Bradley A. Sharpe, MD

Nurse

The nurse reviews the medication list, notices antibiotics and bronchodilators, and wonders why aren't we administering steroids for his COPD? Do any of these medications need to be given now? Is there anything the physician is worried about? What specific things should prompt me to call the physician with an update or change in condition? I'm not sure if it's safe to send the patient down for the ordered radiographic study because he still looks pretty short of breath. I hate paging the physician several times to get these questions answered because I know that person is busy as well. I also know the patient will have questions about the care plans, which I won't be able to answer. I wonder if I should finish administering evening medications for my other patients as I'm running behind schedule on my other tasks.

Respiratory therapist

At the same time, the respiratory therapist (RT) is contacted to assist with nebulizer therapy for the patient. In reviewing the order for bronchodilators, the RT silently asks, do we think he is going to need continuous nebulizers? What is our oxygen saturation goaldo we want him at 90% or above 95%? I wonder if this patient has a history of CO² retention and if I should have a BiPAP machine at the bedside.

Physician

After completing the orders for the patient, the physician remains in the emergency department to admit a different patient with a gastrointestinal bleed. This is the fifth admission in the past few hours. The physician feels the impact of constant paging interruptions. A unit clerk pages asking for clarification about a radiographic study that was ordered. A bedside nurse pages and asks if the physician can come and speak to the family about the diagnosis and treatment plans for an earlier admission (something the nurse is not clear about, either). A second bedside nurse pages, stating a different admission is still tachycardic after 3 liters of intravenous fluids and wants to know whether the fluids should be continued. Finally, the bedside nurse pages about whether the new COPD admission can go off the floor for the ordered chest CT or remain on continuous pulse oximetry because of shortness of breath.

Our case scenario is representative of most non‐surgical admissions to a hospital. The hypothetical questions posed from different provider perspectives are also common and often remain unanswered in a timely fashion. Partly because there is no site to mark and no anesthesia to deliver, the clinical encounter escapes attention as an opportunity for error prevention. In our experience, there are specific times during a hospitalization when communication failures are most likely to compromise patient care: the time of admission, the time of discharge,9 and any time when a patient's clinical condition changes acutely. Whereas handoff communications focus on transitions between providers (eg, shift changes), these circumstances are driven by patient transitions. Indirect communications, such as phone, email, or faxes, are suboptimal forms of communication at such times.10 We believe that there should be an expectation for direct communication at these junctures, and we define these direct communications as Critical Conversations.

Making Critical Conversations Happen

Integrating Critical Conversations into practice requires both buy‐in among providers and a plan for monitoring the interactions. We recommend beginning with educational efforts (eg, at a physician or nurse staff meeting) and reinforcing them with visual cues, such as posters on the unit (Figure 1). These actions promote awareness and generate expectations that this new clinical policy is being supported by clinical and hospital leadership. Our experiences have demonstrated tremendous learning, including numerous anecdotes about the value of Critical Conversations (Table 3). Our implementation efforts also raised a number of questions that ultimately led to improved clarity in later iterations.

Figure 1

Who should be involved in a Critical Conversation?

Identifying which healthcare team members should be involved in Critical Conversations is best determined by the conversation owner. That is, we found communication was most effective when the individual initiating the Critical Conversation directed others who needed to be involved. At admission, the physician writing the admission orders is best suited to make this determination; at a minimum, he or she should engage the bedside nurse but, as in the case example presented, the physician may also need to engage other services in particularly complex situations (eg, respiratory therapy, pharmacy). At time of discharge, there should be a physiciannurse Critical Conversation; however, the owner of the discharge process may determine that other conversations should occur, and this may be inclusive of or driven by a case manager or social worker. Because local culture and practices may drive specific ownership, it's key to outline a protocol for how this should occur. For instance, at admission, we asked the admitting physicians to take responsibility in contacting the bedside nurse. In other venues, this may work more effectively if the bedside nurse pages the physician once the orders are received and reviewed.

Conclusions

We introduced Critical Conversations as an innovative tool and policy that promotes communication and teamwork in a structured format and at a consistent time. Developing formal systems that decrease communication failures in high‐risk circumstances remains a focus in patient safety, evidenced by guidelines for TOs in procedural settings, handoffs in patient care (eg, sign‐out between providers),14, 15 and transitions into and from the hospital setting.16 Furthermore, there is growing evidence that such structured times for communication and teamwork, such as with briefings, can improve efficiency and reduce delays in care.17, 18 However, handoffs, which address provider transitions, and daily multidisciplinary rounds, which bring providers together regularly, are provider‐centered rather than patient‐centered. Critical Conversations focus on times when patients require direct communication about their care plans to ensure safe and high quality outcomes.

Implementation of Critical Conversations provides an opportunity to codify a professional standard for patient‐centered communication at times when it should be expected. Critical Conversations also help build a system that supports a positive safety culture and encourages teamwork and direct communication. This is particularly true at a time when rapid adoption of information technology may have the unintended and opposite effect. For instance, as our hospital moved toward an entirely electronic health record, providers were increasingly relocating from patient care units into remote offices, corner hideaways, or designated computer rooms to complete orders and documentation. Although this may reduce many related errors in these processes and potentially improve communication via shared access to an electronic record, it does allow for less direct communicationa circumstance that traditionally occurs (even informally) when providers share the same clinical work areas. This situation is aggravated where the nurses are unit‐based and other providers (eg, physicians, therapists, case managers) are service‐based.

Integrating Critical Conversations into practice comes with expected challenges, most notably around workflow (eg, adds a step, although may save steps down the line) and the expectations concomitant with any change in standard of care (possible enforcement or auditing of their occurrence). Certain cultural barriers may also play a significant role, such as the presence of hierarchies that can hinder open communication and the related ability to speak up with concerns, as related in the TO literature. Where these cultural barriers highlight historical descriptions of the doctornurse relationship and its effect on patient care,1921 Critical Conversations provide an opportunity to improve such interdisciplinary relationships by providing a shared tool for direct communication.

In summary, we described an innovative communication tool that promotes direct communication at critical junctures during a hospitalization. With the growing complexity of hospital care and greater interdependence between teams that deliver this care, Critical Conversations provide an opportunity to further address the known communication failures that contribute to medical errors.

Nurse

The nurse reviews the medication list, notices antibiotics and bronchodilators, and wonders why aren't we administering steroids for his COPD? Do any of these medications need to be given now? Is there anything the physician is worried about? What specific things should prompt me to call the physician with an update or change in condition? I'm not sure if it's safe to send the patient down for the ordered radiographic study because he still looks pretty short of breath. I hate paging the physician several times to get these questions answered because I know that person is busy as well. I also know the patient will have questions about the care plans, which I won't be able to answer. I wonder if I should finish administering evening medications for my other patients as I'm running behind schedule on my other tasks.

Respiratory therapist

At the same time, the respiratory therapist (RT) is contacted to assist with nebulizer therapy for the patient. In reviewing the order for bronchodilators, the RT silently asks, do we think he is going to need continuous nebulizers? What is our oxygen saturation goaldo we want him at 90% or above 95%? I wonder if this patient has a history of CO² retention and if I should have a BiPAP machine at the bedside.

Physician

After completing the orders for the patient, the physician remains in the emergency department to admit a different patient with a gastrointestinal bleed. This is the fifth admission in the past few hours. The physician feels the impact of constant paging interruptions. A unit clerk pages asking for clarification about a radiographic study that was ordered. A bedside nurse pages and asks if the physician can come and speak to the family about the diagnosis and treatment plans for an earlier admission (something the nurse is not clear about, either). A second bedside nurse pages, stating a different admission is still tachycardic after 3 liters of intravenous fluids and wants to know whether the fluids should be continued. Finally, the bedside nurse pages about whether the new COPD admission can go off the floor for the ordered chest CT or remain on continuous pulse oximetry because of shortness of breath.

