren's disease, lupus, and steroid responsive encephalopathy associated with autoimmune thyroiditis (Hashimoto's encephalopathy). The latter condition is the most common of the group and is suggested by the abnormal thyroid‐stimulating hormone testing, although it may occur in the setting of normal thyroid function. The diagnosis can be confirmed with thyroperoxidase and thyroglobulin antibody testing.

Three days into the hospitalization, her mental status had gradually improved such that she was more consistently awake and oriented to person and place, and she was transferred to a regular nursing unit. Final results from the CSF and blood cultures were negative, as was PCR for HSV. The antimicrobials were discontinued. Routine serum chemistries continued to be unremarkable. Additional studies recommended by the neurologist demonstrated an antithyroperoxidase antibody concentration of 587.1 IU/mL (normal, <5), and antithyroglobulin antibody level of 52.2 IU/mL (normal, <10).

These results confirm the diagnosis of HE which, in addition to its presentation as a recurrent illness, is an important treatable cause of dementia and should be considered in young patients, those with autoimmune and thyroid disorders, and those whose dementia is rapidly progressive. Most cases are thought to be steroid‐responsive, but some studies have defined the disorder based on this responsiveness, resulting in some nonresponders likely being overlooked.

A trial of corticosteroids should be considered if the patient does not quickly return to baseline given the potential morbidities associated with prolonged altered mental status to this degree. Whether initiation of chronic immunosuppression could prevent these attacks in the future is unclear from the literature but should be considered given the recurrent, dramatic presentation in this patient.

A diagnosis of HE was made, and she was prescribed corticosteroids. Twenty‐four hours later, she was alert and fully‐oriented. She was discharged to home on prednisone and seen in follow‐up in neurology clinic 1 month later. She had had no further episodes of confusion or stupor, but because of steroid‐induced hyperglycemia, her corticosteroids were decreased and mycophenolate mofetil added for chronic immunosuppression. Four months after discharge she was neurologically stable but continued to struggle with the adverse effects of chronic corticosteroid treatment.

COMMENTARY

HE is an uncommon condition that can present with a rapidly progressive decline and should be considered in patients who present with recurrent mental status change in the setting of normal imaging studies and routine laboratory results. The entity was initially described by Lord William Russell Brain in 1966, and in the most recent terminology is known as steroid‐responsive encephalopathy associated with autoimmune thyroiditis (SREAT).1 It is characterized by an acute or subacute encephalopathy associated with thyroid autoimmunity. Patients typically present with fluctuating symptoms, episodes of confusion, alterations of consciousness, and rapid cognitive decline.2 Common features include myoclonus, tremor, ataxia, speech disturbance, stroke‐like episodes, increased muscle tone, neuropsychiatric manifestations, and seizures, that in some cases may progress to status epilepticus.3, 4

Although serum antithyroglobulin and antithyroperoxidase antibodies are elevated in HE, their presence is thought to be an epiphenomenon of the condition rather than the direct cause. Supporting this are the facts that the incidence of encephalopathy is not increased in patients with established autoimmune thyroiditis, and the presence of antithyroid antibodies ranges from 5% to 20% in the general population.2, 5 There is also no evidence that thyroid antibodies directly react with brain tissue, and the levels of these antibodies do not correlate with either neurologic manifestations or clinical improvement.2, 4, 5 As HE has been reported in patients with euthyroidism, hypothyroidism, and hyperthyroidism (with hypothyroidismeither subclinical or activemost common), it is also unlikely that the level of thyroid hormones play a role in the etiology of this disease.2, 4, 6

The etiology and pathogenesis of HE are unclear, although an immune‐mediated process is generally implicated, either from an inflammatory vasculitis or as a form of acute disseminated encephalomyelitis.7‐9 Global hypoperfusion on single‐photon emission computed tomography (SPECT) studies has also been reported.10, 11 Patients with HE may have nonspecific evidence of inflammation, including an elevated ESR, CRP, and CSF protein.12 Other laboratory abnormalities may include a mild elevation of liver aminotransferase levels; renal impairment has also been reported in a few cases of HE in the form of glomerulonephritis, and may be related to deposition of immune complexes containing thyroglobulin antigen.6, 12‐14 MRI of the brain is normal or nonspecific in most cases, and the EEG most commonly shows diffuse slowing.

The differential for a rapidly progressive cognitive decline includes CJD, CNS vasculitis, paraneoplastic syndromes, and autoimmune and subacute infectious encephalopathies. In patients with CJD, T2‐weighted imaging may show hyperintense signals in the basal ganglia, while diffusion‐weighted sequences may reveal changes in the cortical ribbon and bilateral thalami.15 In CNS vasculitis, the imaging findings are variable and range from discrete areas of vascular infarcts to hemorrhagic lesions.16 In paraneoplastic and autoimmune encephalopathies (excluding HE), MRI often shows nonenhancing signal intensity changes in the mesial temporal lobes.12 This patient had repeatedly normal MRI studies of the brain, which in combination with the history of tremor, myoclonus, seizures, and interval return to baseline status, helped point to the diagnosis of HE.

Different approaches to treatment of HE have been recommended. As the acronym SREAT suggests, patients typically respond dramatically to high‐dose steroid therapy. Although a number of patients also improve spontaneously, up to 60% of patients experience a relapsing course and require chronic immunosuppressive agents for maintenance therapy, including long‐term steroids and azathioprine.2, 17 Treatment with plasma exchange and intravenous immune globulin have also been reported, but with mixed results.18, 19 Due to her history of multiple relapses, the patient was placed on mycophenolate mofetil for additional maintenance immunosuppression, as her corticosteroid dose was reduced due to adverse effects.

Acute mental status change is a potentially emergent situation that must be evaluated with careful history and studies to exclude life‐threatening metabolic, infectious, and vascular conditions. This patient presented similarly on 2 prior occasions, and each time her physician team evaluated what appeared to be a new onset of altered consciousness, reaching a plausible but ultimately incorrect diagnosis. The patient's third presentation was finally the charm, as her physicians learned of the repeated history of a confusional state, and in particular the return to baseline status, allowing them to create a differential that focused on etiologies of relapsing encephalopathy and make the correct diagnosis.

A 58‐year old woman was brought to the emergency department with confusion. Her husband stated that for several hours she had been drifting in and out at home, and that he had to shout to get her attention. He described no seizure activity, weakness, incontinence, or difficulty speaking, and had noted no complaints of headache, fevers, chest pain, shortness of breath, or gastrointestinal complaints.

Altered mental status in a middle‐aged woman can result from a diverse set of etiologies. A key distinction in the neurological examination will be to assure that the complaint of confusion is accurate as opposed to aphasia; the former is usually indicative of diffuse cerebral dysfunction while the latter suggests a focal lesion in the dominant hemisphere.

The acuity of the change in mental status is important, as are the fluctuations described by the husband. Unwitnessed or nonconvulsive seizure activity can present this way. Toxic/metabolic etiologies, infectious and inflammatory disorders of the central nervous system (CNS), and vascular diseases are also important considerations. Although stroke does not typically present with global encephalopathy, intermittent large vessel occlusion, especially in the posterior circulation, can disrupt cognition in this manner. Following a physical examination, initial workup should focus on toxic/metabolic etiologies, followed rapidly by head imaging if no cause is identified.

Her past medical history was notable for type 2 diabetes mellitus, coronary artery disease, hyperlipidemia, and an unspecified seizure disorder, which according to her husband was diagnosed during a recent hospitalization for a similar presentation. She also had a remote history of venous thromboembolism and antithrombin‐III deficiency. She was unemployed, lived with her husband, and spent most of her time at home. She never smoked, and rarely drank alcohol. Her family history was unobtainable, and her husband denied that she used any illicit drugs. Her medications included pioglitazone, aspirin, simvastatin, pregabalin, ferrous sulfate, levetiracetam, warfarin, and magnesium oxide, and she was allergic to sulfa.

While the differential diagnosis remains broad, 3 elements of the history are potentially relevant. The history of epilepsy based on a similar prior presentation increases the likelihood that the current spell is ictal in nature; examination of previous records would be important in order to document whether these spells have indeed been proven to be epileptic, as many conditions can mimic seizures. Given the history of venous thromboembolism and hypercoagulability, one must consider cerebral venous sinus thrombosis, which can present with global neurologic dysfunction and seizures. Prompt identification, usually via computed tomography (CT) or magnetic resonance angiography, is vital, because anticoagulation can mitigate this potentially life‐threatening illness. Finally, although many medications can cause encephalopathy in overdose, levetiracetam has well‐described cognitive side effects even at usual doses, including encephalopathy, irritability, and depression.

The records from that recent hospitalization remarked that she had presented confused and stuporous. Her potassium had been 2.7 mmol/L, international normalized ration (INR) 3.4, and hemoglobin 8 g/dL; other routine laboratory studies were normal. CT and magnetic resonance imaging (MRI) of the brain had been negative, and electroencephalogram (EEG) reportedly was performed but specific results were unknown. She was discharged alert and oriented 1 week prior to the current presentation on the above medications, including levetiracetam for this newly‐diagnosed seizure disorder.

Previous records confirm that the current presentation is that of a relapsing acute alteration in mental status. Regardless of the EEG findings or response to antiepileptic medications, a seizure disorder should remain a primary consideration, although relapsing inflammatory, toxic/metabolic conditions, and, rarely, vascular disorders can also present in this manner.

The neurologic manifestations of hypokalemia are usually peripheral in nature, including periodic paralysis; confusion accompanying hypokalemia is usually not a result of the low potassium itself but rather due to an underlying toxic or endocrinologic cause. Various causes of anemia can lead to mental status changes; the mean corpuscular volume (MCV) will be particularly helpful given known associations between megaloblastic anemia and confusional states.

On examination, she appeared to be in good health and in no distress. She was afebrile. Her blood pressure was 93/57, pulse 90 beats per minute, respiratory rate 16 per minute, and room air oxygen saturation 100%. She was oriented to her surroundings, but slow in her responses to questioning. There were no cranial nerve, motor, or sensory deficits, or abnormal reflexes or movements. Examination of the head, skin, chest, cardiovascular system, abdomen, and extremities was normal. Serum sodium was 136 mmol/L, creatinine 1.2 mg/dL, calcium 9.3 mg/dL, and glucose 81 mg/dL; other routine blood chemistries were normal. Her white blood cell (WBC) count was 7100/L, hemoglobin 9.2 g/dL with normal MCV, and platelet count 275,000/L. INR was 3.4, and liver function tests were normal. CT of the brain demonstrated no evidence of acute pathology.

Given that her laboratory results (aside from the hemoglobin) and CT were essentially normal, the most common etiology of a recurrent encephalopathy would be a toxic exposure including drugs, alcohol, and environmental toxins or poisons. A comprehensive serum drug screen, including heavy metals, could follow a basic urinary screen for drugs of abuse; specific etiologies may be suggested by patterns of injury seen on MRI such as those seen with carbon monoxide or methanol exposure. Other recurrent metabolic processes include the porphyrias and relapsing inflammatory disorders, which could be entertained if further diagnostics are unrevealing.

An EEG is warranted at this point and is a test that is underutilized in the workup of altered mental status. Patients who have a spell and do not quickly awaken should be considered to be in nonconvulsive status epilepticus until proven otherwise. This can be easily identified on the EEG and is an important entity to recognize quickly. Additional findings on EEG may suggest focal cerebral dysfunction (such as that following a seizure or acute unilateral injury), diffuse encephalopathy (eg, triphasic waves), or fairly specific diagnoses (eg, periodic lateralized epileptiform discharges from the temporal lobes in suspected herpes simplex meningoencephalitis). While the CT of the brain is a reasonable initial screen, MRI is more sensitive for structural disease and should be obtained if no etiology is rapidly identified.

Finally, acute infectious etiologies such as abscess, encephalitis, or meningoencephalitis need to be excluded via lumbar puncture. Spinal fluid examination also can be helpful in the consideration of inflammatory and autoimmune disorders.

Over the next several hours, while still in the emergency department, she became increasingly obtunded, to the point that she was unresponsive to all stimuli. No seizure activity was witnessed, her vital signs were unchanged, and no medications had been administered. She was urgently transferred to a tertiary care center, where, at the time of arrival, she was obtunded and nonverbal, and opened her eyes only to noxious stimuli. She would withdraw all 4 extremities in response to pain. Pupils were 2 mm and symmetrically reactive. Corneal reflexes were normal, and her gag reflex was diminished. Motor tone was decreased in all 4 extremities. No fasciculations were noted. Deep tendon reflexes were present but symmetrically diminished throughout, and Babinski testing demonstrated a withdrawal response bilaterally.

Coma is a state of profound unconsciousness where the patient is unarousable and unaware of her surroundings. Coma can result either from bihemispheric dysfunction or diffuse injury to the reticular activating system in the brainstem, and the physical examination should focus on distinguishing between these 2 sites. Because the nuclei of cranial nerves III through XII (excepting XI) reside in the brainstem, the coma examination emphasizes testing the cranial nerves; although all cranial nerves are not tested in this patient, the ones that are appear to be normal, making bihemispheric dysfunction most likely. Bihemispheric coma most commonly results from diffuse toxic or metabolic etiologies such as intoxication or hepatic encephalopathy, but it can also be caused by bilateral structural lesions (including the bilateral thalami) or ongoing seizure activity.

Although an EEG remains the key test in this patient given her past history and an MRI would prove extremely useful, her deterioration warrants a workup for CNS infection. Since the head CT was negative, it would be prudent to proceed with urgent lumbar puncture (although it should never be performed in a patient with significant coagulopathy due to risks of hemorrhage leading to spinal cord injury). She should be covered empirically with broad spectrum meningeal‐dose antibiotics, including acyclovir, until the results of the spinal fluid examination are known, given that bacterial meningitis and herpes meningoencephalitis carry a high morbidity and mortality if not treated promptly.

Routine blood tests were similar to her labs at the referring emergency room. Ammonia level was 10 mol/L. Urine toxicology screen was negative, and blood tests for ethanol, salicylates, lithium, and acetaminophen were negative. Chest X‐ray and urinalysis were normal, and electrocardiogram was notable only for a sinus tachycardia. Cultures of the blood were obtained and the patient was admitted to the intensive care unit.

Levetiracetam, vancomycin, piperacillin‐tazobactam, and acyclovir were initiated. A lumbar puncture was performed without reversing the anticoagulation, and the procedure was traumatic. The cerebrospinal fluid was bloody, with a clear supernatant. Cell count demonstrated a red blood cell (RBC) count of 1250/L and a WBC count of 9/L, with a WBC differential of 42% neutrophils, 48% lymphocytes, and 8% monocytes. The cerebrospinal fluid (CSF) glucose was 62 mg/dL (with a serum glucose of 74 mg/dL) and protein 41 mg/dL. The CSF Gram stain demonstrated no organisms, and fluid was sent for routine culture and polymerase chain reaction (PCR) to detect herpes simplex virus (HSV). A neurology consultation was urgently requested.

As mentioned, it would have been more appropriate to reverse the patient's anticoagulation prior to lumbar puncture. The absence of xanthochromia suggests that the RBCs seen in the sample were introduced at the time of the lumbar puncture, arguing against a hemorrhagic disorder of the CNS (occasionally seen with herpes simplex encephalitis) or spinal fluid (eg, subarachnoid hemorrhage).

A reasonable rule of thumb to correct for the number of RBCs in a traumatic lumbar puncture is to allow 1 WBC for every 700 RBCs/L. Given this conversion, there are still too many WBCs in this sample, indicating a mild pleocytosis that is approximately one‐half neutrophilic and one‐half lymphocytic. This profile is nonspecific and can occur with a variety of conditions including stroke, seizure, inflammatory disorders, and infections, including viruses such as West Nile virus.

While coverage with acyclovir and broad‐spectrum antibacterials is appropriate, it should be noted that piperacillin‐tazobactam has poor CSF penetration and therefore is not a good choice for empiric coverage of CNS infections.

The neurologist's examination additionally noted multifocal myoclonus with noxious stimuli, most prominent in the face and toes. An urgent EEG demonstrated continuous, slow, generalized triphasic wave activity (Figures 1 and 2); no epileptiform discharges were seen. The erythrocyte sedimentation rate (ESR) was 66 mm/hour (normal, 0‐30), and tests for antinuclear antibodies, serum levetiracetam level, and thyroid function studies were ordered.

Figure 1

Figure 2

Stimulus‐evoked multifocal myoclonus is a general marker of encephalopathy found in many conditions, including hepatic and renal failure, drug intoxication (eg, opiates), neurodegenerative disorders (eg, Creutzfeldt‐Jakob disease [CJD]), and postanoxic injury, the latter of which is termed the Lance‐Adams syndrome.

Triphasic waves on EEG, while commonly associated with hepatic encephalopathy, have a similarly broad differential diagnosis, although in a comatose patient, they must first and foremost be distinguished from the repetitive discharges characteristic of nonconvulsive status epilepticus. In addition to hepatic and renal failure, triphasic waves have also been described in medication toxicity (especially with anticonvulsants, lithium, and cephalosporins), CNS infections (including Lyme disease and West Nile virus), strokes involving the bilateral thalami (usually from deep venous thrombosis), inflammatory disorders (such as Hashimoto's encephalopathy [HE]), and neurodegenerative diseases. It is important to remember that a single EEG does not exclude the possibility of an episodic ictal disorder and longer‐term monitoring would be required to definitively exclude seizures.

At this point, although the myoclonus and triphasic waves most commonly would indicate a toxic/metabolic process, the elevated ESR and CSF pleocytosis argue for an inflammatory or infectious condition. An MRI remains the next most useful test to guide further workup because many such conditions have distinct signatures on MRI.

The following day, she was noted to have periods of alertnessopening her eyes and following some commandsbut at other times she was difficult to arouse or obtunded. Tremulous movements and sporadic myoclonic jerks continued but no focal neurologic signs were found. Although there was increased muscle tone throughout, she was intermittently seen moving her limbs spontaneously, but not to command. No new findings were appreciated on routine laboratory tests. Antinuclear antibody testing was negative. Serum levetiracetam level was 23.5 g/mL (reference range, 545). Serum thyroid‐stimulating hormone was less than _0.005 U/mL, but free T3 was 3.5 pg/mL (normal, 1.8‐4.6) and free T4 was 2.0 ng/dL (normal, 0.71.8). An MRI of the brain was compromised by motion artifact but no significant abnormalities were appreciated.

At this point, a family member in another state disclosed that the patient had also been hospitalized 2 months previously while visiting him. Her chief complaint had been shortness of breath. The records were obtained; a cardiac catheterization had revealed nonobstructive coronary disease, and medical management was recommended. The notes mentioned that during the hospitalization she developed altered mental status with disorientation and shaking. CT and MRI of the brain had been unremarkable. The confusion was not explained, but she was discharged in good condition, alert and fully‐oriented.

The additional history confirms a relapsing encephalopathy, now with at least 3 occurrences. The most common etiologies in the face of a normal MRI and basic labs would be recurrent intoxication or exposures, but the inflammatory CSF profile and elevated ESR are not consistent with this. A variety of inflammatory disorders can present with recurrent encephalopathy, including demyelinating diseases and neurosarcoidosis. Some systemic rheumatologic conditions, such as systemic lupus erythematosus, can present with relapsing neurologic symptoms due to seizures, vasculitis, or cerebritis. Vasculitis would fit this picture as well, except for the normal findings on 2 MRIs. In a patient with such dramatic symptoms of neurologic dysfunction, one would expect to see changes on the MRI of cerebral inflammation with probable ischemia.

Therefore, given the CSF, ESR, clinical course, and unrevealing MRI and EEG, the most likely group of disorders responsible would be the nonvasculitic autoimmune meningoencephalitides, which present with recurrent encephalopathy and feature spontaneous remissions and/or often‐dramatic responses to corticosteroids. Key disorders in this category include Sjogren's disease, lupus, and steroid responsive encephalopathy associated with autoimmune thyroiditis (Hashimoto's encephalopathy). The latter condition is the most common of the group and is suggested by the abnormal thyroid‐stimulating hormone testing, although it may occur in the setting of normal thyroid function. The diagnosis can be confirmed with thyroperoxidase and thyroglobulin antibody testing.

Three days into the hospitalization, her mental status had gradually improved such that she was more consistently awake and oriented to person and place, and she was transferred to a regular nursing unit. Final results from the CSF and blood cultures were negative, as was PCR for HSV. The antimicrobials were discontinued. Routine serum chemistries continued to be unremarkable. Additional studies recommended by the neurologist demonstrated an antithyroperoxidase antibody concentration of 587.1 IU/mL (normal, <5), and antithyroglobulin antibody level of 52.2 IU/mL (normal, <10).

These results confirm the diagnosis of HE which, in addition to its presentation as a recurrent illness, is an important treatable cause of dementia and should be considered in young patients, those with autoimmune and thyroid disorders, and those whose dementia is rapidly progressive. Most cases are thought to be steroid‐responsive, but some studies have defined the disorder based on this responsiveness, resulting in some nonresponders likely being overlooked.

A trial of corticosteroids should be considered if the patient does not quickly return to baseline given the potential morbidities associated with prolonged altered mental status to this degree. Whether initiation of chronic immunosuppression could prevent these attacks in the future is unclear from the literature but should be considered given the recurrent, dramatic presentation in this patient.

A diagnosis of HE was made, and she was prescribed corticosteroids. Twenty‐four hours later, she was alert and fully‐oriented. She was discharged to home on prednisone and seen in follow‐up in neurology clinic 1 month later. She had had no further episodes of confusion or stupor, but because of steroid‐induced hyperglycemia, her corticosteroids were decreased and mycophenolate mofetil added for chronic immunosuppression. Four months after discharge she was neurologically stable but continued to struggle with the adverse effects of chronic corticosteroid treatment.

COMMENTARY

HE is an uncommon condition that can present with a rapidly progressive decline and should be considered in patients who present with recurrent mental status change in the setting of normal imaging studies and routine laboratory results. The entity was initially described by Lord William Russell Brain in 1966, and in the most recent terminology is known as steroid‐responsive encephalopathy associated with autoimmune thyroiditis (SREAT).1 It is characterized by an acute or subacute encephalopathy associated with thyroid autoimmunity. Patients typically present with fluctuating symptoms, episodes of confusion, alterations of consciousness, and rapid cognitive decline.2 Common features include myoclonus, tremor, ataxia, speech disturbance, stroke‐like episodes, increased muscle tone, neuropsychiatric manifestations, and seizures, that in some cases may progress to status epilepticus.3, 4

Although serum antithyroglobulin and antithyroperoxidase antibodies are elevated in HE, their presence is thought to be an epiphenomenon of the condition rather than the direct cause. Supporting this are the facts that the incidence of encephalopathy is not increased in patients with established autoimmune thyroiditis, and the presence of antithyroid antibodies ranges from 5% to 20% in the general population.2, 5 There is also no evidence that thyroid antibodies directly react with brain tissue, and the levels of these antibodies do not correlate with either neurologic manifestations or clinical improvement.2, 4, 5 As HE has been reported in patients with euthyroidism, hypothyroidism, and hyperthyroidism (with hypothyroidismeither subclinical or activemost common), it is also unlikely that the level of thyroid hormones play a role in the etiology of this disease.2, 4, 6

The etiology and pathogenesis of HE are unclear, although an immune‐mediated process is generally implicated, either from an inflammatory vasculitis or as a form of acute disseminated encephalomyelitis.7‐9 Global hypoperfusion on single‐photon emission computed tomography (SPECT) studies has also been reported.10, 11 Patients with HE may have nonspecific evidence of inflammation, including an elevated ESR, CRP, and CSF protein.12 Other laboratory abnormalities may include a mild elevation of liver aminotransferase levels; renal impairment has also been reported in a few cases of HE in the form of glomerulonephritis, and may be related to deposition of immune complexes containing thyroglobulin antigen.6, 12‐14 MRI of the brain is normal or nonspecific in most cases, and the EEG most commonly shows diffuse slowing.

The differential for a rapidly progressive cognitive decline includes CJD, CNS vasculitis, paraneoplastic syndromes, and autoimmune and subacute infectious encephalopathies. In patients with CJD, T2‐weighted imaging may show hyperintense signals in the basal ganglia, while diffusion‐weighted sequences may reveal changes in the cortical ribbon and bilateral thalami.15 In CNS vasculitis, the imaging findings are variable and range from discrete areas of vascular infarcts to hemorrhagic lesions.16 In paraneoplastic and autoimmune encephalopathies (excluding HE), MRI often shows nonenhancing signal intensity changes in the mesial temporal lobes.12 This patient had repeatedly normal MRI studies of the brain, which in combination with the history of tremor, myoclonus, seizures, and interval return to baseline status, helped point to the diagnosis of HE.

Different approaches to treatment of HE have been recommended. As the acronym SREAT suggests, patients typically respond dramatically to high‐dose steroid therapy. Although a number of patients also improve spontaneously, up to 60% of patients experience a relapsing course and require chronic immunosuppressive agents for maintenance therapy, including long‐term steroids and azathioprine.2, 17 Treatment with plasma exchange and intravenous immune globulin have also been reported, but with mixed results.18, 19 Due to her history of multiple relapses, the patient was placed on mycophenolate mofetil for additional maintenance immunosuppression, as her corticosteroid dose was reduced due to adverse effects.

Acute mental status change is a potentially emergent situation that must be evaluated with careful history and studies to exclude life‐threatening metabolic, infectious, and vascular conditions. This patient presented similarly on 2 prior occasions, and each time her physician team evaluated what appeared to be a new onset of altered consciousness, reaching a plausible but ultimately incorrect diagnosis. The patient's third presentation was finally the charm, as her physicians learned of the repeated history of a confusional state, and in particular the return to baseline status, allowing them to create a differential that focused on etiologies of relapsing encephalopathy and make the correct diagnosis.

Physicians play an important role in improving quality improvement (QI) through clinical expertise and leadership.1, 2 The role of the physician leader is dynamic and complex, yet key competencies have been described in terms of personal commitment, professional credibility, QI behaviors and skills, and institutional linkages.3 Several characteristics have also been identified in hospitals successful in implementing QI, including shared goals for improvement, substantial administrative support, use of credible data feedback, and strong physician leadership.4 Cultivating physician leadership in hospital QI via development of these competencies is crucial for ongoing efforts in hospital QI activities.

The American Board of Internal Medicine (ABIM) developed web‐based assessment tools called Practice Improvement Modules (PIMs), as part of the maintenance of certification (MOC) program. Designed to facilitate physician involvement in QI, most PIMs target a single medical condition in the ambulatory practice,5 and involve a medical record audit performed by the physician, a patient survey, and a systems readiness survey. Physicians use the results of this data collection to perform a single test of change, and receive MOC credit when they report on the results of their intervention. Recognizing that QI activities may be different in hospital settings, the ABIM subsequently developed a Hospital‐based PIM (Hospital PIM) that allows physicians to use nationally‐approved hospital‐level performance data to complete the module. The Hospital PIM requires physicians to carry out a single test of change, and report any change in the perception of the environment supporting QI activities.

The ABIM has also begun work on potentially creating a focused pathway for hospitalists in the MOC program.6 Practice‐based learning and improvement (PBLI), systems‐based practice (SBP),7 and QI8 are core competencies of hospital medicine, and assessment of these competencies would be an important component of the new, focused MOC pathway. The Hospital PIM potentially provides an assessment methodology, thus it is important to understand the impact and value of this web‐based assessment tool.

The objective of this study is to explore the impact of the Hospital PIM on physicians participating in hospital‐based QI, including facilitators and barriers to a successful experience. We highlight several case studies to describe this impact, which can be defined as learning about QI, value‐added to practice, or an enhanced QI experience. We also describe 3 pathways suggesting how physician engagement, which is an emerging theme of our research, mediates the impact of the Hospital PIM.

Methods

A nonprobability purposive sample of physicians who completed the Hospital PIM (n = 21) as part of MOC by January 2007 was interviewed using semistructured telephone interviews. At the time of data collection, 771 physicians completed the Hospital PIM, and our sample strategically reflects equal proportions of those currently active on a QI team, as well as those who formed a QI team. Physicians were contacted via e‐mail and telephone to arrange an interview. None of the physicians contacted declined. All physicians were informed about the purpose of the study and provided consent for the ABIM to analyze and report data for purposes of understanding the feasibility of the PIM at improving practice. No physician personal identifiers were used in data analysis.

The interviews focused on four domains: reasons for choosing the module, assembly and role on any quality improvement teams, the value and satisfaction with the experience of completing the Hospital PIM, and prior experience with QI. Interviews were conducted by 1 trained member of the research team, and lasted approximately 30 minutes. Data collection was terminated when theoretical saturation was reached, or when no new data was revealed during the interviews. Interviews were audio‐recorded, and data were transcribed verbatim to facilitate analysis. Through an inductive and iterative approach, 3 members of the research team (including the interviewer) coded the data to identify themes that were consistently grounded in the data.9 These themes were subsequently discussed with a fourth researcher to maximize interrater reliability. Codes were then checked against existing literature to confirm linkages and enhance interpretation.

Results

The mean age of the participants (n = 21) was 42 years and 81% were male. Primary certificates were issued a mean of 13 years prior and completers came from a wide variety of disciplines, hospitals, and areas of expertise. Overall, the majority of physicians found the Hospital PIM to be a valuable experience (n = 17; 81%), which is similar to ABIM Hospital PIM surveillance data in which 75% of all completers said they would recommend the PIM to a colleague.

The impact and value of completing the PIM is illustrated in a variety of ways. For some, particularly those with extensive QI backgrounds, the PIM organized and broadened documentation of ongoing work. Several physicians described their utilization of the Hospital PIM as a byproduct of their hospital's existing cultural norms and interest in QI, and many Hospital PIM projects dovetailed with ongoing hospital QI activities. In these cases, even though the PIM did not stimulate new ideas, there was still value among physicians in receiving recognition for their ongoing QI activities and, perhaps more importantly, in learning by reflecting on their work.10

For others, the Hospital PIM was a catalyst to change. Similar to the role of a catalyst in a chemical reaction, our data suggest that the Hospital PIM facilitated QI by lowering the energy necessary for the process to occur. Many physicians reported the process stimulated new interest in QI, (eg, [The process] gave me some ideas about future QI projects that I plan to do, or described how the experience was an extra boost or stimulus to help change their ways.) These findings are consistent with recent findings on the use of PIMs in residency,11 which describes the Preventive Cardiology PIM as a catalyst to change.

Several physicians were surprised at how easy it was to begin and initiate a QI project, acknowledging the importance of leadership and teamwork (n = 7; 33%). In addition, many physicians highlighted reflective processes (n = 8; 38%) whereby the PIM led to an increased awareness of their clinical environment, or QI in general, including how QI can affect patient care and/or patient outcomes.

The most frequently reported facilitators to a successful PIM experience were familiarity/access to QI resources and staff, institutional support and culture of QI, and documentation of ongoing QI activities. The most frequently reported barrier (n = 9; 43%) was the time that it took to complete the module. Other barriers included a lack of institutional support or negative culture supporting QI activities, a lack of familiarity/access to QI resources and staff, and perceived irrelevance of QI activities to clinical practice.