Our case scenario is representative of most non‐surgical admissions to a hospital. The hypothetical questions posed from different provider perspectives are also common and often remain unanswered in a timely fashion. Partly because there is no site to mark and no anesthesia to deliver, the clinical encounter escapes attention as an opportunity for error prevention. In our experience, there are specific times during a hospitalization when communication failures are most likely to compromise patient care: the time of admission, the time of discharge,9 and any time when a patient's clinical condition changes acutely. Whereas handoff communications focus on transitions between providers (eg, shift changes), these circumstances are driven by patient transitions. Indirect communications, such as phone, email, or faxes, are suboptimal forms of communication at such times.10 We believe that there should be an expectation for direct communication at these junctures, and we define these direct communications as Critical Conversations.

Making Critical Conversations Happen

Integrating Critical Conversations into practice requires both buy‐in among providers and a plan for monitoring the interactions. We recommend beginning with educational efforts (eg, at a physician or nurse staff meeting) and reinforcing them with visual cues, such as posters on the unit (Figure 1). These actions promote awareness and generate expectations that this new clinical policy is being supported by clinical and hospital leadership. Our experiences have demonstrated tremendous learning, including numerous anecdotes about the value of Critical Conversations (Table 3). Our implementation efforts also raised a number of questions that ultimately led to improved clarity in later iterations.

Figure 1

Who should be involved in a Critical Conversation?

Identifying which healthcare team members should be involved in Critical Conversations is best determined by the conversation owner. That is, we found communication was most effective when the individual initiating the Critical Conversation directed others who needed to be involved. At admission, the physician writing the admission orders is best suited to make this determination; at a minimum, he or she should engage the bedside nurse but, as in the case example presented, the physician may also need to engage other services in particularly complex situations (eg, respiratory therapy, pharmacy). At time of discharge, there should be a physiciannurse Critical Conversation; however, the owner of the discharge process may determine that other conversations should occur, and this may be inclusive of or driven by a case manager or social worker. Because local culture and practices may drive specific ownership, it's key to outline a protocol for how this should occur. For instance, at admission, we asked the admitting physicians to take responsibility in contacting the bedside nurse. In other venues, this may work more effectively if the bedside nurse pages the physician once the orders are received and reviewed.

Conclusions

We introduced Critical Conversations as an innovative tool and policy that promotes communication and teamwork in a structured format and at a consistent time. Developing formal systems that decrease communication failures in high‐risk circumstances remains a focus in patient safety, evidenced by guidelines for TOs in procedural settings, handoffs in patient care (eg, sign‐out between providers),14, 15 and transitions into and from the hospital setting.16 Furthermore, there is growing evidence that such structured times for communication and teamwork, such as with briefings, can improve efficiency and reduce delays in care.17, 18 However, handoffs, which address provider transitions, and daily multidisciplinary rounds, which bring providers together regularly, are provider‐centered rather than patient‐centered. Critical Conversations focus on times when patients require direct communication about their care plans to ensure safe and high quality outcomes.

Implementation of Critical Conversations provides an opportunity to codify a professional standard for patient‐centered communication at times when it should be expected. Critical Conversations also help build a system that supports a positive safety culture and encourages teamwork and direct communication. This is particularly true at a time when rapid adoption of information technology may have the unintended and opposite effect. For instance, as our hospital moved toward an entirely electronic health record, providers were increasingly relocating from patient care units into remote offices, corner hideaways, or designated computer rooms to complete orders and documentation. Although this may reduce many related errors in these processes and potentially improve communication via shared access to an electronic record, it does allow for less direct communicationa circumstance that traditionally occurs (even informally) when providers share the same clinical work areas. This situation is aggravated where the nurses are unit‐based and other providers (eg, physicians, therapists, case managers) are service‐based.

Integrating Critical Conversations into practice comes with expected challenges, most notably around workflow (eg, adds a step, although may save steps down the line) and the expectations concomitant with any change in standard of care (possible enforcement or auditing of their occurrence). Certain cultural barriers may also play a significant role, such as the presence of hierarchies that can hinder open communication and the related ability to speak up with concerns, as related in the TO literature. Where these cultural barriers highlight historical descriptions of the doctornurse relationship and its effect on patient care,1921 Critical Conversations provide an opportunity to improve such interdisciplinary relationships by providing a shared tool for direct communication.

In summary, we described an innovative communication tool that promotes direct communication at critical junctures during a hospitalization. With the growing complexity of hospital care and greater interdependence between teams that deliver this care, Critical Conversations provide an opportunity to further address the known communication failures that contribute to medical errors.

Nurse

The nurse reviews the medication list, notices antibiotics and bronchodilators, and wonders why aren't we administering steroids for his COPD? Do any of these medications need to be given now? Is there anything the physician is worried about? What specific things should prompt me to call the physician with an update or change in condition? I'm not sure if it's safe to send the patient down for the ordered radiographic study because he still looks pretty short of breath. I hate paging the physician several times to get these questions answered because I know that person is busy as well. I also know the patient will have questions about the care plans, which I won't be able to answer. I wonder if I should finish administering evening medications for my other patients as I'm running behind schedule on my other tasks.

Respiratory therapist

At the same time, the respiratory therapist (RT) is contacted to assist with nebulizer therapy for the patient. In reviewing the order for bronchodilators, the RT silently asks, do we think he is going to need continuous nebulizers? What is our oxygen saturation goaldo we want him at 90% or above 95%? I wonder if this patient has a history of CO² retention and if I should have a BiPAP machine at the bedside.

Physician

After completing the orders for the patient, the physician remains in the emergency department to admit a different patient with a gastrointestinal bleed. This is the fifth admission in the past few hours. The physician feels the impact of constant paging interruptions. A unit clerk pages asking for clarification about a radiographic study that was ordered. A bedside nurse pages and asks if the physician can come and speak to the family about the diagnosis and treatment plans for an earlier admission (something the nurse is not clear about, either). A second bedside nurse pages, stating a different admission is still tachycardic after 3 liters of intravenous fluids and wants to know whether the fluids should be continued. Finally, the bedside nurse pages about whether the new COPD admission can go off the floor for the ordered chest CT or remain on continuous pulse oximetry because of shortness of breath.

Our case scenario is representative of most non‐surgical admissions to a hospital. The hypothetical questions posed from different provider perspectives are also common and often remain unanswered in a timely fashion. Partly because there is no site to mark and no anesthesia to deliver, the clinical encounter escapes attention as an opportunity for error prevention. In our experience, there are specific times during a hospitalization when communication failures are most likely to compromise patient care: the time of admission, the time of discharge,9 and any time when a patient's clinical condition changes acutely. Whereas handoff communications focus on transitions between providers (eg, shift changes), these circumstances are driven by patient transitions. Indirect communications, such as phone, email, or faxes, are suboptimal forms of communication at such times.10 We believe that there should be an expectation for direct communication at these junctures, and we define these direct communications as Critical Conversations.

Making Critical Conversations Happen

Integrating Critical Conversations into practice requires both buy‐in among providers and a plan for monitoring the interactions. We recommend beginning with educational efforts (eg, at a physician or nurse staff meeting) and reinforcing them with visual cues, such as posters on the unit (Figure 1). These actions promote awareness and generate expectations that this new clinical policy is being supported by clinical and hospital leadership. Our experiences have demonstrated tremendous learning, including numerous anecdotes about the value of Critical Conversations (Table 3). Our implementation efforts also raised a number of questions that ultimately led to improved clarity in later iterations.

Figure 1

Who should be involved in a Critical Conversation?

Identifying which healthcare team members should be involved in Critical Conversations is best determined by the conversation owner. That is, we found communication was most effective when the individual initiating the Critical Conversation directed others who needed to be involved. At admission, the physician writing the admission orders is best suited to make this determination; at a minimum, he or she should engage the bedside nurse but, as in the case example presented, the physician may also need to engage other services in particularly complex situations (eg, respiratory therapy, pharmacy). At time of discharge, there should be a physiciannurse Critical Conversation; however, the owner of the discharge process may determine that other conversations should occur, and this may be inclusive of or driven by a case manager or social worker. Because local culture and practices may drive specific ownership, it's key to outline a protocol for how this should occur. For instance, at admission, we asked the admitting physicians to take responsibility in contacting the bedside nurse. In other venues, this may work more effectively if the bedside nurse pages the physician once the orders are received and reviewed.