Discussion

This study describes experiences for a small number of early‐completers of the Hospital PIM. For many, impact is described as an increased awareness of the hospital clinical environment, particularly an awareness of ongoing QI activities. For others, the primary impact was learning through an increased appreciation of the importance of QI activities and understanding of basic QI procedures (ie, interdisciplinary teamwork, enhanced communication, and documentation, buy‐in, using data). Still others described impact as an enhanced QI experience via reflection on current QI work or catalyzing change in their hospital environment. Further exploration of these findings will be important to determine the full impact of the PIM. An unanticipated finding, however, was the emerging theme of the role of physician engagement in mediating a successful experience with the Hospital PIM.

Prior research on physician engagement more generally demonstrates that increased physician engagement enhances interaction with nursing and other office staff,12 improves overall physician alignment,13 enhances QI,2 and may heighten physicians' willingness to participate in hospital administration and policy.14 Our data support these findings and further describe the importance of engagement in QI activities, whether it be through assembling and working in a team, helping analyze hospital systems, navigating existing institutional linkages, or simply becoming the creative initiative on a QI project. For completers of the Hospital PIM, engaging in any aspect of the QI process facilitates a successful PIM experience as documented by impact, and may stimulate physician leadership and hospital level change as well. Nonengaging physicians, in contrast, had a negative experience and documented little or no impact as a result of completing the PIM.

As our findings illustrate, engagement may not be a fixed construct, and may be acquired or generated through the QI process. In this context, physicians with varying levels of QI experience and expertise may learn and find value in completing the Hospital PIM, provided they become engaged with the process. Internal (ie, personal commitment, buy‐in, perceived relevance) and external (ie, hospital QI culture, access to QI team, access to data) factors may influence the degree of satisfaction and success with the Hospital PIM experience, thus maximizing facilitators and overcoming the barriers is also important for a positive outcome.

There are important limitations to this study. Most importantly, we acknowledge that the quality of the resources available between hospitals is highly variable. Therefore, our subjective assessment of whether or not someone was actively engaged is largely dependent on the quality of the available resources, and in a resource‐poor environment, this may not be a fair reflection of their engagement. We further recognize that coming to any broad or conclusive findings about the impact of the PIM is difficult given the qualitative nature of this study. However, our findings do suggest that the Hospital PIM may promote learning and value to completing physicians, especially those that engage in the QI process. Future studies should further explore the described impact and the relationship between engagement and QI.

To further enhance the Hospital PIM, consideration of prerequisite criteria for future completers, such as documentation of engagement, adequate access to hospital QI resources, and significant clinical work in the hospital setting, may be warranted. Additionally, consideration for alternative means of MOC credit may be warranted for physicians who demonstrate a proficiency in QI activities or who work in hospitals that efficiently participate in QI activities, such as a Health Maintenance Organization, for whom completion of the Hospital PIM may be redundant.

In conclusion, our findings suggest that the Hospital PIM is a useful component of MOC for appropriate groups of physicians despite the unique aspects of the Hospital PIM using hospital‐level outcomes data. Many physicians in our sample found it to be useful as a catalyst for learning about QI activities, which was facilitated through active engagement with the PIM QI process. While ongoing study is needed, it is anticipated that the findings from this study will help to inform the proposed pathway of focused practice in hospital medicine as part of MOC, particularly activities geared toward assessing competency in QI.

Physicians play an important role in improving quality improvement (QI) through clinical expertise and leadership.1, 2 The role of the physician leader is dynamic and complex, yet key competencies have been described in terms of personal commitment, professional credibility, QI behaviors and skills, and institutional linkages.3 Several characteristics have also been identified in hospitals successful in implementing QI, including shared goals for improvement, substantial administrative support, use of credible data feedback, and strong physician leadership.4 Cultivating physician leadership in hospital QI via development of these competencies is crucial for ongoing efforts in hospital QI activities.

The American Board of Internal Medicine (ABIM) developed web‐based assessment tools called Practice Improvement Modules (PIMs), as part of the maintenance of certification (MOC) program. Designed to facilitate physician involvement in QI, most PIMs target a single medical condition in the ambulatory practice,5 and involve a medical record audit performed by the physician, a patient survey, and a systems readiness survey. Physicians use the results of this data collection to perform a single test of change, and receive MOC credit when they report on the results of their intervention. Recognizing that QI activities may be different in hospital settings, the ABIM subsequently developed a Hospital‐based PIM (Hospital PIM) that allows physicians to use nationally‐approved hospital‐level performance data to complete the module. The Hospital PIM requires physicians to carry out a single test of change, and report any change in the perception of the environment supporting QI activities.

The ABIM has also begun work on potentially creating a focused pathway for hospitalists in the MOC program.6 Practice‐based learning and improvement (PBLI), systems‐based practice (SBP),7 and QI8 are core competencies of hospital medicine, and assessment of these competencies would be an important component of the new, focused MOC pathway. The Hospital PIM potentially provides an assessment methodology, thus it is important to understand the impact and value of this web‐based assessment tool.

The objective of this study is to explore the impact of the Hospital PIM on physicians participating in hospital‐based QI, including facilitators and barriers to a successful experience. We highlight several case studies to describe this impact, which can be defined as learning about QI, value‐added to practice, or an enhanced QI experience. We also describe 3 pathways suggesting how physician engagement, which is an emerging theme of our research, mediates the impact of the Hospital PIM.

Methods

A nonprobability purposive sample of physicians who completed the Hospital PIM (n = 21) as part of MOC by January 2007 was interviewed using semistructured telephone interviews. At the time of data collection, 771 physicians completed the Hospital PIM, and our sample strategically reflects equal proportions of those currently active on a QI team, as well as those who formed a QI team. Physicians were contacted via e‐mail and telephone to arrange an interview. None of the physicians contacted declined. All physicians were informed about the purpose of the study and provided consent for the ABIM to analyze and report data for purposes of understanding the feasibility of the PIM at improving practice. No physician personal identifiers were used in data analysis.

The interviews focused on four domains: reasons for choosing the module, assembly and role on any quality improvement teams, the value and satisfaction with the experience of completing the Hospital PIM, and prior experience with QI. Interviews were conducted by 1 trained member of the research team, and lasted approximately 30 minutes. Data collection was terminated when theoretical saturation was reached, or when no new data was revealed during the interviews. Interviews were audio‐recorded, and data were transcribed verbatim to facilitate analysis. Through an inductive and iterative approach, 3 members of the research team (including the interviewer) coded the data to identify themes that were consistently grounded in the data.9 These themes were subsequently discussed with a fourth researcher to maximize interrater reliability. Codes were then checked against existing literature to confirm linkages and enhance interpretation.

Results

The mean age of the participants (n = 21) was 42 years and 81% were male. Primary certificates were issued a mean of 13 years prior and completers came from a wide variety of disciplines, hospitals, and areas of expertise. Overall, the majority of physicians found the Hospital PIM to be a valuable experience (n = 17; 81%), which is similar to ABIM Hospital PIM surveillance data in which 75% of all completers said they would recommend the PIM to a colleague.

The impact and value of completing the PIM is illustrated in a variety of ways. For some, particularly those with extensive QI backgrounds, the PIM organized and broadened documentation of ongoing work. Several physicians described their utilization of the Hospital PIM as a byproduct of their hospital's existing cultural norms and interest in QI, and many Hospital PIM projects dovetailed with ongoing hospital QI activities. In these cases, even though the PIM did not stimulate new ideas, there was still value among physicians in receiving recognition for their ongoing QI activities and, perhaps more importantly, in learning by reflecting on their work.10

For others, the Hospital PIM was a catalyst to change. Similar to the role of a catalyst in a chemical reaction, our data suggest that the Hospital PIM facilitated QI by lowering the energy necessary for the process to occur. Many physicians reported the process stimulated new interest in QI, (eg, [The process] gave me some ideas about future QI projects that I plan to do, or described how the experience was an extra boost or stimulus to help change their ways.) These findings are consistent with recent findings on the use of PIMs in residency,11 which describes the Preventive Cardiology PIM as a catalyst to change.

Several physicians were surprised at how easy it was to begin and initiate a QI project, acknowledging the importance of leadership and teamwork (n = 7; 33%). In addition, many physicians highlighted reflective processes (n = 8; 38%) whereby the PIM led to an increased awareness of their clinical environment, or QI in general, including how QI can affect patient care and/or patient outcomes.

The most frequently reported facilitators to a successful PIM experience were familiarity/access to QI resources and staff, institutional support and culture of QI, and documentation of ongoing QI activities. The most frequently reported barrier (n = 9; 43%) was the time that it took to complete the module. Other barriers included a lack of institutional support or negative culture supporting QI activities, a lack of familiarity/access to QI resources and staff, and perceived irrelevance of QI activities to clinical practice.

Discussion

This study describes experiences for a small number of early‐completers of the Hospital PIM. For many, impact is described as an increased awareness of the hospital clinical environment, particularly an awareness of ongoing QI activities. For others, the primary impact was learning through an increased appreciation of the importance of QI activities and understanding of basic QI procedures (ie, interdisciplinary teamwork, enhanced communication, and documentation, buy‐in, using data). Still others described impact as an enhanced QI experience via reflection on current QI work or catalyzing change in their hospital environment. Further exploration of these findings will be important to determine the full impact of the PIM. An unanticipated finding, however, was the emerging theme of the role of physician engagement in mediating a successful experience with the Hospital PIM.

Prior research on physician engagement more generally demonstrates that increased physician engagement enhances interaction with nursing and other office staff,12 improves overall physician alignment,13 enhances QI,2 and may heighten physicians' willingness to participate in hospital administration and policy.14 Our data support these findings and further describe the importance of engagement in QI activities, whether it be through assembling and working in a team, helping analyze hospital systems, navigating existing institutional linkages, or simply becoming the creative initiative on a QI project. For completers of the Hospital PIM, engaging in any aspect of the QI process facilitates a successful PIM experience as documented by impact, and may stimulate physician leadership and hospital level change as well. Nonengaging physicians, in contrast, had a negative experience and documented little or no impact as a result of completing the PIM.

As our findings illustrate, engagement may not be a fixed construct, and may be acquired or generated through the QI process. In this context, physicians with varying levels of QI experience and expertise may learn and find value in completing the Hospital PIM, provided they become engaged with the process. Internal (ie, personal commitment, buy‐in, perceived relevance) and external (ie, hospital QI culture, access to QI team, access to data) factors may influence the degree of satisfaction and success with the Hospital PIM experience, thus maximizing facilitators and overcoming the barriers is also important for a positive outcome.

There are important limitations to this study. Most importantly, we acknowledge that the quality of the resources available between hospitals is highly variable. Therefore, our subjective assessment of whether or not someone was actively engaged is largely dependent on the quality of the available resources, and in a resource‐poor environment, this may not be a fair reflection of their engagement. We further recognize that coming to any broad or conclusive findings about the impact of the PIM is difficult given the qualitative nature of this study. However, our findings do suggest that the Hospital PIM may promote learning and value to completing physicians, especially those that engage in the QI process. Future studies should further explore the described impact and the relationship between engagement and QI.

To further enhance the Hospital PIM, consideration of prerequisite criteria for future completers, such as documentation of engagement, adequate access to hospital QI resources, and significant clinical work in the hospital setting, may be warranted. Additionally, consideration for alternative means of MOC credit may be warranted for physicians who demonstrate a proficiency in QI activities or who work in hospitals that efficiently participate in QI activities, such as a Health Maintenance Organization, for whom completion of the Hospital PIM may be redundant.

In conclusion, our findings suggest that the Hospital PIM is a useful component of MOC for appropriate groups of physicians despite the unique aspects of the Hospital PIM using hospital‐level outcomes data. Many physicians in our sample found it to be useful as a catalyst for learning about QI activities, which was facilitated through active engagement with the PIM QI process. While ongoing study is needed, it is anticipated that the findings from this study will help to inform the proposed pathway of focused practice in hospital medicine as part of MOC, particularly activities geared toward assessing competency in QI.

Physicians play an important role in improving quality improvement (QI) through clinical expertise and leadership.1, 2 The role of the physician leader is dynamic and complex, yet key competencies have been described in terms of personal commitment, professional credibility, QI behaviors and skills, and institutional linkages.3 Several characteristics have also been identified in hospitals successful in implementing QI, including shared goals for improvement, substantial administrative support, use of credible data feedback, and strong physician leadership.4 Cultivating physician leadership in hospital QI via development of these competencies is crucial for ongoing efforts in hospital QI activities.

The American Board of Internal Medicine (ABIM) developed web‐based assessment tools called Practice Improvement Modules (PIMs), as part of the maintenance of certification (MOC) program. Designed to facilitate physician involvement in QI, most PIMs target a single medical condition in the ambulatory practice,5 and involve a medical record audit performed by the physician, a patient survey, and a systems readiness survey. Physicians use the results of this data collection to perform a single test of change, and receive MOC credit when they report on the results of their intervention. Recognizing that QI activities may be different in hospital settings, the ABIM subsequently developed a Hospital‐based PIM (Hospital PIM) that allows physicians to use nationally‐approved hospital‐level performance data to complete the module. The Hospital PIM requires physicians to carry out a single test of change, and report any change in the perception of the environment supporting QI activities.

The ABIM has also begun work on potentially creating a focused pathway for hospitalists in the MOC program.6 Practice‐based learning and improvement (PBLI), systems‐based practice (SBP),7 and QI8 are core competencies of hospital medicine, and assessment of these competencies would be an important component of the new, focused MOC pathway. The Hospital PIM potentially provides an assessment methodology, thus it is important to understand the impact and value of this web‐based assessment tool.

The objective of this study is to explore the impact of the Hospital PIM on physicians participating in hospital‐based QI, including facilitators and barriers to a successful experience. We highlight several case studies to describe this impact, which can be defined as learning about QI, value‐added to practice, or an enhanced QI experience. We also describe 3 pathways suggesting how physician engagement, which is an emerging theme of our research, mediates the impact of the Hospital PIM.

Methods

A nonprobability purposive sample of physicians who completed the Hospital PIM (n = 21) as part of MOC by January 2007 was interviewed using semistructured telephone interviews. At the time of data collection, 771 physicians completed the Hospital PIM, and our sample strategically reflects equal proportions of those currently active on a QI team, as well as those who formed a QI team. Physicians were contacted via e‐mail and telephone to arrange an interview. None of the physicians contacted declined. All physicians were informed about the purpose of the study and provided consent for the ABIM to analyze and report data for purposes of understanding the feasibility of the PIM at improving practice. No physician personal identifiers were used in data analysis.

The interviews focused on four domains: reasons for choosing the module, assembly and role on any quality improvement teams, the value and satisfaction with the experience of completing the Hospital PIM, and prior experience with QI. Interviews were conducted by 1 trained member of the research team, and lasted approximately 30 minutes. Data collection was terminated when theoretical saturation was reached, or when no new data was revealed during the interviews. Interviews were audio‐recorded, and data were transcribed verbatim to facilitate analysis. Through an inductive and iterative approach, 3 members of the research team (including the interviewer) coded the data to identify themes that were consistently grounded in the data.9 These themes were subsequently discussed with a fourth researcher to maximize interrater reliability. Codes were then checked against existing literature to confirm linkages and enhance interpretation.

Results

The mean age of the participants (n = 21) was 42 years and 81% were male. Primary certificates were issued a mean of 13 years prior and completers came from a wide variety of disciplines, hospitals, and areas of expertise. Overall, the majority of physicians found the Hospital PIM to be a valuable experience (n = 17; 81%), which is similar to ABIM Hospital PIM surveillance data in which 75% of all completers said they would recommend the PIM to a colleague.