Conclusions

We introduced Critical Conversations as an innovative tool and policy that promotes communication and teamwork in a structured format and at a consistent time. Developing formal systems that decrease communication failures in high‐risk circumstances remains a focus in patient safety, evidenced by guidelines for TOs in procedural settings, handoffs in patient care (eg, sign‐out between providers),14, 15 and transitions into and from the hospital setting.16 Furthermore, there is growing evidence that such structured times for communication and teamwork, such as with briefings, can improve efficiency and reduce delays in care.17, 18 However, handoffs, which address provider transitions, and daily multidisciplinary rounds, which bring providers together regularly, are provider‐centered rather than patient‐centered. Critical Conversations focus on times when patients require direct communication about their care plans to ensure safe and high quality outcomes.

Implementation of Critical Conversations provides an opportunity to codify a professional standard for patient‐centered communication at times when it should be expected. Critical Conversations also help build a system that supports a positive safety culture and encourages teamwork and direct communication. This is particularly true at a time when rapid adoption of information technology may have the unintended and opposite effect. For instance, as our hospital moved toward an entirely electronic health record, providers were increasingly relocating from patient care units into remote offices, corner hideaways, or designated computer rooms to complete orders and documentation. Although this may reduce many related errors in these processes and potentially improve communication via shared access to an electronic record, it does allow for less direct communicationa circumstance that traditionally occurs (even informally) when providers share the same clinical work areas. This situation is aggravated where the nurses are unit‐based and other providers (eg, physicians, therapists, case managers) are service‐based.

Integrating Critical Conversations into practice comes with expected challenges, most notably around workflow (eg, adds a step, although may save steps down the line) and the expectations concomitant with any change in standard of care (possible enforcement or auditing of their occurrence). Certain cultural barriers may also play a significant role, such as the presence of hierarchies that can hinder open communication and the related ability to speak up with concerns, as related in the TO literature. Where these cultural barriers highlight historical descriptions of the doctornurse relationship and its effect on patient care,1921 Critical Conversations provide an opportunity to improve such interdisciplinary relationships by providing a shared tool for direct communication.

In summary, we described an innovative communication tool that promotes direct communication at critical junctures during a hospitalization. With the growing complexity of hospital care and greater interdependence between teams that deliver this care, Critical Conversations provide an opportunity to further address the known communication failures that contribute to medical errors.

Problem Identification

The University of California, San Francisco (UCSF) Medical Center operates nearly 800 beds across 2 hospitals (Parnassus and Mount Zion campuses). The UCSF Division of Hospital Medicine (DHM) provides care on the teaching service (90% of all ward months covered by a hospitalist faculty), a nonhousestaff medical service based at Mount Zion,4 a palliative care service,10 a medical consultation service, a neurosurgical comanagement service, a procedure service, and comanagement on advanced heart failure and cancer services. Like many AHM groups, ours has experienced explosive growth, more than doubling in faculty size in 3 years (50+ faculty by July 2010).

In addition, many of our new faculty joined the division directly after residency training whereas our early hospitalists were mostly former chief residents and/or fellowship‐trained. During a 2‐year period, our division lost several faculty to burnout from clinically heavy positions or because they felt their ultimate academic success was in doubt. During a 2008 divisional retreat, the single greatest need identified was to invest in the development of our first‐year faculty who were felt to be at greatest risk for burnout, dissatisfaction, and failing to integrate into the divisional mission. Based on this result, we set out to develop a program to meet this pressing need.

Needs Assessment

We formed a FD steering committee comprised of faculty from all ranks and career paths in our division (eg, educator, administrator, and investigator), with overrepresentation of recent hires to ascertain how best to meet their needs. Information from the division retreat provided the basis for the program and its priorities. The FD steering committee then outlined ideas that guided program development, which included:

• 1

  New faculty should be required to meet regularly with assigned faculty mentors during their first year, and expectations for that relationship should be outlined for both parties

• 2

  New faculty should be required to attend dedicated sessions that build their teaching skills

• 3

  New faculty should receive a specially designed first year curriculum to provide learnings focused on high‐yield and relevant topics

• 4

  New faculty should receive a set of goals, or scholarly expectations, for their first year that would foster a partnership between individual faculty and the division to meet those goals

• 5

  The division should create new structures for FD that promote collaboration, sharing of personal and professional growth and challenges, and a culture of continuous learning

• 6

  All of the activities that comprise our new FD program must be aligned with our stated mission: to provide the highest quality clinical care, education, system improvements, and research that benefit our patients and trainees by developing successful academic hospitalist faculty.

Program Goals and Objectives

Our DHM FD program established the goal to provide our new faculty with clinical, educational, leadership, and scholarly skills that would promote academic output and foster work satisfaction. From a broader divisional standpoint, the goal was simply to create new FD structures that fostered the division's commitment to the program. The primary objectives of the program were for new faculty to:

• Increase their knowledge, skills, and attitudes about key academic hospitalist domains following participation in the program;

• Demonstrate successful production of scholarly output, participation in a hospital committee, and participation in a quality or safety improvement initiative by the end of their first year;

• Report high levels of satisfaction with the FD program and their first year on faculty.

Program Development Principles

We began by conducting a literature review to draw on the successes and lessons learned from existing FD programs, particularly in large departments, academic centers, and the hospitalist field.1115 We focused our program development on a set of FD principles, which included instructional improvement, organizational development, the development of professional academic skills, and the teaching of specific content.11 Furthermore, whereas many FD programs traditionally focus on mentoring or a longitudinal set of seminars, we believed a multifaceted approach could help shift our culture towards one that prioritized FD and generated a sense of community. We hoped this cultural shift would create an environment that increased faculty satisfaction with their work, with their colleagues, and in our division.

This context drove us to build programmatic activities that not only targeted new faculty, the initial focus of our planning efforts, but also the division more broadly. We wanted to adopt known strategies (eg, mentoring relationships, teaching methods for FD, and grand rounds) but also weave in new ones that targeted AHM and our Division. It was clear that successful programs used a variety of instructional methods, and often combined methods, to create active and engaged faculty. We similarly wanted to create venues for didactic and small‐group learning, but also opportunities for peer learning and facilitated discussions around important topics. Allowing new faculty to learn from each other, and having them observe more senior faculty do the same, would be an important and explicit programmatic element.

Program Description and Implementation

All new faculty meet with Divisional leadership (RMW/BAS), administrative staff (they receive an orientation binder that highlights frequently asked questions and provides service‐specific orientation documents), and the Director of FD (NLS). The latter introduces the DHM FD Program and provides the road map for their first year (Supporting Information). The checklist serves to orient, guide, and emphasize the various programmatic goals, expectations, and logistics. Discussion focuses on the activities targeted to new faculty followed by wider divisional offerings. New faculty activities include:

Program Evaluation

Our evaluation focused on measuring the FD program's impact on our new faculty. We tracked their success in completing the stated scholarly expectations and surveyed them about their satisfaction with the programmatic activities, their first year on faculty, and their preparation for year 2. Prior to implementing the program, we surveyed the previous 2 years of new faculty to provide a comparison.

Problem Identification

The University of California, San Francisco (UCSF) Medical Center operates nearly 800 beds across 2 hospitals (Parnassus and Mount Zion campuses). The UCSF Division of Hospital Medicine (DHM) provides care on the teaching service (90% of all ward months covered by a hospitalist faculty), a nonhousestaff medical service based at Mount Zion,4 a palliative care service,10 a medical consultation service, a neurosurgical comanagement service, a procedure service, and comanagement on advanced heart failure and cancer services. Like many AHM groups, ours has experienced explosive growth, more than doubling in faculty size in 3 years (50+ faculty by July 2010).

In addition, many of our new faculty joined the division directly after residency training whereas our early hospitalists were mostly former chief residents and/or fellowship‐trained. During a 2‐year period, our division lost several faculty to burnout from clinically heavy positions or because they felt their ultimate academic success was in doubt. During a 2008 divisional retreat, the single greatest need identified was to invest in the development of our first‐year faculty who were felt to be at greatest risk for burnout, dissatisfaction, and failing to integrate into the divisional mission. Based on this result, we set out to develop a program to meet this pressing need.

Needs Assessment

We formed a FD steering committee comprised of faculty from all ranks and career paths in our division (eg, educator, administrator, and investigator), with overrepresentation of recent hires to ascertain how best to meet their needs. Information from the division retreat provided the basis for the program and its priorities. The FD steering committee then outlined ideas that guided program development, which included:

• 1

  New faculty should be required to meet regularly with assigned faculty mentors during their first year, and expectations for that relationship should be outlined for both parties

• 2

  New faculty should be required to attend dedicated sessions that build their teaching skills

• 3

  New faculty should receive a specially designed first year curriculum to provide learnings focused on high‐yield and relevant topics

• 4

  New faculty should receive a set of goals, or scholarly expectations, for their first year that would foster a partnership between individual faculty and the division to meet those goals

• 5

  The division should create new structures for FD that promote collaboration, sharing of personal and professional growth and challenges, and a culture of continuous learning

• 6

  All of the activities that comprise our new FD program must be aligned with our stated mission: to provide the highest quality clinical care, education, system improvements, and research that benefit our patients and trainees by developing successful academic hospitalist faculty.

Program Goals and Objectives

Our DHM FD program established the goal to provide our new faculty with clinical, educational, leadership, and scholarly skills that would promote academic output and foster work satisfaction. From a broader divisional standpoint, the goal was simply to create new FD structures that fostered the division's commitment to the program. The primary objectives of the program were for new faculty to:

• Increase their knowledge, skills, and attitudes about key academic hospitalist domains following participation in the program;

• Demonstrate successful production of scholarly output, participation in a hospital committee, and participation in a quality or safety improvement initiative by the end of their first year;

• Report high levels of satisfaction with the FD program and their first year on faculty.

Program Development Principles

We began by conducting a literature review to draw on the successes and lessons learned from existing FD programs, particularly in large departments, academic centers, and the hospitalist field.1115 We focused our program development on a set of FD principles, which included instructional improvement, organizational development, the development of professional academic skills, and the teaching of specific content.11 Furthermore, whereas many FD programs traditionally focus on mentoring or a longitudinal set of seminars, we believed a multifaceted approach could help shift our culture towards one that prioritized FD and generated a sense of community. We hoped this cultural shift would create an environment that increased faculty satisfaction with their work, with their colleagues, and in our division.

This context drove us to build programmatic activities that not only targeted new faculty, the initial focus of our planning efforts, but also the division more broadly. We wanted to adopt known strategies (eg, mentoring relationships, teaching methods for FD, and grand rounds) but also weave in new ones that targeted AHM and our Division. It was clear that successful programs used a variety of instructional methods, and often combined methods, to create active and engaged faculty. We similarly wanted to create venues for didactic and small‐group learning, but also opportunities for peer learning and facilitated discussions around important topics. Allowing new faculty to learn from each other, and having them observe more senior faculty do the same, would be an important and explicit programmatic element.

Program Description and Implementation

All new faculty meet with Divisional leadership (RMW/BAS), administrative staff (they receive an orientation binder that highlights frequently asked questions and provides service‐specific orientation documents), and the Director of FD (NLS). The latter introduces the DHM FD Program and provides the road map for their first year (Supporting Information). The checklist serves to orient, guide, and emphasize the various programmatic goals, expectations, and logistics. Discussion focuses on the activities targeted to new faculty followed by wider divisional offerings. New faculty activities include:

Program Evaluation

Our evaluation focused on measuring the FD program's impact on our new faculty. We tracked their success in completing the stated scholarly expectations and surveyed them about their satisfaction with the programmatic activities, their first year on faculty, and their preparation for year 2. Prior to implementing the program, we surveyed the previous 2 years of new faculty to provide a comparison.

Problem Identification

The University of California, San Francisco (UCSF) Medical Center operates nearly 800 beds across 2 hospitals (Parnassus and Mount Zion campuses). The UCSF Division of Hospital Medicine (DHM) provides care on the teaching service (90% of all ward months covered by a hospitalist faculty), a nonhousestaff medical service based at Mount Zion,4 a palliative care service,10 a medical consultation service, a neurosurgical comanagement service, a procedure service, and comanagement on advanced heart failure and cancer services. Like many AHM groups, ours has experienced explosive growth, more than doubling in faculty size in 3 years (50+ faculty by July 2010).

In addition, many of our new faculty joined the division directly after residency training whereas our early hospitalists were mostly former chief residents and/or fellowship‐trained. During a 2‐year period, our division lost several faculty to burnout from clinically heavy positions or because they felt their ultimate academic success was in doubt. During a 2008 divisional retreat, the single greatest need identified was to invest in the development of our first‐year faculty who were felt to be at greatest risk for burnout, dissatisfaction, and failing to integrate into the divisional mission. Based on this result, we set out to develop a program to meet this pressing need.

Needs Assessment

We formed a FD steering committee comprised of faculty from all ranks and career paths in our division (eg, educator, administrator, and investigator), with overrepresentation of recent hires to ascertain how best to meet their needs. Information from the division retreat provided the basis for the program and its priorities. The FD steering committee then outlined ideas that guided program development, which included:

• 1

  New faculty should be required to meet regularly with assigned faculty mentors during their first year, and expectations for that relationship should be outlined for both parties

• 2

  New faculty should be required to attend dedicated sessions that build their teaching skills

• 3

  New faculty should receive a specially designed first year curriculum to provide learnings focused on high‐yield and relevant topics

• 4

  New faculty should receive a set of goals, or scholarly expectations, for their first year that would foster a partnership between individual faculty and the division to meet those goals

• 5

  The division should create new structures for FD that promote collaboration, sharing of personal and professional growth and challenges, and a culture of continuous learning

• 6

  All of the activities that comprise our new FD program must be aligned with our stated mission: to provide the highest quality clinical care, education, system improvements, and research that benefit our patients and trainees by developing successful academic hospitalist faculty.

Program Goals and Objectives

Our DHM FD program established the goal to provide our new faculty with clinical, educational, leadership, and scholarly skills that would promote academic output and foster work satisfaction. From a broader divisional standpoint, the goal was simply to create new FD structures that fostered the division's commitment to the program. The primary objectives of the program were for new faculty to:

• Increase their knowledge, skills, and attitudes about key academic hospitalist domains following participation in the program;

• Demonstrate successful production of scholarly output, participation in a hospital committee, and participation in a quality or safety improvement initiative by the end of their first year;

• Report high levels of satisfaction with the FD program and their first year on faculty.

Program Development Principles

We began by conducting a literature review to draw on the successes and lessons learned from existing FD programs, particularly in large departments, academic centers, and the hospitalist field.1115 We focused our program development on a set of FD principles, which included instructional improvement, organizational development, the development of professional academic skills, and the teaching of specific content.11 Furthermore, whereas many FD programs traditionally focus on mentoring or a longitudinal set of seminars, we believed a multifaceted approach could help shift our culture towards one that prioritized FD and generated a sense of community. We hoped this cultural shift would create an environment that increased faculty satisfaction with their work, with their colleagues, and in our division.

This context drove us to build programmatic activities that not only targeted new faculty, the initial focus of our planning efforts, but also the division more broadly. We wanted to adopt known strategies (eg, mentoring relationships, teaching methods for FD, and grand rounds) but also weave in new ones that targeted AHM and our Division. It was clear that successful programs used a variety of instructional methods, and often combined methods, to create active and engaged faculty. We similarly wanted to create venues for didactic and small‐group learning, but also opportunities for peer learning and facilitated discussions around important topics. Allowing new faculty to learn from each other, and having them observe more senior faculty do the same, would be an important and explicit programmatic element.

Program Description and Implementation

All new faculty meet with Divisional leadership (RMW/BAS), administrative staff (they receive an orientation binder that highlights frequently asked questions and provides service‐specific orientation documents), and the Director of FD (NLS). The latter introduces the DHM FD Program and provides the road map for their first year (Supporting Information). The checklist serves to orient, guide, and emphasize the various programmatic goals, expectations, and logistics. Discussion focuses on the activities targeted to new faculty followed by wider divisional offerings. New faculty activities include:

Program Evaluation

Our evaluation focused on measuring the FD program's impact on our new faculty. We tracked their success in completing the stated scholarly expectations and surveyed them about their satisfaction with the programmatic activities, their first year on faculty, and their preparation for year 2. Prior to implementing the program, we surveyed the previous 2 years of new faculty to provide a comparison.