The impact and value of completing the PIM is illustrated in a variety of ways. For some, particularly those with extensive QI backgrounds, the PIM organized and broadened documentation of ongoing work. Several physicians described their utilization of the Hospital PIM as a byproduct of their hospital's existing cultural norms and interest in QI, and many Hospital PIM projects dovetailed with ongoing hospital QI activities. In these cases, even though the PIM did not stimulate new ideas, there was still value among physicians in receiving recognition for their ongoing QI activities and, perhaps more importantly, in learning by reflecting on their work.10

For others, the Hospital PIM was a catalyst to change. Similar to the role of a catalyst in a chemical reaction, our data suggest that the Hospital PIM facilitated QI by lowering the energy necessary for the process to occur. Many physicians reported the process stimulated new interest in QI, (eg, [The process] gave me some ideas about future QI projects that I plan to do, or described how the experience was an extra boost or stimulus to help change their ways.) These findings are consistent with recent findings on the use of PIMs in residency,11 which describes the Preventive Cardiology PIM as a catalyst to change.

Several physicians were surprised at how easy it was to begin and initiate a QI project, acknowledging the importance of leadership and teamwork (n = 7; 33%). In addition, many physicians highlighted reflective processes (n = 8; 38%) whereby the PIM led to an increased awareness of their clinical environment, or QI in general, including how QI can affect patient care and/or patient outcomes.

The most frequently reported facilitators to a successful PIM experience were familiarity/access to QI resources and staff, institutional support and culture of QI, and documentation of ongoing QI activities. The most frequently reported barrier (n = 9; 43%) was the time that it took to complete the module. Other barriers included a lack of institutional support or negative culture supporting QI activities, a lack of familiarity/access to QI resources and staff, and perceived irrelevance of QI activities to clinical practice.

Discussion

This study describes experiences for a small number of early‐completers of the Hospital PIM. For many, impact is described as an increased awareness of the hospital clinical environment, particularly an awareness of ongoing QI activities. For others, the primary impact was learning through an increased appreciation of the importance of QI activities and understanding of basic QI procedures (ie, interdisciplinary teamwork, enhanced communication, and documentation, buy‐in, using data). Still others described impact as an enhanced QI experience via reflection on current QI work or catalyzing change in their hospital environment. Further exploration of these findings will be important to determine the full impact of the PIM. An unanticipated finding, however, was the emerging theme of the role of physician engagement in mediating a successful experience with the Hospital PIM.

Prior research on physician engagement more generally demonstrates that increased physician engagement enhances interaction with nursing and other office staff,12 improves overall physician alignment,13 enhances QI,2 and may heighten physicians' willingness to participate in hospital administration and policy.14 Our data support these findings and further describe the importance of engagement in QI activities, whether it be through assembling and working in a team, helping analyze hospital systems, navigating existing institutional linkages, or simply becoming the creative initiative on a QI project. For completers of the Hospital PIM, engaging in any aspect of the QI process facilitates a successful PIM experience as documented by impact, and may stimulate physician leadership and hospital level change as well. Nonengaging physicians, in contrast, had a negative experience and documented little or no impact as a result of completing the PIM.

As our findings illustrate, engagement may not be a fixed construct, and may be acquired or generated through the QI process. In this context, physicians with varying levels of QI experience and expertise may learn and find value in completing the Hospital PIM, provided they become engaged with the process. Internal (ie, personal commitment, buy‐in, perceived relevance) and external (ie, hospital QI culture, access to QI team, access to data) factors may influence the degree of satisfaction and success with the Hospital PIM experience, thus maximizing facilitators and overcoming the barriers is also important for a positive outcome.

There are important limitations to this study. Most importantly, we acknowledge that the quality of the resources available between hospitals is highly variable. Therefore, our subjective assessment of whether or not someone was actively engaged is largely dependent on the quality of the available resources, and in a resource‐poor environment, this may not be a fair reflection of their engagement. We further recognize that coming to any broad or conclusive findings about the impact of the PIM is difficult given the qualitative nature of this study. However, our findings do suggest that the Hospital PIM may promote learning and value to completing physicians, especially those that engage in the QI process. Future studies should further explore the described impact and the relationship between engagement and QI.

To further enhance the Hospital PIM, consideration of prerequisite criteria for future completers, such as documentation of engagement, adequate access to hospital QI resources, and significant clinical work in the hospital setting, may be warranted. Additionally, consideration for alternative means of MOC credit may be warranted for physicians who demonstrate a proficiency in QI activities or who work in hospitals that efficiently participate in QI activities, such as a Health Maintenance Organization, for whom completion of the Hospital PIM may be redundant.

In conclusion, our findings suggest that the Hospital PIM is a useful component of MOC for appropriate groups of physicians despite the unique aspects of the Hospital PIM using hospital‐level outcomes data. Many physicians in our sample found it to be useful as a catalyst for learning about QI activities, which was facilitated through active engagement with the PIM QI process. While ongoing study is needed, it is anticipated that the findings from this study will help to inform the proposed pathway of focused practice in hospital medicine as part of MOC, particularly activities geared toward assessing competency in QI.

In recent years, medical educators have recognized the importance of the inclusion of patient‐centered care in the medical school curriculum.1 There is an increased awareness of the importance of patient involvement in medical decision‐making, as well as a realization that patient‐centered care positively affects patient satisfaction and outcomes measures.2 The Institute of Medicine, in its 2001 report Crossing the Quality Chasm: A New Health System for the 21st Century, included patient‐centered care as one of the areas for the development of quality measures, defining it as providing care that is respectful of and responsive to individual patient preferences, needs, and values and ensuring that patient values guide all clinical decisions.3 Several organizations, such as the Institute for Healthcare Improvement,4 have initiatives related to achieving the goals of patient‐centered care. While not a new phenomenon, patient‐centered care is permeating many areas of the healthcare delivery system.5 The recognition of patient‐centered care as both a desirable and measurable outcome of the healthcare enterprise has also renewed interest in the field of medical humanities as a valid tool for the advancement of patient‐centered initiatives and goals.

The Basis for Patient‐centered Care

Stewart et al.,6 in their book Patient‐Centered Medicine: Transforming the Clinical Method, identify 6 essential components of the patient‐centered clinical method: exploration of the disease and illness experience; understanding of the whole person; finding common ground; incorporating prevention and health promotion; enhancing the patient‐doctor relationship; and being realistic. These recommendations seek to improve the patient‐physician relationship by empowering the patient to be an active participant in his/her own health care.

The American Academy of Pediatrics recently released a policy statement7 outlining the benefits of family‐centered care in patient‐family outcomes, as well as in staff satisfaction. The statement also highlights the importance of bedside rounding by the attending physician and the healthcare team. Bedside rounding involves the patient in management discussions and decision making, while allowing for the unfiltered exchange of information. Nurses, therapists, and ancillary staff involved in the care of the patient also participate in the presentation. After the presentation, goals for the hospitalization are established, and the patient and family are asked for permission to implement the plan of care. Educational discussions regarding the patient's diagnosis usually take place outside the room, unless the patient's physical exam warrants a bedside teaching moment. Regardless of the format, patient‐involvement in the decision process is the central objective.

The Basis for the Medical Humanities

At the same time, there is renewed interest in the inclusion of the humanities in medicine. There is a perceived gap between the technological emphasis of the current medical school curriculum and the human values integral to the patient‐physician relationship.8 Namely, there is a growing concern that medical technology has suffused medical education with a sort of trade mentality in which doctors are trained in the latest scientific medical breakthroughs without the proper contextualization of the patient as the center of the healthcare enterprise.9 The National Endowment for the Humanities, the Association of American Medical Colleges, the Accreditation Council on Graduate Medical Education, and the Society for Health and Human Values have all called for increased emphasis of the humanities in medical education.10

What are the medical humanities? There is no clear‐cut definition. Felice Aull11 correlates the term with various so‐called liberal arts disciplines and their application to medical education and practice. She develops the notion that the medical humanities contribute to medical education in areas pertinent to patient‐centered care: insight into the human condition, development of observational and analytical skills, development of empathy and self‐reflection, and intercultural understanding.

In general, the medical humanities provide broad educational perspectives, and allow learners to develop skills critical to the development of a humane approach to patients.12 The paradigm relies on the assumption that exposure to the humanities in medical schoolin the form of formal lectures dealing with topics such as philosophy and literature, or through the role‐modeling interactions of teaching physicians and their perceived empathy to their patientswill allow the students to become more humane, and therefore, better doctors.13 The medical humanities seek to equip doctors with the critical thinking incumbent to the conversation about human values in a scientific field, and to explore questions of value and purpose critical in the medical setting.14

Shared Values

What are the values associated with the medical humanities that make them ideal for the teaching of patient‐centered care? Lester Friedman15 delineates 2 domains pertaining to the intrinsic values of the medical humanities. He identifies an affective domain, corresponding to a loose interpretation of the traditional affective perspective identified in the patient‐physician relationship; and a cognitive domain, a refocusing of medicine to its so‐called traditional professional roots, contrasted with the trade mentality of some in the profession today. Donnie Self16 identifies 2 currents of thought regarding the medical humanities: the affective approach, related to the development of compassion, sensitivity, empathy between the patients and their care providers; and the cognitive approach, which pertains to the development of logical and critical thinking required by medical education. These correspond to the integral attributes of patient‐centered care: incorporating the patients' ideas and affective responses to their illness; and establishing common ground and goals agreed upon by both patient and physician.17

The medical humanities are used to address such patient‐centered issues as end‐of‐life care in children,18 the physician‐patient narrative interaction,19 and the patients' role in their own health care.20 Medical humanities courses are also used in the training of cultural competence.21 Of course, the best‐known contribution of the medical humanities to patient‐centered care is the continuing importance of bioethics programs and their interrelation with other humanities fields.22

One of the goals of patient‐centered care is the elimination of perceived barriers of communication between patient and healthcare providers, in order to create a partnership aimed at improving healthcare outcomes.2 Since one of the fundamental aspects of medical training is learning the language of medicine,23 enhancing the communication skills of future physicians is part of the educational goals pursued by medical humanities programs.24 Language and its applicationsfor example, courses on medical interviewing or narrative medicineserve as the link between patient care and the medical humanities. Effective communication with patients is a measurable predictor of patient satisfaction, patient outcomes, and occurrences of malpractice litigation.17 For example, a study examining communication behaviors between physicians and the occurrence of malpractice claims25 found that doctors who were not sued spent more time with patients, educating patients about what to expect, and asking patients their understanding and opinion of the situation. Another study26 demonstrated that a patient‐focused approach improved the management of asthma, decreasing emergency room visits and hospitalizations. Therefore, it is not surprising that the Institute of Medicine's Committee on Behavioral and Social Sciences in Medical School Curricula identified basic and complex communications skills as priorities for inclusion in the medical curriculum.27

Doubts About the Process

There are doubts as to whether the medical humanities can really instill humanistic qualities in doctors. There are also questions about the physician‐centric focus of the medical humanities.28 This physician‐centric attitude runs counter to the intent of the medical humanities. Edmund Pellegrino and David Thomasma29 define the patient‐physician interaction as a human relationship where 1 person in need of healing seeks out another who professes to heal, or to assist in healing. The act of medicine ties these 2 persons together. While acknowledging the basic imbalance of the physician‐patient relationship, Pellegrino and Thomasma29 strive to close the gap by establishing medicine as a relation of mutual consent to effect individualized well‐being by working in, with, and through the body. The individualized exercise of well‐being, framed in, with, and through the body of the patient is similar to the description used by Stewart et al.6 of the patient as the unit of analysis delineating patient‐centered care, as it incorporates the interactive components proposed for a successful patient‐centered interaction.

There is also confusion between the teaching of humanities in medical schoolfor example, courses in history of medicine, narrative medicine, and medicine and the artsand the attempt to train humanistic physicians.30 Although an examination of humanities texts is certainly useful, the focus of the teaching of the medical humanities should evolve beyond a simple lucubration based on liberal studies, to a focused interaction between patient and physician, and a recentralization of the patient as the focus of that relationship.31

Conclusions

There is general agreement that a humane doctor is a better doctor. There is less agreement on how to measure the impact of a humanities education, as a qualitative assessment of satisfactory health care.19, 22, 25 There has been great growth in the teaching of medical humanities in medical schools. Most of the focus has been on the inclusion of humanities textssuch as literary, philosophical, and historical documentsas tools to establish a correlation between the arts and medicine, in hopes that the clarification of such association will provide medical students a broad‐based assessment, a so‐called world‐view, from which they can become introspective and humanistic when faced with their patients.³² Although this is a desirable goal, the driving force behind the medical humanities should shift to a quantifiable, evidence‐based assessment of its goals. A tool to achieve this verification is through the process of patient‐centered care. There is evidence to suggest patient‐centered care improves satisfaction and outcomes measures. It also refocuses care on the patient, which is the same goal of the medical humanities. By focusing on the patient, instead of the physician, the medical humanities will gain verification and validation within the academic healthcare environment.

In recent years, medical educators have recognized the importance of the inclusion of patient‐centered care in the medical school curriculum.1 There is an increased awareness of the importance of patient involvement in medical decision‐making, as well as a realization that patient‐centered care positively affects patient satisfaction and outcomes measures.2 The Institute of Medicine, in its 2001 report Crossing the Quality Chasm: A New Health System for the 21st Century, included patient‐centered care as one of the areas for the development of quality measures, defining it as providing care that is respectful of and responsive to individual patient preferences, needs, and values and ensuring that patient values guide all clinical decisions.3 Several organizations, such as the Institute for Healthcare Improvement,4 have initiatives related to achieving the goals of patient‐centered care. While not a new phenomenon, patient‐centered care is permeating many areas of the healthcare delivery system.5 The recognition of patient‐centered care as both a desirable and measurable outcome of the healthcare enterprise has also renewed interest in the field of medical humanities as a valid tool for the advancement of patient‐centered initiatives and goals.

The Basis for Patient‐centered Care

Stewart et al.,6 in their book Patient‐Centered Medicine: Transforming the Clinical Method, identify 6 essential components of the patient‐centered clinical method: exploration of the disease and illness experience; understanding of the whole person; finding common ground; incorporating prevention and health promotion; enhancing the patient‐doctor relationship; and being realistic. These recommendations seek to improve the patient‐physician relationship by empowering the patient to be an active participant in his/her own health care.

The American Academy of Pediatrics recently released a policy statement7 outlining the benefits of family‐centered care in patient‐family outcomes, as well as in staff satisfaction. The statement also highlights the importance of bedside rounding by the attending physician and the healthcare team. Bedside rounding involves the patient in management discussions and decision making, while allowing for the unfiltered exchange of information. Nurses, therapists, and ancillary staff involved in the care of the patient also participate in the presentation. After the presentation, goals for the hospitalization are established, and the patient and family are asked for permission to implement the plan of care. Educational discussions regarding the patient's diagnosis usually take place outside the room, unless the patient's physical exam warrants a bedside teaching moment. Regardless of the format, patient‐involvement in the decision process is the central objective.

The Basis for the Medical Humanities

At the same time, there is renewed interest in the inclusion of the humanities in medicine. There is a perceived gap between the technological emphasis of the current medical school curriculum and the human values integral to the patient‐physician relationship.8 Namely, there is a growing concern that medical technology has suffused medical education with a sort of trade mentality in which doctors are trained in the latest scientific medical breakthroughs without the proper contextualization of the patient as the center of the healthcare enterprise.9 The National Endowment for the Humanities, the Association of American Medical Colleges, the Accreditation Council on Graduate Medical Education, and the Society for Health and Human Values have all called for increased emphasis of the humanities in medical education.10

What are the medical humanities? There is no clear‐cut definition. Felice Aull11 correlates the term with various so‐called liberal arts disciplines and their application to medical education and practice. She develops the notion that the medical humanities contribute to medical education in areas pertinent to patient‐centered care: insight into the human condition, development of observational and analytical skills, development of empathy and self‐reflection, and intercultural understanding.

In general, the medical humanities provide broad educational perspectives, and allow learners to develop skills critical to the development of a humane approach to patients.12 The paradigm relies on the assumption that exposure to the humanities in medical schoolin the form of formal lectures dealing with topics such as philosophy and literature, or through the role‐modeling interactions of teaching physicians and their perceived empathy to their patientswill allow the students to become more humane, and therefore, better doctors.13 The medical humanities seek to equip doctors with the critical thinking incumbent to the conversation about human values in a scientific field, and to explore questions of value and purpose critical in the medical setting.14

Shared Values

What are the values associated with the medical humanities that make them ideal for the teaching of patient‐centered care? Lester Friedman15 delineates 2 domains pertaining to the intrinsic values of the medical humanities. He identifies an affective domain, corresponding to a loose interpretation of the traditional affective perspective identified in the patient‐physician relationship; and a cognitive domain, a refocusing of medicine to its so‐called traditional professional roots, contrasted with the trade mentality of some in the profession today. Donnie Self16 identifies 2 currents of thought regarding the medical humanities: the affective approach, related to the development of compassion, sensitivity, empathy between the patients and their care providers; and the cognitive approach, which pertains to the development of logical and critical thinking required by medical education. These correspond to the integral attributes of patient‐centered care: incorporating the patients' ideas and affective responses to their illness; and establishing common ground and goals agreed upon by both patient and physician.17

The medical humanities are used to address such patient‐centered issues as end‐of‐life care in children,18 the physician‐patient narrative interaction,19 and the patients' role in their own health care.20 Medical humanities courses are also used in the training of cultural competence.21 Of course, the best‐known contribution of the medical humanities to patient‐centered care is the continuing importance of bioethics programs and their interrelation with other humanities fields.22

One of the goals of patient‐centered care is the elimination of perceived barriers of communication between patient and healthcare providers, in order to create a partnership aimed at improving healthcare outcomes.2 Since one of the fundamental aspects of medical training is learning the language of medicine,23 enhancing the communication skills of future physicians is part of the educational goals pursued by medical humanities programs.24 Language and its applicationsfor example, courses on medical interviewing or narrative medicineserve as the link between patient care and the medical humanities. Effective communication with patients is a measurable predictor of patient satisfaction, patient outcomes, and occurrences of malpractice litigation.17 For example, a study examining communication behaviors between physicians and the occurrence of malpractice claims25 found that doctors who were not sued spent more time with patients, educating patients about what to expect, and asking patients their understanding and opinion of the situation. Another study26 demonstrated that a patient‐focused approach improved the management of asthma, decreasing emergency room visits and hospitalizations. Therefore, it is not surprising that the Institute of Medicine's Committee on Behavioral and Social Sciences in Medical School Curricula identified basic and complex communications skills as priorities for inclusion in the medical curriculum.27

Doubts About the Process

There are doubts as to whether the medical humanities can really instill humanistic qualities in doctors. There are also questions about the physician‐centric focus of the medical humanities.28 This physician‐centric attitude runs counter to the intent of the medical humanities. Edmund Pellegrino and David Thomasma29 define the patient‐physician interaction as a human relationship where 1 person in need of healing seeks out another who professes to heal, or to assist in healing. The act of medicine ties these 2 persons together. While acknowledging the basic imbalance of the physician‐patient relationship, Pellegrino and Thomasma29 strive to close the gap by establishing medicine as a relation of mutual consent to effect individualized well‐being by working in, with, and through the body. The individualized exercise of well‐being, framed in, with, and through the body of the patient is similar to the description used by Stewart et al.6 of the patient as the unit of analysis delineating patient‐centered care, as it incorporates the interactive components proposed for a successful patient‐centered interaction.

There is also confusion between the teaching of humanities in medical schoolfor example, courses in history of medicine, narrative medicine, and medicine and the artsand the attempt to train humanistic physicians.30 Although an examination of humanities texts is certainly useful, the focus of the teaching of the medical humanities should evolve beyond a simple lucubration based on liberal studies, to a focused interaction between patient and physician, and a recentralization of the patient as the focus of that relationship.31

Conclusions

There is general agreement that a humane doctor is a better doctor. There is less agreement on how to measure the impact of a humanities education, as a qualitative assessment of satisfactory health care.19, 22, 25 There has been great growth in the teaching of medical humanities in medical schools. Most of the focus has been on the inclusion of humanities textssuch as literary, philosophical, and historical documentsas tools to establish a correlation between the arts and medicine, in hopes that the clarification of such association will provide medical students a broad‐based assessment, a so‐called world‐view, from which they can become introspective and humanistic when faced with their patients.³² Although this is a desirable goal, the driving force behind the medical humanities should shift to a quantifiable, evidence‐based assessment of its goals. A tool to achieve this verification is through the process of patient‐centered care. There is evidence to suggest patient‐centered care improves satisfaction and outcomes measures. It also refocuses care on the patient, which is the same goal of the medical humanities. By focusing on the patient, instead of the physician, the medical humanities will gain verification and validation within the academic healthcare environment.

In recent years, medical educators have recognized the importance of the inclusion of patient‐centered care in the medical school curriculum.1 There is an increased awareness of the importance of patient involvement in medical decision‐making, as well as a realization that patient‐centered care positively affects patient satisfaction and outcomes measures.2 The Institute of Medicine, in its 2001 report Crossing the Quality Chasm: A New Health System for the 21st Century, included patient‐centered care as one of the areas for the development of quality measures, defining it as providing care that is respectful of and responsive to individual patient preferences, needs, and values and ensuring that patient values guide all clinical decisions.3 Several organizations, such as the Institute for Healthcare Improvement,4 have initiatives related to achieving the goals of patient‐centered care. While not a new phenomenon, patient‐centered care is permeating many areas of the healthcare delivery system.5 The recognition of patient‐centered care as both a desirable and measurable outcome of the healthcare enterprise has also renewed interest in the field of medical humanities as a valid tool for the advancement of patient‐centered initiatives and goals.

The Basis for Patient‐centered Care

Stewart et al.,6 in their book Patient‐Centered Medicine: Transforming the Clinical Method, identify 6 essential components of the patient‐centered clinical method: exploration of the disease and illness experience; understanding of the whole person; finding common ground; incorporating prevention and health promotion; enhancing the patient‐doctor relationship; and being realistic. These recommendations seek to improve the patient‐physician relationship by empowering the patient to be an active participant in his/her own health care.

The American Academy of Pediatrics recently released a policy statement7 outlining the benefits of family‐centered care in patient‐family outcomes, as well as in staff satisfaction. The statement also highlights the importance of bedside rounding by the attending physician and the healthcare team. Bedside rounding involves the patient in management discussions and decision making, while allowing for the unfiltered exchange of information. Nurses, therapists, and ancillary staff involved in the care of the patient also participate in the presentation. After the presentation, goals for the hospitalization are established, and the patient and family are asked for permission to implement the plan of care. Educational discussions regarding the patient's diagnosis usually take place outside the room, unless the patient's physical exam warrants a bedside teaching moment. Regardless of the format, patient‐involvement in the decision process is the central objective.

The Basis for the Medical Humanities

At the same time, there is renewed interest in the inclusion of the humanities in medicine. There is a perceived gap between the technological emphasis of the current medical school curriculum and the human values integral to the patient‐physician relationship.8 Namely, there is a growing concern that medical technology has suffused medical education with a sort of trade mentality in which doctors are trained in the latest scientific medical breakthroughs without the proper contextualization of the patient as the center of the healthcare enterprise.9 The National Endowment for the Humanities, the Association of American Medical Colleges, the Accreditation Council on Graduate Medical Education, and the Society for Health and Human Values have all called for increased emphasis of the humanities in medical education.10

What are the medical humanities? There is no clear‐cut definition. Felice Aull11 correlates the term with various so‐called liberal arts disciplines and their application to medical education and practice. She develops the notion that the medical humanities contribute to medical education in areas pertinent to patient‐centered care: insight into the human condition, development of observational and analytical skills, development of empathy and self‐reflection, and intercultural understanding.

In general, the medical humanities provide broad educational perspectives, and allow learners to develop skills critical to the development of a humane approach to patients.12 The paradigm relies on the assumption that exposure to the humanities in medical schoolin the form of formal lectures dealing with topics such as philosophy and literature, or through the role‐modeling interactions of teaching physicians and their perceived empathy to their patientswill allow the students to become more humane, and therefore, better doctors.13 The medical humanities seek to equip doctors with the critical thinking incumbent to the conversation about human values in a scientific field, and to explore questions of value and purpose critical in the medical setting.14

Shared Values

What are the values associated with the medical humanities that make them ideal for the teaching of patient‐centered care? Lester Friedman15 delineates 2 domains pertaining to the intrinsic values of the medical humanities. He identifies an affective domain, corresponding to a loose interpretation of the traditional affective perspective identified in the patient‐physician relationship; and a cognitive domain, a refocusing of medicine to its so‐called traditional professional roots, contrasted with the trade mentality of some in the profession today. Donnie Self16 identifies 2 currents of thought regarding the medical humanities: the affective approach, related to the development of compassion, sensitivity, empathy between the patients and their care providers; and the cognitive approach, which pertains to the development of logical and critical thinking required by medical education. These correspond to the integral attributes of patient‐centered care: incorporating the patients' ideas and affective responses to their illness; and establishing common ground and goals agreed upon by both patient and physician.17

The medical humanities are used to address such patient‐centered issues as end‐of‐life care in children,18 the physician‐patient narrative interaction,19 and the patients' role in their own health care.20 Medical humanities courses are also used in the training of cultural competence.21 Of course, the best‐known contribution of the medical humanities to patient‐centered care is the continuing importance of bioethics programs and their interrelation with other humanities fields.22

One of the goals of patient‐centered care is the elimination of perceived barriers of communication between patient and healthcare providers, in order to create a partnership aimed at improving healthcare outcomes.2 Since one of the fundamental aspects of medical training is learning the language of medicine,23 enhancing the communication skills of future physicians is part of the educational goals pursued by medical humanities programs.24 Language and its applicationsfor example, courses on medical interviewing or narrative medicineserve as the link between patient care and the medical humanities. Effective communication with patients is a measurable predictor of patient satisfaction, patient outcomes, and occurrences of malpractice litigation.17 For example, a study examining communication behaviors between physicians and the occurrence of malpractice claims25 found that doctors who were not sued spent more time with patients, educating patients about what to expect, and asking patients their understanding and opinion of the situation. Another study26 demonstrated that a patient‐focused approach improved the management of asthma, decreasing emergency room visits and hospitalizations. Therefore, it is not surprising that the Institute of Medicine's Committee on Behavioral and Social Sciences in Medical School Curricula identified basic and complex communications skills as priorities for inclusion in the medical curriculum.27

Doubts About the Process

There are doubts as to whether the medical humanities can really instill humanistic qualities in doctors. There are also questions about the physician‐centric focus of the medical humanities.28 This physician‐centric attitude runs counter to the intent of the medical humanities. Edmund Pellegrino and David Thomasma29 define the patient‐physician interaction as a human relationship where 1 person in need of healing seeks out another who professes to heal, or to assist in healing. The act of medicine ties these 2 persons together. While acknowledging the basic imbalance of the physician‐patient relationship, Pellegrino and Thomasma29 strive to close the gap by establishing medicine as a relation of mutual consent to effect individualized well‐being by working in, with, and through the body. The individualized exercise of well‐being, framed in, with, and through the body of the patient is similar to the description used by Stewart et al.6 of the patient as the unit of analysis delineating patient‐centered care, as it incorporates the interactive components proposed for a successful patient‐centered interaction.

There is also confusion between the teaching of humanities in medical schoolfor example, courses in history of medicine, narrative medicine, and medicine and the artsand the attempt to train humanistic physicians.30 Although an examination of humanities texts is certainly useful, the focus of the teaching of the medical humanities should evolve beyond a simple lucubration based on liberal studies, to a focused interaction between patient and physician, and a recentralization of the patient as the focus of that relationship.31

Conclusions

There is general agreement that a humane doctor is a better doctor. There is less agreement on how to measure the impact of a humanities education, as a qualitative assessment of satisfactory health care.19, 22, 25 There has been great growth in the teaching of medical humanities in medical schools. Most of the focus has been on the inclusion of humanities textssuch as literary, philosophical, and historical documentsas tools to establish a correlation between the arts and medicine, in hopes that the clarification of such association will provide medical students a broad‐based assessment, a so‐called world‐view, from which they can become introspective and humanistic when faced with their patients.³² Although this is a desirable goal, the driving force behind the medical humanities should shift to a quantifiable, evidence‐based assessment of its goals. A tool to achieve this verification is through the process of patient‐centered care. There is evidence to suggest patient‐centered care improves satisfaction and outcomes measures. It also refocuses care on the patient, which is the same goal of the medical humanities. By focusing on the patient, instead of the physician, the medical humanities will gain verification and validation within the academic healthcare environment.

Since contrast nephropathy (CN) was recognized more than 50 years ago,1 there have been continuous efforts to chemically modify radiocontrast agents to be less nephrotoxic. Although radiocontrast media have indeed become safer, which reduces the likelihood of CN per procedure, the indications for radiocontrast administration have dramatically increased, since over 80 million doses are delivered in the world annually.2, 3 Furthermore, the number of patients with CN risks, which are mainly chronic renal insufficiency (CRI) and diabetes (Table 1), has also grown. Currently, more than 26 million people are estimated to have CRI in the United States4 and 200 million people have diabetes worldwide.5 The combination of increased radiocontrast administration frequency and greater prevalence of at‐risk patients is likely to result in continued increases in CN events.

The incidence of CN varies between studies, depending on risk factors of the cohort and definition of CN, but figures have been reported to be as high as 50% in studies enriched with CRI and diabetic patients. However, a very recent study disputes such high incidence rates by demonstrating that patients receiving no radiocontrast media had a similar frequency of serum creatinine increases compared to a comparable group of historical CN patients.6 This study emphasizes that conventional definitions of CN, eg, 25% increase in serum creatinine above baseline, may be too conservative.

A retrospective study of 7586 patients showed 22% in‐hospital mortality in patients who developed CN vs. 1.4% in those who did not, after adjusting for comorbidities. One‐ and 5‐year mortality rates were also higher in the CN group (12.1% vs. 3.7% and 44.6% vs. 14.5%, respectively).7 Another study of 1826 patients, who underwent coronary artery intervention procedures, showed that 14.4% developed CN and 0.8% required hemodialysis. Mortality was 1.1% in patients who did not develop CN, 7.1% in those with CN, and 35.7% in the hemodialysis‐treated CN group.8 Moreover, studies by several other groups also support the position that CN is associated with increased in‐hospital and long‐term mortality.911 Although radiocontrast administration may not be a causal risk factor for mortality, since at‐risk patients have a number of comorbidities, radiocontrast media should nevertheless at least be viewed as an important marker of acute kidney injury and death risk.

Despite the enhanced morbidity and mortality associated with CN, there are no strict guidelines for prevention of CN. Part of the reason is that the literature is controversial regarding most prevention strategies. Several interventions are commonly proposed to help prevent CN, including discriminate selection of the type of radiocontrast, N‐acetylcysteine, volume expansion with saline and/or NaHCO₃, and prophylactic hemofiltration. The major purpose of this review is to discuss these different approaches to CN prevention, with the ultimate goal of offering discrete recommendations.

The basis of this semisystematic review was a literature search using the PubMed database (www.ncbi.nlm.nih.gov/sites/entrez) to identify studies published in English language journals between January 1966 and July 2008 comparing regimens for prophylaxis of CN. Search terms included contrast, radiocontrast, radioiodinated AND nephropathy, nephrotoxicity, renal failure, kidney injury AND N‐acetylcysteine, Mucomyst, sodium bicarbonate, NaHCO₃, hemofiltration, continuous venovenous hemofiltration (CVVH), theophylline, statin, ascorbic acid, dopamine, fenoldopam, adenosine, and endothelin. The total number of articles that met the search criteria exceeded 3000 and over 200 in high profile biomedical journals were scrutinized in detail. The articles cited in this review were independently considered by 2 authors (B.G.A. and J.R.S.) to have the greatest impact.