Differential Diagnosis

Given the nonspecific nature of the symptoms and signs associated with CAP, there is no single clinical feature or combination of clinical features that adequately rules in or out the diagnosis of CAP. Consequently, the differential diagnosis to be considered in patients with suspected CAP is broad. Noninfectious diseases can often present with similar clinical syndromes; these include congestive heart failure, exacerbation of chronic obstructive pulmonary disease (COPD), asthma, pulmonary embolism, and hypersensitivity pneumonitis. These diseases can often be distinguished with a thorough history and physical examination.

In addition, other upper‐ and lower‐airway infectious diseases can have similar nonspecific signs and symptoms. In particular, pneumonia must often be differentiated from acute bronchitis, which as a diagnosis accounts for up to 40% of patients evaluated for cough (versus 5% for pneumonia).10 Patients with acute bronchitis frequently do not present with high fevers or hypoxia and in general will not benefit from antibiotic therapy.13 Patients believed to have community‐acquired pneumonia might also be suffering from other pneumonia syndromes including aspiration pneumonia, postobstructive pneumonia, and pneumonia in immunocompromised patients (eg, those with HIV, on steroids, receiving chemotherapy). Determining the correct diagnosis can have implications for therapy and prognosis.

Diagnostic Studies

The diagnosis of community‐acquired pneumonia requires that a patient have both signs and symptoms consistent with pulmonary infection and evidence of a new radiographic infiltrate. Therefore, most guidelines recommend that all patients with a possible diagnosis of CAP be evaluated with chest radiography.1, 14, 15

The specific radiographic findings in community‐acquired pneumonia range from lobar consolidation to hazy focal infiltrate to diffuse bilateral interstitial opacities (see Figure 1). Although chest radiography has traditionally been considered the gold standard for the diagnosis of CAP, its exact performance characteristics are unknown, and it is clearly not 100% sensitive or 100% specific. The utility of the chest radiograph can be limited by patient body habitus, underlying lung disease, or dehydration. Computed tomography (CT) scanning, although not recommended for routine use, can identify pulmonary consolidation in up to 30% of patients with a normal or equivocal chest radiograph in whom pneumonia is suspected and can also identify complications of pneumonia including an empyema or pulmonary abscess.16

Figure 1

Limitations in the performance of the chest radiograph have resulted in an interest in the diagnostic performance of serologic markers of infection such as C‐reactive protein (CRP), procalcitonin, and soluble triggering receptor expressed on myeloid cells (s‐TREM).1719 Preliminary evidence suggests these inflammatory markers may ultimately prove useful in differentiating infectious from noninfectious pulmonary processes, but regular use of these new tests cannot currently be recommended.

Most expert guidelines state that 2 sets of blood cultures should be taken and analyzed prior to antibiotic administration in all patients admitted to the hospital with suspected community‐acquired pneumonia.1, 14, 15 Isolation of bacteria from blood cultures in CAP is a very specific way to identify a causative organism in order to subsequently narrow therapy and also identifies a high‐risk group of patients because bacteremia is associated with increased mortality. Obtaining blood cultures within 24 hours of admission has been associated with 10% lower odds of 30‐day mortality in patients with CAP,20 and as a result, drawing blood cultures prior to antibiotic administration is a national quality indicator for CAP.

There are, however, a number of problems with the routine acquisition of blood cultures in all patients admitted with CAP. Practically, the cultures can be difficult to obtain, can potentially delay the initiation of antibiotics, and are often contaminated, which has been shown to increase both cost and length of stay.21, 22 The yield is generally low: the true‐positive bacteremia rate for admitted patients with CAP ranges from 6% to 9%, and the culture results rarely change management or outcomes.23, 24 Given these limitations, many have argued that blood cultures should be obtained with a more targeted approach. A recent study used a Medicare database to create a decision‐support tool to help maximize the value of blood cultures in CAP.25 The predictors of a positive blood culture are shown in Table 3. Not obtaining cultures on patients who had received prior antibiotics or had no risk factors resulted in about 40% fewer overall cultures while identifying approximately 90% of bacteremias. In their guidelines, the British Thoracic Society (BTS) advocates a similar strategy, recommending blood cultures be omitted in nonsevere pneumonia and in patients without comorbidities.15, 26 Although recommendations vary for non‐severe CAP in hospitalized patients, all professional society guidelines agree that blood cultures should be obtained in critically ill patients, and if cultures are obtained, they should be drawn prior to antibiotics.1, 14, 15, 26

Substantial controversy surrounds the utility of routine sputum gram stains and cultures for patients admitted to the hospital with CAP. The Infectious Disease Society of America (IDSA) and the British Thoracic Society (BTS) both recommend that all patients admitted to the hospital with community‐acquired pneumonia should have a gram stain and culture of expectorated sputum.1, 15, 26 Both organizations argue sputum collection is a simple and inexpensive procedure that can potentially identify pathogenic organisms and can affect both initial and long‐term antibiotic therapy. Most notably, they highlight gram stain specificity of greater than 80% for pneumococcal pneumonia. Conversely, the American Thoracic Society (ATS) argues that sputum gram stains and cultures generally have low sensitivity, specificity, and positive predictive value.14 Furthermore, they argue the utility of sputum testing is also limited practically; in one study 30% of patients could not produce an adequate sputum specimen and up to 30% had received prior antibiotic therapy, substantially reducing the yield.27 In another study, good‐quality sputum with a predominant morphotype could be obtained in only 14% of patients admitted with CAP.28 However, targeting sputum analysis to patients who have not received prior antibiotics and are able to produce an adequate sample improved the yield significantly.29 In addition, with increasing rates of antibiotic resistance among common community isolates (ie, S. Pneumoniae) and the increasing prevalence of infecting organisms not targeted by routine empiric therapy (methicillin‐resistant Staphylococcus Aureus [MRSA]), isolation of potential causative pathogens is increasingly important. We believe that severely ill patients with CAP (such as patients admitted to the ICU), as well as patients with identifiable risk factors for uncommon or drug‐resistant pathogens (eg, Pseudomonas aeruginosa, enteric gram‐negative rods, MRSA, etc.) should have sputum sent for gram stain and culture. Ideally, sputum obtained for gram stain and culture should be:

• 1

  Prior to antibiotic therapy,

• 2

  A deep‐cough, expectorated specimen,

• 3

  A purulent specimen (>25 polymorphonucleacytes and less than 10 squamous cells per high‐powered field), and

• 4

  Rapidly transported to the laboratory.

Subsequent gram stain and culture results should be interpreted in the specific clinical context and antibiotic choices targeted appropriately.

Alternative Diagnostic Tests

In recent years, there has been growth in additional diagnostic tests targeting specific organisms. The pneumococcal urinary antigen assay is a relatively sensitive (50%80%) and highly specific (90%) test for the detection of pneumococcal pneumonia, when compared with conventional diagnostic methods.27 The test is simple, convenient, rapid ( 15 min), and, with its high specificity, may allow for more focused antimicrobial therapy early in management. Current limitations include the possibility of false‐positive tests in patients colonized with S. pneumoniae or infected with other streptococcal species, as well as the inability to determine antibiotic sensitivity from positive tests. Updated IDSA and BTS guidelines state pneumococcal urinary antigen testing is an acceptable adjunct to other diagnostic tests, but blood and sputum analyses should still be performed.26, 27 For patients with suspected Legionella pneumonia (primarily critically ill and immunocompromised patients or in association with regional outbreaks), the urinary Legionella antigen assay is the test of choice, which detects 80%95% of community‐acquired cases of Legionnaires' disease with a specificity of 90%.27

During the winter months (typically from October to March), rapid antigen testing for influenza is generally recommended for patients with signs or symptoms consistent with influenza.27 The sensitivity of these tests is approximately 50%70%, so negative test results do not exclude the diagnosis, but results can be important epidemiologically and therapeutically (differentiating influenza A and B strains).27 Diagnostic tests targeting other common CAP pathogens, such as serologic tests for Mycoplasma pneumoniae or Chlamydia spp, should not be routinely performed. Testing for less common causative pathogens such as Mycobacterium tuberculosis should only be employed in the appropriate clinical setting.