We report on prophylactic maneuvers that are either commonly considered by nephrology consultants or contain sufficient data to warrant a meta‐analysis study. Studies involving statins, ascorbic acid, dopamine analogs, endothelin antagonists, and theophylline were not addressed in this review due to insufficient, inconclusive, or predominantly negative data. Clinical characteristics and pathogenesis of CN, which include vasoconstriction, ischemia, production of oxygen free radicals, tubular cell apoptosis, and intratubular obstruction, are also not discussed in detail, but have recently been reviewed.12

Risk Associated with Different Types of Radiocontrast Media

There are 3 generations of radiocontrast media: hyperosmolar (14001800 mosm/kg), low osmolar (500850 mosm/kg) and isoosmolar (290 mosm/kg). Note that the low osmolar agents have lower osmolarity relative to the hyperosmolar agents, but are still hyperosmolar compared to serum. Multiple studies have compared effects of radiocontrast with different osmolarities.

The Iohexol Cooperative Study was a double‐blind, randomized, controlled trial (RCT) that randomized 1196 patients to Iohexol (low osmolarity) or diatrizoate (hyperosmolar). Definition of CN was a rise in serum creatinine by >1 mg/dL within 48 to 72 hours after the radiocontrast exposure. Results were in favor of the iohexol group (3% developed CN vs. 7% in the diatrizoate group; P = 0.002). Subgroup analysis of patients with CRI and CRI plus diabetes also revealed less CN in the iohexol vs. diatrizoate groups (7% vs. 16% and 12% vs. 27%, respectively).13

An earlier non‐RCT of 303 patients undergoing femoral angiography compared iohexol/emoxaglate (both are low osmolar) to diatrizoate (hyperosmolar). Six different CN definitions were used. Each comprised a combination of different magnitudes of rise of serum creatinine over various periods of time. Overall, the incidence of CN was 7% in the low osmolar group vs. 26% in the hyperosmolar group (P = 0.001). In subgroup analysis of patients with CRI, and of patients with diabetes, less CN was again observed with low osmolar agents (10% vs. 41%, P = 0.017, in the CRI group; and 10% vs. 31%, P = 0.012, in the diabetes group). Analysis of subjects with baseline serum creatinine <1.5 mg/dL showed no differences between the 2 groups, emphasizing that prior CRI is an important CN risk factor.14

The RECOVER study was a double‐blind RCT of 300 patients undergoing coronary angiography, who were randomized to iodixanol (isoosmolar) or ioxaglate (low osmolarity). CN was defined as a rise in serum creatinine by >25% or 0.5 mg/dL at 24 and 48 hours. CN incidence was 7.9% in the iodixanol group vs. 17% in the ioxaglate group (P = 0.021). Subgroup analyses of patients stratified by severe CRI, diabetes, and contrast volume also favored iodixanol.15 In a similarly designed, double‐blind RCT involving 129 patients with diabetes and CRI randomized to iodixanol or iohexol, CN developed in 3% of iodixanol group vs. 26% in the iohexol group (P = 0.002).16 These results were further supported by a very recent double‐blind RCT comparing iodixanol to iopromide (low osmolar) in 117 patients with baseline serum creatinine 1.5 mg/dL undergoing CT scans. The incidence of serum creatinine increases of 0.5 mg/dL or 25% above baseline or glomerular filtration rate (GFR) reduction of 5 mL/minute was significantly lower in the iodixanol group (P = 0.04, 0.01, and 0.04 respectively).17

In contrast to these reports, 2 double‐blind RCTs showed no differences in CN incidence between iodixanol and low osmolar agents. One study compared iodixanol to iopromide in 64 patients undergoing intravenous pyelography (IVP),18 and the other included 16 nondiabetic patients with CRI and compared iodixanol to iohexol.19 However, because of the small sample sizes, it is likely that neither study is adequately powered to detect differences in outcome between the 2 types of radiocontrast media.

Given the importance of the issue and conflicting results from individual studies, a meta‐analysis of 16 double‐blind RCTs was performed.20 This study included 2727 patients undergoing angiography, compared iodixanol (isoosmolar) to a variety of low osmolar agents, and demonstrated that iodixanol was less nephrotoxic compared to the low osmolarity agents in CRI patients (2.8% vs. 8.4%; P = 0.001) and in patients with CRI plus diabetes (3.5% vs. 15.5%; P = 0.003). Independent predictors of CN were CRI, CRI plus diabetes, and the use of low‐osmolarity media, whereas diabetes, age, and radiocontrast volume were not statistically significant independent predictors.20

Taken together, we conclude that isoosmolar media represents the lowest risk for CN. An additional benefit is that isoosmolar media, on the basis of diminished osmotic load, is less likely to precipitate extracellular fluid volume overload, which is particularly germane for patients with CRI, who have diminished capacity to excrete solute loads. Therefore, we recommend using isoosmolar media, particularly in patients at high risk for CN, such as those with CRI, especially due to diabetic nephropathy.

Oral N‐Acetylcysteine

Based primarily upon in vitro evidence, N‐acetylcysteine (NAC) may theoretically prevent CN by direct antioxidant and vasodilatory effects. However, in vivo, NAC is rapidly metabolized and inactivated by the liver. Therefore, it has been postulated that the mechanism of action may be indirect, and the cysteine metabolite of NAC may stimulate glutathione synthesis, which then inhibits cellular oxidation.21

The first clinical trial to address the prophylactic role of NAC in CN was an RCT of 83 patients with CRI (mean serum creatinine [Cr] = 2.4) undergoing CT scans, who were randomized to NAC plus 0.45% NaCl vs. placebo and 0.9% NaCl.22 The NAC dose was 600 mg orally twice daily for 2 doses before and 2 doses after the procedure. Intravenous fluids were started 12 hours before and stopped 12 hours after the procedure and infused at a rate of 1 ml/kg/hour. CN definition was rise in serum creatinine by >0.5 mg/dL at 48 hours. The results were statistically significant, with a relative risk of CN = 0.1 (95%CI, 0.020.9) in subjects treated with NAC.

There have been many subsequent reports that have evaluated NAC in small numbers of patients with mild to moderate CRI. In general, results from these trials have been inconsistent, which has led to several meta‐analyses to delineate NAC efficacy in CN prevention. The most recent and largest meta‐analysis included 26 NAC RCTs, and revealed a statistically significant benefit from NAC (relative risk [RR] = 0.62; 95%CI, 0.440.88).23 Twelve other meta‐analyses, which incorporated fewer studies, have been published,2435 and 7 of the 12 reported a benefit from NAC.25, 27, 29, 30, 32, 34, 35

Although meta‐analysis is considered the most accepted strategy to define conclusions from multiple trials, conflicting results between NAC meta‐analyses highlight the possibility that this approach may still not provide resolution to clinical questions, especially when inclusion criteria differ between meta‐analyses. Therefore, as discussed by Bagshaw et al.,36 meta‐analyses are not always a panacea, and should be avoided if the trials to be included exhibit significant statistical or clinical heterogeneity, as is the case with studies involving NAC prophylaxis of CN. Finally, because meta‐analyses require pooling of data from published studies, which tend to be positive, the possibility of publication bias exists.

In summary, conclusions from trials to assess efficacy of oral NAC in the prevention of CN have been inconsistent, though there has been a general trend toward benefit. Factors contributing to inconsistent results include variable definitions of CN, degree of CRI and diabetes in the cohort, amount and type of contrast used, NAC dosing and intravenous hydration protocols. As a result, a large multicenter RCT would certainly be helpful. However, the size of such trial might be cost‐prohibitive, and unlikely to be underwritten by the pharmaceutical industry because the patent for NAC has expired.36

Intravenous NAC

In addition to the vast literature on oral NAC for CN prophylaxis, there are now studies that have also evaluated efficacy of intravenous NAC. In one of the largest double‐blind RCTs,37 487 patients (mean baseline serum creatinine = 1.6 mg/dL) were randomized to NAC 500 mg intravenously vs. placebo before cardiac catheterization. Both groups received the same hydration protocols. The study was stopped when an interim analysis determined that there was no advantage to NAC (CN incidence, which was defined as a decrease in creatinine clearance by >5 mL/minute at days 1 to 8 postprocedure, was 23.3% vs. 20.7% in the placebo group).

The RAPPID study examined higher intravenous NAC doses in 80 patients undergoing cardiac catheterization.38 Subjects in this study were randomized to either NAC (150 mg/kg in 500 mL 0.9% NaCl before procedure and then 50 mg/kg in 500 mL 0.9% NaCl over 4 hours after procedure) or 0.9% NaCl 12 hours before and 12 hours after procedure. Despite the relatively small study size, intravenous NAC demonstrated a significant benefit in the prevention of CN (RR = 0.28; P = 0.045) defined as rise in serum creatinine by >25% at 2 or 4 days postexposure. Hypersensitivity‐like reactions were observed in 14.5% of patients receiving intravenous NAC, but symptoms were easily recognized and managed.38

A recent study of 354 patients undergoing primary angioplasty evaluated the combination of intravenous and oral NAC in different doses.39 Patients were randomized to 3 groups: (1) NAC 600 mg intravenously once before procedure and then 600 mg orally twice daily for 48 hours; (2) NAC 1200 mg intravenously once before procedure and then 1200 mg orally twice daily for 48 hours; or (3) placebo. The primary outcome was increase in Cr by >25% and secondary outcomes were in‐hospital death and a composite score that included death and need for renal replacement therapy. Results were significantly in favor of the 1200‐mg NAC regimen across all outcomes (P = <0.001, 0.02, and 0.002, respectively). It should be emphasized that this study was restricted to patients undergoing primary angioplasty, which is an emergent procedure. As a result, implementation of this protocol would necessarily require rapid administration of intravenous NAC prior to the procedure, which might even require maintenance of NAC stocks within the catheterization laboratory.

Because there is a trend toward benefit from oral NAC and the benefit from intravenous NAC in trials from limited settings, and both NAC formulations are inexpensive and safe, we recommend that NAC should be included in CN prophylaxis protocols.

Extracellular Fluid Volume Expansion

Since publication of work by Solomon et al.,40 which demonstrated a benefit of intravenous hydration with 0.45% NaCl in the prevention of CN in a group of CKD patients, it has been considered standard practice to prescribe intravenous fluid regimens for CN prophylaxis in high‐risk subjects. In the largest study to test the effect of different hydration protocols for CN prevention, Mueller et al.41 randomized 1620 patients with normal baseline serum creatinine to intravenous 0.9 % NaCl vs. 0.45% NaCl. The definition of CN was rise in serum creatinine by >0.5 mg/dL at 48 hours and the incidence was 0.7% for the 0.9% NaCl group and 2% for the 0.45% NaCl group (P = 0.04).

More recently, several trials have examined the relative efficacy of intravenous NaCl vs. NaHCO₃ for CN prophylaxis. In the first NaHCO₃ RCT, Merten et al.42 compared 0.9% NaHCO₃ to 0.9% NaCl infusion in a population with a mean serum creatinine of 1.8 mg/dL. Both groups received 3 mL/kg intravenous bolus over 1 hour before the radiographic procedure followed by 1 mL/kg/hour for 6 hours. Urine pH was measured to confirm alkalinization of urine in the NaHCO₃‐treated patients and the primary end point was increase in serum creatinine by >25% within 48 hours. The study was terminated early (after enrollment of 119 patients) when the interim analysis showed CN incidence was 1.7% in the NaHCO₃ group vs. 13.6% in the NaCl group (P = 0.02).

These results were corroborated in the recent REMEDIAL trial,43 which enrolled 326 patients with serum creatinine >2 mg/dL, who were randomized to 1 of 3 arms. One group received intravenous saline (0.9% NaCl for 12 hours before and 12 hours after the procedure) and oral NAC; a second group received intravenous NaHCO₃ (3 mL/kg intravenous bolus over 1 hour before the radiographic procedure followed by 1 mL/kg/hour for 6 hours) and oral NAC; and a third group received intravenous 0.9% NaCl plus oral NAC and ascorbic acid. Patients had similar baseline characteristics and the primary end point was an increase in serum creatinine by >25% within 48 hours. The best results were observed in the NaHCO₃ plus NAC group; 1.9% developed CN in this group vs. 9.9% in the NaCl plus NAC group vs. 10.3% in the NaCl plus NAC plus ascorbic acid group (P = 0.019). Three additional prospective but smaller studies also showed the superiority of NaHCO₃.4446

In contrast to studies supporting a role for prophylactic NaHCO₃, a recent RCT showed no superiority of NaHCO₃ infusion regimens.47 In this trial, 352 patients undergoing coronary angiography were randomized to receive either NaHCO₃ or 0.9% NaCl. Both solutions were administered at rates of 3 mL/kg for 1 hour before the procedure and 1.5 mL/kg/hour for 4 hours postprocedure. The primary endpoint (>25% decrease in estimated GFR during the first 4 days after contrast exposure) was met in 13.3% of NaHCO₃ group vs. 14.6% of the 0.9% NaCl group (P = 0.82). Moreover, there were no differences in the rates of secondary outcomes, which included death, dialysis, and cardiovascular and cerebrovascular events.

Results from a very recent retrospective cohort study of 7977 patients demonstrated that NaHCO₃ infusion was associated with increased risk of CN compared to no treatment (odds ratio [OR] = 3.1; P < 0.001), whereas NAC alone or in combination with NaHCO₃ was associated with no significant difference in the incidence of CN.48 However, multiple weaknesses associated with the retrospective study design, such as inclusion of few patients at high CN risk, and acceptance of serum creatinine values within 7 days before and after the contrast procedure, which likely captures causes of acute kidney injury other than CN, preclude abandonment of NaHCO₃ prophylaxis for CN solely on the basis of this study.

In an effort to resolve the conflicting NaHCO₃ prophylaxis literature, a meta‐analysis was recently conducted.49 This study encompassed 1307 patients enrolled in 7 RCTs that examined outcomes for NaHCO₃ vs. saline prevention of CN. The main finding was a significant benefit of NaHCO₃ for protection against CN (RR = 0.37; P = 0.005). No benefit of NaHCO₃ infusion could be shown for postprocedure renal replacement therapy or death.

Therefore, based upon the results of multiple prospective trials, the recent meta‐analysis, the relative safety of NaHCO₃ infusion with appropriate monitoring, and a plausible biological mechanism whereby bicarbonate may have antioxidant properties and scavenge oxygen‐derived free radicals, which have been implicated in CN pathogenesis,50 we advocate a prophylactic regimen employing NaHCO₃ for patients at high risk for CN.

Renal Replacement Therapies

Two studies have been conducted by the same group (Marenzi et al.51, 52) to examine efficacy of continuous hemofiltration (CVVH) in preventing CN (hemofiltration clears solute by convection, and involves administration of a HCO‐rich replacement solution, whereas hemodialysis clears solute by both diffusion and convection, and there is routinely no replacement fluid). In the first study,51 114 patients with baseline serum creatinine greater than 2 mg/dL undergoing coronary angiography were randomized to hemofiltration vs. 0.9% NaCl infusion. Isovolemic hemofiltration was implemented for 4 to 6 hours before and 18 to 24 hours after the radiographic procedure. The primary endpoint was increase in creatinine by >25% within 72 hours. CN incidence was 5% in the hemofiltration group vs. 50% in the 0.9% NaCl group (P < 0.001). The secondary outcomes including in‐hospital mortality, 1‐year mortality and temporary renal replacement were also superior in the hemofiltration group. In the second study, the same investigators compared 2 different hemofiltration protocols, using the same definition of CN.52 Patients with baseline creatinine clearance <30 mL/minute (n = 92) were randomized to 0.9% NaCl infusion, postprocedure isovolemic hemofiltration only, or preprocedure plus postprocedure hemofiltration (same protocol as previous study). The incidence of CN was significantly lower in the preprocedure plus postprocedure hemofiltration group (3% vs. 26% in the postprocedure hemofiltration group vs. 40% in the 0.9% NaCl without hemofiltration group; P = 0.0001). The preprocedure plus postprocedure hemofiltration group also had reductions in in‐hospital mortality and temporary renal replacement therapy rates.

Although the mechanism of hemofiltration prevention of CN is unknown, it is certainly not enhanced clearance of contrast material, inasmuch as hemofiltration was discontinued during the angiography procedure in all protocols, and radiocontrast was therefore not cleared by hemofiltration until the process was reinstituted. Furthermore, the second study indicates that the major benefit was derived from the preprocedure hemofiltration component. Contributing factors might be control of extracellular pH and redox potential with bicarbonate replacement fluid during hemofiltration. Important confounding issues to consider are that patients receiving hemofiltration were in controlled, monitored settings and thus received more intensive care than the hydration group, and that serum creatinine, the major outcome parameter, is cleared by hemofiltration. Before hemofiltration can be recommended as routine prophylactic therapy for CN, the data will need to be corroborated by other groups, preferably involving larger numbers of study subjects and including cost‐benefit analyses.

Multiple small studies have examined the possibility that dialysis immediately following radiocontrast exposure could prevent CN, presumably by accelerating radiocontrast clearance. Most of these reports were negative, including a well‐designed meta‐analysis of RCTs, which showed no benefit of hemodialysis.53 Of note, one report suggested that hemodialysis might be potentially harmful.54 The single prospective trial that showed benefit from prophylactic hemodialysis analyzed 82 patients with advanced CRI (baseline creatinine clearance 13 mL/minute) referred for coronary angiography55 These subjects were randomly assigned to intravenous 0.9% NaCl and hemodialysis vs. intravenous 0.9% NaCl alone. Subsequent renal replacement therapy was required in 35% of control patients and in only 2% in the prophylactic dialysis group. One potential limitation of this study is that the investigators were more cognizant of volume status in the hemodialysis group to avoid fluid shifts and volume depletion during dialysis, while the control group appeared to experience no comparable intravascular volume management. Moreover, this study was conducted in patients with extremely advanced renal insufficiency, and therefore does not reflect the vast majority of patients at risk for CN.

Conclusions

CN is associated with increased morbidity and mortality, and efforts to minimize CN are therefore warranted. However, the overwhelming majority of CN trials were designed to investigate the effects of prophylaxis strategies on surrogate endpoints for estimates of GFR. Therefore, conclusions regarding the effect of these regimens on definitive outcomes, such as death and vascular events, cannot be drawn. On balance, there is evidence that oral and intravenous NAC, as well as extracellular volume expansion with intravenous NaHCO₃ are effective measures to prevent CN, whereas the data for renal replacement therapies are more equivocal. We emphasize though, that the literature on this topic is vast, and includes a large number of conflicting studies, including multiple meta‐analyses. As a result, we refrain from being too dogmatic about the best approach, and therefore cautiously offer the following recommendations for prevention of CN.

The first step is to identify high‐risk patients, who are most likely to benefit from prophylaxis (Table 1). Although risk stratification was not the focus of this review, Mehran et al.56 have developed a scoring system to quantitatively predict CN risk, with weighted parameters including CRI, diabetes, radiocontrast volume, age, hypotension, congestive heart failure, treatment with an intraaortic balloon pump, and anemia. For low‐risk patients, hydration with saline is probably adequate. For high‐risk patients, it would be prudent to initially consider whether sufficient information could be obtained from an alternative, noncontrasted radiologic procedure. If not, it would behoove the prescribing physician to then treat modifiable risk factors, as well as to discontinue potentially nephrotoxic medications.

In high‐risk patients undergoing radiocontrast procedures, we recommend using NAC and volume expansion with NaHCO₃ (Table 2). Although the evidence for this combined approach is limited,43 we believe it is biologically consistent, since the rationale for both strategies is primarily modification of redox state and inhibition of oxygen free radical generation. Because NAC formulations are generally effective, safe and inexpensive ($0.04 for 600 mg oral NAC and $24 for 1200 mg intravenous NAC at our hospital), we recommend the protocol used by Marenzi et al.,39 NAC 1200 mg intravenously once before procedure and then 1200 mg orally twice daily for 48 hours, as prophylaxis for all contrast procedures in high‐risk patients. However, we recognize that this regimen would require formal evaluation in procedures other than emergent coronary artery angioplasty before it could be enthusiastically endorsed. Therefore, if intravenous NAC is not available and/or the procedure is not emergent, NAC 6001200 mg orally twice a day, 2 doses before and 2 doses after the procedure would be a rational alternative. For the NaHCO₃ infusion, we recommend 3 mL/kg for 1 hour before the procedure, followed by 1 mL/kg/hour for 6 hours after.

Hemofiltration is labor‐intensive, expensive, and not readily available in all hospitals that renders it difficult to endorse as a definitive or routine CN prophylaxis modality. However, if a patient is already undergoing acute dialysis with catheter vascular access, it would be reasonable to consider CVVH 6 hours before and for 24 hours after the procedure (Table 2).

For high‐risk patients, we recommend minimizing the radiocontrast dose (reviewed in Ref.57). Although the dose has not consistently been identified as a risk factor (Table 1), we envision no harm in reducing the dose, particularly if adequate information can be obtained by other means, eg, coronary angiogram accompanied by an echocardiogram, rather than a ventriculogram. We would also consider the use of isoosmolar media in high‐risk patients, since the data are relatively compelling.20 Low doses of isoosmolar media should be particularly beneficial to patients with preexisting hypertension or congestive heart failure, for which the osmotic load and excess extracellular volume expansion might be deleterious. However, because isoosmolar media is expensive, a detailed cost‐benefit analysis would be required before definitive recommendations could be made, especially for patients at lower risk for CN. Finally, because of the complex literature, as well as budgetary issues, we encourage communication between the physician ordering the contrast study and the operator (radiologist or cardiologist) concerning the type of procedure and contrast media to be used.

Since contrast nephropathy (CN) was recognized more than 50 years ago,1 there have been continuous efforts to chemically modify radiocontrast agents to be less nephrotoxic. Although radiocontrast media have indeed become safer, which reduces the likelihood of CN per procedure, the indications for radiocontrast administration have dramatically increased, since over 80 million doses are delivered in the world annually.2, 3 Furthermore, the number of patients with CN risks, which are mainly chronic renal insufficiency (CRI) and diabetes (Table 1), has also grown. Currently, more than 26 million people are estimated to have CRI in the United States4 and 200 million people have diabetes worldwide.5 The combination of increased radiocontrast administration frequency and greater prevalence of at‐risk patients is likely to result in continued increases in CN events.

The incidence of CN varies between studies, depending on risk factors of the cohort and definition of CN, but figures have been reported to be as high as 50% in studies enriched with CRI and diabetic patients. However, a very recent study disputes such high incidence rates by demonstrating that patients receiving no radiocontrast media had a similar frequency of serum creatinine increases compared to a comparable group of historical CN patients.6 This study emphasizes that conventional definitions of CN, eg, 25% increase in serum creatinine above baseline, may be too conservative.

A retrospective study of 7586 patients showed 22% in‐hospital mortality in patients who developed CN vs. 1.4% in those who did not, after adjusting for comorbidities. One‐ and 5‐year mortality rates were also higher in the CN group (12.1% vs. 3.7% and 44.6% vs. 14.5%, respectively).7 Another study of 1826 patients, who underwent coronary artery intervention procedures, showed that 14.4% developed CN and 0.8% required hemodialysis. Mortality was 1.1% in patients who did not develop CN, 7.1% in those with CN, and 35.7% in the hemodialysis‐treated CN group.8 Moreover, studies by several other groups also support the position that CN is associated with increased in‐hospital and long‐term mortality.911 Although radiocontrast administration may not be a causal risk factor for mortality, since at‐risk patients have a number of comorbidities, radiocontrast media should nevertheless at least be viewed as an important marker of acute kidney injury and death risk.

Despite the enhanced morbidity and mortality associated with CN, there are no strict guidelines for prevention of CN. Part of the reason is that the literature is controversial regarding most prevention strategies. Several interventions are commonly proposed to help prevent CN, including discriminate selection of the type of radiocontrast, N‐acetylcysteine, volume expansion with saline and/or NaHCO₃, and prophylactic hemofiltration. The major purpose of this review is to discuss these different approaches to CN prevention, with the ultimate goal of offering discrete recommendations.

The basis of this semisystematic review was a literature search using the PubMed database (www.ncbi.nlm.nih.gov/sites/entrez) to identify studies published in English language journals between January 1966 and July 2008 comparing regimens for prophylaxis of CN. Search terms included contrast, radiocontrast, radioiodinated AND nephropathy, nephrotoxicity, renal failure, kidney injury AND N‐acetylcysteine, Mucomyst, sodium bicarbonate, NaHCO₃, hemofiltration, continuous venovenous hemofiltration (CVVH), theophylline, statin, ascorbic acid, dopamine, fenoldopam, adenosine, and endothelin. The total number of articles that met the search criteria exceeded 3000 and over 200 in high profile biomedical journals were scrutinized in detail. The articles cited in this review were independently considered by 2 authors (B.G.A. and J.R.S.) to have the greatest impact.

We report on prophylactic maneuvers that are either commonly considered by nephrology consultants or contain sufficient data to warrant a meta‐analysis study. Studies involving statins, ascorbic acid, dopamine analogs, endothelin antagonists, and theophylline were not addressed in this review due to insufficient, inconclusive, or predominantly negative data. Clinical characteristics and pathogenesis of CN, which include vasoconstriction, ischemia, production of oxygen free radicals, tubular cell apoptosis, and intratubular obstruction, are also not discussed in detail, but have recently been reviewed.12

Risk Associated with Different Types of Radiocontrast Media

There are 3 generations of radiocontrast media: hyperosmolar (14001800 mosm/kg), low osmolar (500850 mosm/kg) and isoosmolar (290 mosm/kg). Note that the low osmolar agents have lower osmolarity relative to the hyperosmolar agents, but are still hyperosmolar compared to serum. Multiple studies have compared effects of radiocontrast with different osmolarities.

The Iohexol Cooperative Study was a double‐blind, randomized, controlled trial (RCT) that randomized 1196 patients to Iohexol (low osmolarity) or diatrizoate (hyperosmolar). Definition of CN was a rise in serum creatinine by >1 mg/dL within 48 to 72 hours after the radiocontrast exposure. Results were in favor of the iohexol group (3% developed CN vs. 7% in the diatrizoate group; P = 0.002). Subgroup analysis of patients with CRI and CRI plus diabetes also revealed less CN in the iohexol vs. diatrizoate groups (7% vs. 16% and 12% vs. 27%, respectively).13

An earlier non‐RCT of 303 patients undergoing femoral angiography compared iohexol/emoxaglate (both are low osmolar) to diatrizoate (hyperosmolar). Six different CN definitions were used. Each comprised a combination of different magnitudes of rise of serum creatinine over various periods of time. Overall, the incidence of CN was 7% in the low osmolar group vs. 26% in the hyperosmolar group (P = 0.001). In subgroup analysis of patients with CRI, and of patients with diabetes, less CN was again observed with low osmolar agents (10% vs. 41%, P = 0.017, in the CRI group; and 10% vs. 31%, P = 0.012, in the diabetes group). Analysis of subjects with baseline serum creatinine <1.5 mg/dL showed no differences between the 2 groups, emphasizing that prior CRI is an important CN risk factor.14

The RECOVER study was a double‐blind RCT of 300 patients undergoing coronary angiography, who were randomized to iodixanol (isoosmolar) or ioxaglate (low osmolarity). CN was defined as a rise in serum creatinine by >25% or 0.5 mg/dL at 24 and 48 hours. CN incidence was 7.9% in the iodixanol group vs. 17% in the ioxaglate group (P = 0.021). Subgroup analyses of patients stratified by severe CRI, diabetes, and contrast volume also favored iodixanol.15 In a similarly designed, double‐blind RCT involving 129 patients with diabetes and CRI randomized to iodixanol or iohexol, CN developed in 3% of iodixanol group vs. 26% in the iohexol group (P = 0.002).16 These results were further supported by a very recent double‐blind RCT comparing iodixanol to iopromide (low osmolar) in 117 patients with baseline serum creatinine 1.5 mg/dL undergoing CT scans. The incidence of serum creatinine increases of 0.5 mg/dL or 25% above baseline or glomerular filtration rate (GFR) reduction of 5 mL/minute was significantly lower in the iodixanol group (P = 0.04, 0.01, and 0.04 respectively).17

In contrast to these reports, 2 double‐blind RCTs showed no differences in CN incidence between iodixanol and low osmolar agents. One study compared iodixanol to iopromide in 64 patients undergoing intravenous pyelography (IVP),18 and the other included 16 nondiabetic patients with CRI and compared iodixanol to iohexol.19 However, because of the small sample sizes, it is likely that neither study is adequately powered to detect differences in outcome between the 2 types of radiocontrast media.

Given the importance of the issue and conflicting results from individual studies, a meta‐analysis of 16 double‐blind RCTs was performed.20 This study included 2727 patients undergoing angiography, compared iodixanol (isoosmolar) to a variety of low osmolar agents, and demonstrated that iodixanol was less nephrotoxic compared to the low osmolarity agents in CRI patients (2.8% vs. 8.4%; P = 0.001) and in patients with CRI plus diabetes (3.5% vs. 15.5%; P = 0.003). Independent predictors of CN were CRI, CRI plus diabetes, and the use of low‐osmolarity media, whereas diabetes, age, and radiocontrast volume were not statistically significant independent predictors.20

Taken together, we conclude that isoosmolar media represents the lowest risk for CN. An additional benefit is that isoosmolar media, on the basis of diminished osmotic load, is less likely to precipitate extracellular fluid volume overload, which is particularly germane for patients with CRI, who have diminished capacity to excrete solute loads. Therefore, we recommend using isoosmolar media, particularly in patients at high risk for CN, such as those with CRI, especially due to diabetic nephropathy.

Oral N‐Acetylcysteine

Based primarily upon in vitro evidence, N‐acetylcysteine (NAC) may theoretically prevent CN by direct antioxidant and vasodilatory effects. However, in vivo, NAC is rapidly metabolized and inactivated by the liver. Therefore, it has been postulated that the mechanism of action may be indirect, and the cysteine metabolite of NAC may stimulate glutathione synthesis, which then inhibits cellular oxidation.21

The first clinical trial to address the prophylactic role of NAC in CN was an RCT of 83 patients with CRI (mean serum creatinine [Cr] = 2.4) undergoing CT scans, who were randomized to NAC plus 0.45% NaCl vs. placebo and 0.9% NaCl.22 The NAC dose was 600 mg orally twice daily for 2 doses before and 2 doses after the procedure. Intravenous fluids were started 12 hours before and stopped 12 hours after the procedure and infused at a rate of 1 ml/kg/hour. CN definition was rise in serum creatinine by >0.5 mg/dL at 48 hours. The results were statistically significant, with a relative risk of CN = 0.1 (95%CI, 0.020.9) in subjects treated with NAC.

There have been many subsequent reports that have evaluated NAC in small numbers of patients with mild to moderate CRI. In general, results from these trials have been inconsistent, which has led to several meta‐analyses to delineate NAC efficacy in CN prevention. The most recent and largest meta‐analysis included 26 NAC RCTs, and revealed a statistically significant benefit from NAC (relative risk [RR] = 0.62; 95%CI, 0.440.88).23 Twelve other meta‐analyses, which incorporated fewer studies, have been published,2435 and 7 of the 12 reported a benefit from NAC.25, 27, 29, 30, 32, 34, 35

Although meta‐analysis is considered the most accepted strategy to define conclusions from multiple trials, conflicting results between NAC meta‐analyses highlight the possibility that this approach may still not provide resolution to clinical questions, especially when inclusion criteria differ between meta‐analyses. Therefore, as discussed by Bagshaw et al.,36 meta‐analyses are not always a panacea, and should be avoided if the trials to be included exhibit significant statistical or clinical heterogeneity, as is the case with studies involving NAC prophylaxis of CN. Finally, because meta‐analyses require pooling of data from published studies, which tend to be positive, the possibility of publication bias exists.