Initial Treatment

Once the admission decision is made and the initial diagnostic tests are completed (including blood and sputum cultures), patients with presumed community‐acquired pneumonia should receive necessary supportive care (O₂, intravenous fluids, etc.) and prompt antimicrobial therapy. Antibiotics should be administered within 4 hours of arrival to patients with suspected CAP, as such prompt administration may be associated with shorter in‐hospital stays and decreased 30‐day mortality.34, 35 Regulatory organizations such as the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) and the Center for Medicare Services (CMS) have made delivery of antibiotics in less than 4 hours a hospital quality measure.

Despite diagnostic testing, the specific etiologic agent causing the pneumonia of a patient remains unknown in up to 75% of those admitted to the hospital.14 Most expert guidelines therefore recommend broad‐spectrum empiric therapy targeting both the typical and the atypical organisms that commonly cause CAP (Table 1).

Recommendations for empiric antibiotics are driven by 2 key factors: antibiotic resistance by S. pneumoniae and the results of studies of CAP treatment outcomes. Historically, patients with suspected community‐acquired pneumonia were treated with penicillin with generally good outcomes. Recently, the rate of S. pneumoniae isolates resistant to penicillin has risen dramatically in the United States, ranging between 20% and 30%, with high‐level resistance (MIC 4 mg/L) as high as 5.7%.36, 37 Concurrently, the rates of resistance of S. pneumoniae to many other antibiotics commonly used to treat CAP have also risen.37 Despite increasing resistance overall, most U.S. pneumococcal isolates have low resistance to third‐generation cephalosporins and fluoroquinolones with enhanced activity against S. pneumoniae.3638 In addition, despite increasing resistance by pneumococcal isolates to penicillin, several observational studies have shown that regardless of initial therapy, resistance to penicillin as well as third‐generation cephalosporins is not associated with higher mortality or worse outcomes when controlled for other risk factors for drug resistance.39, 40 An exception to that rule is pneumococcal isolates that are very highly resistant to PCN (MIC 4 mg/dL). At least one study has shown that patients with such isolates may be at higher risk for adverse outcomes and should probably not be treated with penicillins.1, 14, 15, 41 However, nationally, fewer than 6% of pneumococcal isolates have this level of resistance.37

The rationale for empiric broad‐spectrum coverage against both typical and atypical organisms has arisen from many retrospective and observational studies that have suggested that there is clinical benefit and improved outcomes with such regimens. One large retrospective study showed that in elderly patients with CAP, fluoroquinolone monotherapy was associated with lower 30‐day mortality when compared to monotherapy with a third‐generation cephalosporin.34 Adding an extended‐spectrum macrolide (eg, azithromycin) to an extended‐spectrum ‐lactam (eg, ceftriaxone) in the treatment of patients hospitalized with nonsevere CAP also appears to be associated with improved outcomes. Adding a macrolide has resulted in shorter lengths of stay (LOS), less treatment failure, and lower mortality.34, 4244 Similarly, according to unpublished observations, adding doxycycline to a ‐lactam as initial therapy was associated with a benefit of decreased mortality.45 The presumed etiology of the benefit has been the addition of specific coverage of atypical organisms, such as Mycoplasma pneumoniae and Chlamydia pneumoniae, which are common causes of CAP (Table 1). Others have proposed that the benefit of therapy with macrolides may be derived from the inherent anti‐inflammatory properties of macrolides.46 Because research has shown a benefit of dual versus monotherapy across a spectrum of antibiotics, others have proposed the benefit is simply a result of receiving double antibiotic coverage. In particular, 2 studies found a benefit of reduced mortality from combination therapy over monotherapy in bacteremic pneumococcal pneumonia.47, 48

Yet the accumulated evidence for adding coverage of atypical organisms has been only retrospective and observational. Because of this, the recommendation to routinely add antibiotics active against atypical organisms has been questioned by some. Two recent meta‐analyses and a systematic review examined all the available data on the need for atypical coverage in the treatment of patients with community‐acquired pneumonia.4951 Surprisingly, none showed a benefit in clinical efficacy or survival in patients treated with agents active against both atypical and typical organisms when compared to regimens with only typical coverage. In subset analyses, there was a benefit to providing empiric atypical coverage in patients subsequently shown to have Legionella spp. as a causative pathogen. However, this organism was uncommon in all 3 studies. Unfortunately, most studies included in the meta‐analyses compared fluoroquinolone or macrolide monotherapy with third‐generation cephalosporin monotherapy. There have been no high‐quality randomized, controlled trials of the treatment of hospitalized patients with CAP assessing combination therapy covering both typical and atypical organisms with monotherapy targeting typical organisms alone. High‐quality trials are warranted.

Despite the recent articles questioning the importance of atypical coverage, citing the substantial retrospective data and the general inability to identify causative organisms in most cases of CAP, adding a second agent with atypical coverage to a ‐lactam currently appears to be the most efficacious empiric treatment for CAP. Nearly all expert guidelines for the management of community‐acquired pneumonia recommend this empiric approach.1, 14, 27

Table 4 displays our recommendations for the treatment of community‐acquired pneumonia requiring hospitalization. Before implementation of these guidelines, hospitalists should consult with their infectious disease experts and consider local resistance patterns. In general, a typical adult patient with non‐severe CAP without additional risk factors should receive a parenteral extended‐spectrum ‐lactam plus either doxycycline or an advanced macrolide (see Table 4). Extended‐spectrum ‐lactams include cefotaxime, ceftriaxone, ampicillin‐sulbactam, and ertapenem. A respiratory fluoroquinolone as a single agent can be used for non‐ICU patients with CAP, but some agencies, including the Centers for Disease Control (CDC), discourage routine use of these agents in all patients secondary to concerns about cost and increasing gram‐negative rod fluoroquinolone resistance.52, 53

Patients hospitalized with severe CAP who require ICU‐level care are at increased risk of Legionella spp. and drug‐resistant S. pneumoniae, which must be reflected in their initial antibiotic therapy.5 Patients with severe pneumonia should receive an intravenous extended‐spectrum ‐lactam plus either an intravenous macrolide or an intravenous respiratory fluoroquinolone.

All patients with severe CAP who are admitted to the intensive care unit should be routinely screened for risk factors for Pseudomonas aeruginosa. The known risk factors for pseudomonal infection are: bronchiectasis, immunosuppression including more than 10 mg/day of prednisone, malnutrition, and treatment with broad‐spectrum antibiotics in the last month.14 Those at risk for Pseudomonas aeruginosa or other resistant gram‐negative rod infection should be treated with an antipseudomonal ‐lactam plus an antipseudomonal fluoroquinolone. Many patients with severe CAP have risk factors for MRSA infection including recent prolonged hospitalization, recent use of broad‐spectrum antibiotics, and significant underlying lung disease, which should be considered in choosing initial antibiotic therapy.54 In addition, there have been reports of patients without underlying risk factors presenting with severe community‐acquired MRSA pneumonia. Many of these patients were younger and the MRSA pneumonia was associated with a necrotizing or cavitary disease requiring prolonged ICU stays.5558 In such cases or if an institution's rate of methicillin resistance in S. aureus community isolates is high (>15%20%), it may be appropriate to add initial empiric MRSA coverage for patients admitted to the ICU with CAP.55

Some patients will have unique risk factors and clinical presentations, which may require modification of these empiric recommendations. Several studies found 5%15% of cases of community‐acquired pneumonia to be aspiration pneumonia.57 Risk factors for aspiration events include, among others, dysphagia, history of stroke, altered level of consciousness, poor dentition, and tube feeding. Aspiration pneumonia traditionally was believed to be secondary to oral anaerobes, but recent research suggests gram‐positive cocci and gram‐negative rods are the predominant organisms.58 Antibiotic therapy in patients with clear aspiration pneumonia should be directed at these microbes with an extended‐spectrum ‐lactam (eg, ceftriaxone) or a respiratory fluoroquinolone (eg, levofloxacin or moxifloxacin). Anaerobic bacterial coverage can be added in patients with severe periodontal disease, alcoholism, concern for empyema, or evidence of aspiration with pulmonary abscess.58

Patients residing in long‐term care facilities are at high risk of contracting pneumonia. The microbiology of infections acquired in nursing facilities is similar to that in hospital‐acquired cases.59, 60 As a result, patients who develop pneumonia in institutional settings such as nursing homes should be treated with broad‐spectrum antibiotics, including coverage for MRSA.

Subsequent Treatment

Initial empiric antibiotic treatment should be modified based on the results of diagnostic testing. Although the specific etiologic agent is determined in only 25% of cases of CAP,35 when an organism is isolated, antibiotic coverage should be narrowed to cover that particular organism with an antibiotic with adequate lung penetration. Evidence suggests clinicians often do not adjust or narrow antibiotics based on sensitivity results, potentially breeding resistant organisms.61

Patients hospitalized with CAP usually improve quickly if they receive early, appropriate antibiotic therapy and supportive care. Excluding patients with severe CAP requiring intensive care unit admission, most patients resolve their tachycardia, tachypnea, and fever by day 2 or 3.62 Recent practice experience, evidence, and published guidelines14, 27 all indicate that patients can safely be transitioned to oral antibiotic therapy earlier in their hospital course. Table 5 outlines criteria that can be used to identify patients who have had an adequate response to parenteral therapy and can be considered for a switch to oral antibiotics. If these criteria are met, patients have less than a 1% chance of clinical deterioration necessitating admission to an ICU or transitional care unit.62 When an etiologic organism is not identified, oral therapy should reflect a spectrum of coverage to that of the initial intravenous therapy. In some cases, this may require use of more than one oral agent. We have had success, however, transitioning non‐ICU patients initially treated with intravenous ceftriaxone plus oral doxycycline, typically for 4872 hours, to oral doxycycline monotherapy at discharge.45

There have been a limited number of high‐quality randomized trials examining the optimal duration of treatment for community‐acquired pneumonia. Most practice guidelines recommend 710 days for patients with CAP requiring hospitalization, with 14 days for documented Mycoplasma pneumoniae or Chlamydia pneumoniae. One recent randomized trial of patients with mild to severe CAP showed a short course of high‐dose levofloxacin (750 mg daily 5 days) was at least as effective as normal dosing (500 mg daily 10 days).63 Clinical experience with high‐dose levofloxacin is limited, but this regimen can be considered because it may reduce costs and exposure to antibiotics. When diagnosed, Legionella is usually treated for 1021 days, but 14 days is adequate with macrolides because of their long half‐life.27 Patients with more virulent pathogens like Staphylococcus aureus or Pseudomonas aeruginosa or other suppurative complications should be treated for at least 14 days.1, 14, 15, 27 In determining length of therapy, clinicians should use these durations of treatment as guides, and to individualize therapy, they should always consider patient age and frailty, comorbid conditions, severity of illness, and hospital course.

Failure to Respond

Although most patients hospitalized for CAP will improve rapidly and reach clinical stability in 23 days, some patients fail to respond. Some studies have estimated that failure to improve or clinical deterioration occurs in 5%10% of patients in the first 23 days.64 The common reasons for clinical decline or nonresponse to treatment, highlighted in Table 6, are:

• 1

  Incorrect diagnosis: Illnesses such as congestive heart failure, pulmonary embolism, neoplasms, and hypersensitivity pneumonitis can mimick CAP.

• 2

  Inadequate antibiotic selection: The etiologic agent may be resistant to empiric antibiotic selections. Examples would include methicillin‐resistant Staphylococcus aureus (MRSA) or multiresistant gram‐negative bacilli.

• 3

  Unusual pathogen: CAP syndromes can be caused by myriad unusual organisms including Pneumocystis jirovecii, mycobacterium tuberculosis, endemic fungal infections (eg, coccidioidomycosis), and nocardiosis.

• 4

  Complications of pneumonia: Specific complications of CAP include empyema, pulmonary abscess, extrapulmonary spread including meningitis or endocarditis, or other organ dysfunctions such as renal failure or myocardial infarction.

• 5

  Inadequate host response: Despite appropriate antibiotic and supportive therapy, patients with CAP often fail to respond.

Progressive pneumonia despite appropriate therapy and empyema were the most common causes of failure to respond in the first 72 hours in a recent study.64 Risk factors for early failure were older age (>65 years), Pneumonia Severity Index > 90, Legionella pneumonia, gram‐negative pneumonia, and initial antimicrobial therapy discordant with final culture and susceptibility results. The initial evaluation of the nonresponding patient should address these common causes and is likely to include additional imaging (CT), sampling of potential extrapulmonary infection (thoracentesis), and, in some cases, bronchoscopy.

Differential Diagnosis

Given the nonspecific nature of the symptoms and signs associated with CAP, there is no single clinical feature or combination of clinical features that adequately rules in or out the diagnosis of CAP. Consequently, the differential diagnosis to be considered in patients with suspected CAP is broad. Noninfectious diseases can often present with similar clinical syndromes; these include congestive heart failure, exacerbation of chronic obstructive pulmonary disease (COPD), asthma, pulmonary embolism, and hypersensitivity pneumonitis. These diseases can often be distinguished with a thorough history and physical examination.

In addition, other upper‐ and lower‐airway infectious diseases can have similar nonspecific signs and symptoms. In particular, pneumonia must often be differentiated from acute bronchitis, which as a diagnosis accounts for up to 40% of patients evaluated for cough (versus 5% for pneumonia).10 Patients with acute bronchitis frequently do not present with high fevers or hypoxia and in general will not benefit from antibiotic therapy.13 Patients believed to have community‐acquired pneumonia might also be suffering from other pneumonia syndromes including aspiration pneumonia, postobstructive pneumonia, and pneumonia in immunocompromised patients (eg, those with HIV, on steroids, receiving chemotherapy). Determining the correct diagnosis can have implications for therapy and prognosis.

Diagnostic Studies

The diagnosis of community‐acquired pneumonia requires that a patient have both signs and symptoms consistent with pulmonary infection and evidence of a new radiographic infiltrate. Therefore, most guidelines recommend that all patients with a possible diagnosis of CAP be evaluated with chest radiography.1, 14, 15

The specific radiographic findings in community‐acquired pneumonia range from lobar consolidation to hazy focal infiltrate to diffuse bilateral interstitial opacities (see Figure 1). Although chest radiography has traditionally been considered the gold standard for the diagnosis of CAP, its exact performance characteristics are unknown, and it is clearly not 100% sensitive or 100% specific. The utility of the chest radiograph can be limited by patient body habitus, underlying lung disease, or dehydration. Computed tomography (CT) scanning, although not recommended for routine use, can identify pulmonary consolidation in up to 30% of patients with a normal or equivocal chest radiograph in whom pneumonia is suspected and can also identify complications of pneumonia including an empyema or pulmonary abscess.16

Figure 1

Limitations in the performance of the chest radiograph have resulted in an interest in the diagnostic performance of serologic markers of infection such as C‐reactive protein (CRP), procalcitonin, and soluble triggering receptor expressed on myeloid cells (s‐TREM).1719 Preliminary evidence suggests these inflammatory markers may ultimately prove useful in differentiating infectious from noninfectious pulmonary processes, but regular use of these new tests cannot currently be recommended.

Most expert guidelines state that 2 sets of blood cultures should be taken and analyzed prior to antibiotic administration in all patients admitted to the hospital with suspected community‐acquired pneumonia.1, 14, 15 Isolation of bacteria from blood cultures in CAP is a very specific way to identify a causative organism in order to subsequently narrow therapy and also identifies a high‐risk group of patients because bacteremia is associated with increased mortality. Obtaining blood cultures within 24 hours of admission has been associated with 10% lower odds of 30‐day mortality in patients with CAP,20 and as a result, drawing blood cultures prior to antibiotic administration is a national quality indicator for CAP.