In summary, conclusions from trials to assess efficacy of oral NAC in the prevention of CN have been inconsistent, though there has been a general trend toward benefit. Factors contributing to inconsistent results include variable definitions of CN, degree of CRI and diabetes in the cohort, amount and type of contrast used, NAC dosing and intravenous hydration protocols. As a result, a large multicenter RCT would certainly be helpful. However, the size of such trial might be cost‐prohibitive, and unlikely to be underwritten by the pharmaceutical industry because the patent for NAC has expired.36

Intravenous NAC

In addition to the vast literature on oral NAC for CN prophylaxis, there are now studies that have also evaluated efficacy of intravenous NAC. In one of the largest double‐blind RCTs,37 487 patients (mean baseline serum creatinine = 1.6 mg/dL) were randomized to NAC 500 mg intravenously vs. placebo before cardiac catheterization. Both groups received the same hydration protocols. The study was stopped when an interim analysis determined that there was no advantage to NAC (CN incidence, which was defined as a decrease in creatinine clearance by >5 mL/minute at days 1 to 8 postprocedure, was 23.3% vs. 20.7% in the placebo group).

The RAPPID study examined higher intravenous NAC doses in 80 patients undergoing cardiac catheterization.38 Subjects in this study were randomized to either NAC (150 mg/kg in 500 mL 0.9% NaCl before procedure and then 50 mg/kg in 500 mL 0.9% NaCl over 4 hours after procedure) or 0.9% NaCl 12 hours before and 12 hours after procedure. Despite the relatively small study size, intravenous NAC demonstrated a significant benefit in the prevention of CN (RR = 0.28; P = 0.045) defined as rise in serum creatinine by >25% at 2 or 4 days postexposure. Hypersensitivity‐like reactions were observed in 14.5% of patients receiving intravenous NAC, but symptoms were easily recognized and managed.38

A recent study of 354 patients undergoing primary angioplasty evaluated the combination of intravenous and oral NAC in different doses.39 Patients were randomized to 3 groups: (1) NAC 600 mg intravenously once before procedure and then 600 mg orally twice daily for 48 hours; (2) NAC 1200 mg intravenously once before procedure and then 1200 mg orally twice daily for 48 hours; or (3) placebo. The primary outcome was increase in Cr by >25% and secondary outcomes were in‐hospital death and a composite score that included death and need for renal replacement therapy. Results were significantly in favor of the 1200‐mg NAC regimen across all outcomes (P = <0.001, 0.02, and 0.002, respectively). It should be emphasized that this study was restricted to patients undergoing primary angioplasty, which is an emergent procedure. As a result, implementation of this protocol would necessarily require rapid administration of intravenous NAC prior to the procedure, which might even require maintenance of NAC stocks within the catheterization laboratory.

Because there is a trend toward benefit from oral NAC and the benefit from intravenous NAC in trials from limited settings, and both NAC formulations are inexpensive and safe, we recommend that NAC should be included in CN prophylaxis protocols.

Extracellular Fluid Volume Expansion

Since publication of work by Solomon et al.,40 which demonstrated a benefit of intravenous hydration with 0.45% NaCl in the prevention of CN in a group of CKD patients, it has been considered standard practice to prescribe intravenous fluid regimens for CN prophylaxis in high‐risk subjects. In the largest study to test the effect of different hydration protocols for CN prevention, Mueller et al.41 randomized 1620 patients with normal baseline serum creatinine to intravenous 0.9 % NaCl vs. 0.45% NaCl. The definition of CN was rise in serum creatinine by >0.5 mg/dL at 48 hours and the incidence was 0.7% for the 0.9% NaCl group and 2% for the 0.45% NaCl group (P = 0.04).

More recently, several trials have examined the relative efficacy of intravenous NaCl vs. NaHCO₃ for CN prophylaxis. In the first NaHCO₃ RCT, Merten et al.42 compared 0.9% NaHCO₃ to 0.9% NaCl infusion in a population with a mean serum creatinine of 1.8 mg/dL. Both groups received 3 mL/kg intravenous bolus over 1 hour before the radiographic procedure followed by 1 mL/kg/hour for 6 hours. Urine pH was measured to confirm alkalinization of urine in the NaHCO₃‐treated patients and the primary end point was increase in serum creatinine by >25% within 48 hours. The study was terminated early (after enrollment of 119 patients) when the interim analysis showed CN incidence was 1.7% in the NaHCO₃ group vs. 13.6% in the NaCl group (P = 0.02).

These results were corroborated in the recent REMEDIAL trial,43 which enrolled 326 patients with serum creatinine >2 mg/dL, who were randomized to 1 of 3 arms. One group received intravenous saline (0.9% NaCl for 12 hours before and 12 hours after the procedure) and oral NAC; a second group received intravenous NaHCO₃ (3 mL/kg intravenous bolus over 1 hour before the radiographic procedure followed by 1 mL/kg/hour for 6 hours) and oral NAC; and a third group received intravenous 0.9% NaCl plus oral NAC and ascorbic acid. Patients had similar baseline characteristics and the primary end point was an increase in serum creatinine by >25% within 48 hours. The best results were observed in the NaHCO₃ plus NAC group; 1.9% developed CN in this group vs. 9.9% in the NaCl plus NAC group vs. 10.3% in the NaCl plus NAC plus ascorbic acid group (P = 0.019). Three additional prospective but smaller studies also showed the superiority of NaHCO₃.4446

In contrast to studies supporting a role for prophylactic NaHCO₃, a recent RCT showed no superiority of NaHCO₃ infusion regimens.47 In this trial, 352 patients undergoing coronary angiography were randomized to receive either NaHCO₃ or 0.9% NaCl. Both solutions were administered at rates of 3 mL/kg for 1 hour before the procedure and 1.5 mL/kg/hour for 4 hours postprocedure. The primary endpoint (>25% decrease in estimated GFR during the first 4 days after contrast exposure) was met in 13.3% of NaHCO₃ group vs. 14.6% of the 0.9% NaCl group (P = 0.82). Moreover, there were no differences in the rates of secondary outcomes, which included death, dialysis, and cardiovascular and cerebrovascular events.

Results from a very recent retrospective cohort study of 7977 patients demonstrated that NaHCO₃ infusion was associated with increased risk of CN compared to no treatment (odds ratio [OR] = 3.1; P < 0.001), whereas NAC alone or in combination with NaHCO₃ was associated with no significant difference in the incidence of CN.48 However, multiple weaknesses associated with the retrospective study design, such as inclusion of few patients at high CN risk, and acceptance of serum creatinine values within 7 days before and after the contrast procedure, which likely captures causes of acute kidney injury other than CN, preclude abandonment of NaHCO₃ prophylaxis for CN solely on the basis of this study.

In an effort to resolve the conflicting NaHCO₃ prophylaxis literature, a meta‐analysis was recently conducted.49 This study encompassed 1307 patients enrolled in 7 RCTs that examined outcomes for NaHCO₃ vs. saline prevention of CN. The main finding was a significant benefit of NaHCO₃ for protection against CN (RR = 0.37; P = 0.005). No benefit of NaHCO₃ infusion could be shown for postprocedure renal replacement therapy or death.

Therefore, based upon the results of multiple prospective trials, the recent meta‐analysis, the relative safety of NaHCO₃ infusion with appropriate monitoring, and a plausible biological mechanism whereby bicarbonate may have antioxidant properties and scavenge oxygen‐derived free radicals, which have been implicated in CN pathogenesis,50 we advocate a prophylactic regimen employing NaHCO₃ for patients at high risk for CN.

Renal Replacement Therapies

Two studies have been conducted by the same group (Marenzi et al.51, 52) to examine efficacy of continuous hemofiltration (CVVH) in preventing CN (hemofiltration clears solute by convection, and involves administration of a HCO‐rich replacement solution, whereas hemodialysis clears solute by both diffusion and convection, and there is routinely no replacement fluid). In the first study,51 114 patients with baseline serum creatinine greater than 2 mg/dL undergoing coronary angiography were randomized to hemofiltration vs. 0.9% NaCl infusion. Isovolemic hemofiltration was implemented for 4 to 6 hours before and 18 to 24 hours after the radiographic procedure. The primary endpoint was increase in creatinine by >25% within 72 hours. CN incidence was 5% in the hemofiltration group vs. 50% in the 0.9% NaCl group (P < 0.001). The secondary outcomes including in‐hospital mortality, 1‐year mortality and temporary renal replacement were also superior in the hemofiltration group. In the second study, the same investigators compared 2 different hemofiltration protocols, using the same definition of CN.52 Patients with baseline creatinine clearance <30 mL/minute (n = 92) were randomized to 0.9% NaCl infusion, postprocedure isovolemic hemofiltration only, or preprocedure plus postprocedure hemofiltration (same protocol as previous study). The incidence of CN was significantly lower in the preprocedure plus postprocedure hemofiltration group (3% vs. 26% in the postprocedure hemofiltration group vs. 40% in the 0.9% NaCl without hemofiltration group; P = 0.0001). The preprocedure plus postprocedure hemofiltration group also had reductions in in‐hospital mortality and temporary renal replacement therapy rates.

Although the mechanism of hemofiltration prevention of CN is unknown, it is certainly not enhanced clearance of contrast material, inasmuch as hemofiltration was discontinued during the angiography procedure in all protocols, and radiocontrast was therefore not cleared by hemofiltration until the process was reinstituted. Furthermore, the second study indicates that the major benefit was derived from the preprocedure hemofiltration component. Contributing factors might be control of extracellular pH and redox potential with bicarbonate replacement fluid during hemofiltration. Important confounding issues to consider are that patients receiving hemofiltration were in controlled, monitored settings and thus received more intensive care than the hydration group, and that serum creatinine, the major outcome parameter, is cleared by hemofiltration. Before hemofiltration can be recommended as routine prophylactic therapy for CN, the data will need to be corroborated by other groups, preferably involving larger numbers of study subjects and including cost‐benefit analyses.

Multiple small studies have examined the possibility that dialysis immediately following radiocontrast exposure could prevent CN, presumably by accelerating radiocontrast clearance. Most of these reports were negative, including a well‐designed meta‐analysis of RCTs, which showed no benefit of hemodialysis.53 Of note, one report suggested that hemodialysis might be potentially harmful.54 The single prospective trial that showed benefit from prophylactic hemodialysis analyzed 82 patients with advanced CRI (baseline creatinine clearance 13 mL/minute) referred for coronary angiography55 These subjects were randomly assigned to intravenous 0.9% NaCl and hemodialysis vs. intravenous 0.9% NaCl alone. Subsequent renal replacement therapy was required in 35% of control patients and in only 2% in the prophylactic dialysis group. One potential limitation of this study is that the investigators were more cognizant of volume status in the hemodialysis group to avoid fluid shifts and volume depletion during dialysis, while the control group appeared to experience no comparable intravascular volume management. Moreover, this study was conducted in patients with extremely advanced renal insufficiency, and therefore does not reflect the vast majority of patients at risk for CN.

Conclusions

CN is associated with increased morbidity and mortality, and efforts to minimize CN are therefore warranted. However, the overwhelming majority of CN trials were designed to investigate the effects of prophylaxis strategies on surrogate endpoints for estimates of GFR. Therefore, conclusions regarding the effect of these regimens on definitive outcomes, such as death and vascular events, cannot be drawn. On balance, there is evidence that oral and intravenous NAC, as well as extracellular volume expansion with intravenous NaHCO₃ are effective measures to prevent CN, whereas the data for renal replacement therapies are more equivocal. We emphasize though, that the literature on this topic is vast, and includes a large number of conflicting studies, including multiple meta‐analyses. As a result, we refrain from being too dogmatic about the best approach, and therefore cautiously offer the following recommendations for prevention of CN.

The first step is to identify high‐risk patients, who are most likely to benefit from prophylaxis (Table 1). Although risk stratification was not the focus of this review, Mehran et al.56 have developed a scoring system to quantitatively predict CN risk, with weighted parameters including CRI, diabetes, radiocontrast volume, age, hypotension, congestive heart failure, treatment with an intraaortic balloon pump, and anemia. For low‐risk patients, hydration with saline is probably adequate. For high‐risk patients, it would be prudent to initially consider whether sufficient information could be obtained from an alternative, noncontrasted radiologic procedure. If not, it would behoove the prescribing physician to then treat modifiable risk factors, as well as to discontinue potentially nephrotoxic medications.

In high‐risk patients undergoing radiocontrast procedures, we recommend using NAC and volume expansion with NaHCO₃ (Table 2). Although the evidence for this combined approach is limited,43 we believe it is biologically consistent, since the rationale for both strategies is primarily modification of redox state and inhibition of oxygen free radical generation. Because NAC formulations are generally effective, safe and inexpensive ($0.04 for 600 mg oral NAC and $24 for 1200 mg intravenous NAC at our hospital), we recommend the protocol used by Marenzi et al.,39 NAC 1200 mg intravenously once before procedure and then 1200 mg orally twice daily for 48 hours, as prophylaxis for all contrast procedures in high‐risk patients. However, we recognize that this regimen would require formal evaluation in procedures other than emergent coronary artery angioplasty before it could be enthusiastically endorsed. Therefore, if intravenous NAC is not available and/or the procedure is not emergent, NAC 6001200 mg orally twice a day, 2 doses before and 2 doses after the procedure would be a rational alternative. For the NaHCO₃ infusion, we recommend 3 mL/kg for 1 hour before the procedure, followed by 1 mL/kg/hour for 6 hours after.

Hemofiltration is labor‐intensive, expensive, and not readily available in all hospitals that renders it difficult to endorse as a definitive or routine CN prophylaxis modality. However, if a patient is already undergoing acute dialysis with catheter vascular access, it would be reasonable to consider CVVH 6 hours before and for 24 hours after the procedure (Table 2).

For high‐risk patients, we recommend minimizing the radiocontrast dose (reviewed in Ref.57). Although the dose has not consistently been identified as a risk factor (Table 1), we envision no harm in reducing the dose, particularly if adequate information can be obtained by other means, eg, coronary angiogram accompanied by an echocardiogram, rather than a ventriculogram. We would also consider the use of isoosmolar media in high‐risk patients, since the data are relatively compelling.20 Low doses of isoosmolar media should be particularly beneficial to patients with preexisting hypertension or congestive heart failure, for which the osmotic load and excess extracellular volume expansion might be deleterious. However, because isoosmolar media is expensive, a detailed cost‐benefit analysis would be required before definitive recommendations could be made, especially for patients at lower risk for CN. Finally, because of the complex literature, as well as budgetary issues, we encourage communication between the physician ordering the contrast study and the operator (radiologist or cardiologist) concerning the type of procedure and contrast media to be used.

Since contrast nephropathy (CN) was recognized more than 50 years ago,1 there have been continuous efforts to chemically modify radiocontrast agents to be less nephrotoxic. Although radiocontrast media have indeed become safer, which reduces the likelihood of CN per procedure, the indications for radiocontrast administration have dramatically increased, since over 80 million doses are delivered in the world annually.2, 3 Furthermore, the number of patients with CN risks, which are mainly chronic renal insufficiency (CRI) and diabetes (Table 1), has also grown. Currently, more than 26 million people are estimated to have CRI in the United States4 and 200 million people have diabetes worldwide.5 The combination of increased radiocontrast administration frequency and greater prevalence of at‐risk patients is likely to result in continued increases in CN events.