There are, however, a number of problems with the routine acquisition of blood cultures in all patients admitted with CAP. Practically, the cultures can be difficult to obtain, can potentially delay the initiation of antibiotics, and are often contaminated, which has been shown to increase both cost and length of stay.21, 22 The yield is generally low: the true‐positive bacteremia rate for admitted patients with CAP ranges from 6% to 9%, and the culture results rarely change management or outcomes.23, 24 Given these limitations, many have argued that blood cultures should be obtained with a more targeted approach. A recent study used a Medicare database to create a decision‐support tool to help maximize the value of blood cultures in CAP.25 The predictors of a positive blood culture are shown in Table 3. Not obtaining cultures on patients who had received prior antibiotics or had no risk factors resulted in about 40% fewer overall cultures while identifying approximately 90% of bacteremias. In their guidelines, the British Thoracic Society (BTS) advocates a similar strategy, recommending blood cultures be omitted in nonsevere pneumonia and in patients without comorbidities.15, 26 Although recommendations vary for non‐severe CAP in hospitalized patients, all professional society guidelines agree that blood cultures should be obtained in critically ill patients, and if cultures are obtained, they should be drawn prior to antibiotics.1, 14, 15, 26

Substantial controversy surrounds the utility of routine sputum gram stains and cultures for patients admitted to the hospital with CAP. The Infectious Disease Society of America (IDSA) and the British Thoracic Society (BTS) both recommend that all patients admitted to the hospital with community‐acquired pneumonia should have a gram stain and culture of expectorated sputum.1, 15, 26 Both organizations argue sputum collection is a simple and inexpensive procedure that can potentially identify pathogenic organisms and can affect both initial and long‐term antibiotic therapy. Most notably, they highlight gram stain specificity of greater than 80% for pneumococcal pneumonia. Conversely, the American Thoracic Society (ATS) argues that sputum gram stains and cultures generally have low sensitivity, specificity, and positive predictive value.14 Furthermore, they argue the utility of sputum testing is also limited practically; in one study 30% of patients could not produce an adequate sputum specimen and up to 30% had received prior antibiotic therapy, substantially reducing the yield.27 In another study, good‐quality sputum with a predominant morphotype could be obtained in only 14% of patients admitted with CAP.28 However, targeting sputum analysis to patients who have not received prior antibiotics and are able to produce an adequate sample improved the yield significantly.29 In addition, with increasing rates of antibiotic resistance among common community isolates (ie, S. Pneumoniae) and the increasing prevalence of infecting organisms not targeted by routine empiric therapy (methicillin‐resistant Staphylococcus Aureus [MRSA]), isolation of potential causative pathogens is increasingly important. We believe that severely ill patients with CAP (such as patients admitted to the ICU), as well as patients with identifiable risk factors for uncommon or drug‐resistant pathogens (eg, Pseudomonas aeruginosa, enteric gram‐negative rods, MRSA, etc.) should have sputum sent for gram stain and culture. Ideally, sputum obtained for gram stain and culture should be:

• 1

  Prior to antibiotic therapy,

• 2

  A deep‐cough, expectorated specimen,

• 3

  A purulent specimen (>25 polymorphonucleacytes and less than 10 squamous cells per high‐powered field), and

• 4

  Rapidly transported to the laboratory.

Subsequent gram stain and culture results should be interpreted in the specific clinical context and antibiotic choices targeted appropriately.

Alternative Diagnostic Tests

In recent years, there has been growth in additional diagnostic tests targeting specific organisms. The pneumococcal urinary antigen assay is a relatively sensitive (50%80%) and highly specific (90%) test for the detection of pneumococcal pneumonia, when compared with conventional diagnostic methods.27 The test is simple, convenient, rapid ( 15 min), and, with its high specificity, may allow for more focused antimicrobial therapy early in management. Current limitations include the possibility of false‐positive tests in patients colonized with S. pneumoniae or infected with other streptococcal species, as well as the inability to determine antibiotic sensitivity from positive tests. Updated IDSA and BTS guidelines state pneumococcal urinary antigen testing is an acceptable adjunct to other diagnostic tests, but blood and sputum analyses should still be performed.26, 27 For patients with suspected Legionella pneumonia (primarily critically ill and immunocompromised patients or in association with regional outbreaks), the urinary Legionella antigen assay is the test of choice, which detects 80%95% of community‐acquired cases of Legionnaires' disease with a specificity of 90%.27

During the winter months (typically from October to March), rapid antigen testing for influenza is generally recommended for patients with signs or symptoms consistent with influenza.27 The sensitivity of these tests is approximately 50%70%, so negative test results do not exclude the diagnosis, but results can be important epidemiologically and therapeutically (differentiating influenza A and B strains).27 Diagnostic tests targeting other common CAP pathogens, such as serologic tests for Mycoplasma pneumoniae or Chlamydia spp, should not be routinely performed. Testing for less common causative pathogens such as Mycobacterium tuberculosis should only be employed in the appropriate clinical setting.

Initial Treatment

Once the admission decision is made and the initial diagnostic tests are completed (including blood and sputum cultures), patients with presumed community‐acquired pneumonia should receive necessary supportive care (O₂, intravenous fluids, etc.) and prompt antimicrobial therapy. Antibiotics should be administered within 4 hours of arrival to patients with suspected CAP, as such prompt administration may be associated with shorter in‐hospital stays and decreased 30‐day mortality.34, 35 Regulatory organizations such as the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) and the Center for Medicare Services (CMS) have made delivery of antibiotics in less than 4 hours a hospital quality measure.

Despite diagnostic testing, the specific etiologic agent causing the pneumonia of a patient remains unknown in up to 75% of those admitted to the hospital.14 Most expert guidelines therefore recommend broad‐spectrum empiric therapy targeting both the typical and the atypical organisms that commonly cause CAP (Table 1).

Recommendations for empiric antibiotics are driven by 2 key factors: antibiotic resistance by S. pneumoniae and the results of studies of CAP treatment outcomes. Historically, patients with suspected community‐acquired pneumonia were treated with penicillin with generally good outcomes. Recently, the rate of S. pneumoniae isolates resistant to penicillin has risen dramatically in the United States, ranging between 20% and 30%, with high‐level resistance (MIC 4 mg/L) as high as 5.7%.36, 37 Concurrently, the rates of resistance of S. pneumoniae to many other antibiotics commonly used to treat CAP have also risen.37 Despite increasing resistance overall, most U.S. pneumococcal isolates have low resistance to third‐generation cephalosporins and fluoroquinolones with enhanced activity against S. pneumoniae.3638 In addition, despite increasing resistance by pneumococcal isolates to penicillin, several observational studies have shown that regardless of initial therapy, resistance to penicillin as well as third‐generation cephalosporins is not associated with higher mortality or worse outcomes when controlled for other risk factors for drug resistance.39, 40 An exception to that rule is pneumococcal isolates that are very highly resistant to PCN (MIC 4 mg/dL). At least one study has shown that patients with such isolates may be at higher risk for adverse outcomes and should probably not be treated with penicillins.1, 14, 15, 41 However, nationally, fewer than 6% of pneumococcal isolates have this level of resistance.37

The rationale for empiric broad‐spectrum coverage against both typical and atypical organisms has arisen from many retrospective and observational studies that have suggested that there is clinical benefit and improved outcomes with such regimens. One large retrospective study showed that in elderly patients with CAP, fluoroquinolone monotherapy was associated with lower 30‐day mortality when compared to monotherapy with a third‐generation cephalosporin.34 Adding an extended‐spectrum macrolide (eg, azithromycin) to an extended‐spectrum ‐lactam (eg, ceftriaxone) in the treatment of patients hospitalized with nonsevere CAP also appears to be associated with improved outcomes. Adding a macrolide has resulted in shorter lengths of stay (LOS), less treatment failure, and lower mortality.34, 4244 Similarly, according to unpublished observations, adding doxycycline to a ‐lactam as initial therapy was associated with a benefit of decreased mortality.45 The presumed etiology of the benefit has been the addition of specific coverage of atypical organisms, such as Mycoplasma pneumoniae and Chlamydia pneumoniae, which are common causes of CAP (Table 1). Others have proposed that the benefit of therapy with macrolides may be derived from the inherent anti‐inflammatory properties of macrolides.46 Because research has shown a benefit of dual versus monotherapy across a spectrum of antibiotics, others have proposed the benefit is simply a result of receiving double antibiotic coverage. In particular, 2 studies found a benefit of reduced mortality from combination therapy over monotherapy in bacteremic pneumococcal pneumonia.47, 48

Yet the accumulated evidence for adding coverage of atypical organisms has been only retrospective and observational. Because of this, the recommendation to routinely add antibiotics active against atypical organisms has been questioned by some. Two recent meta‐analyses and a systematic review examined all the available data on the need for atypical coverage in the treatment of patients with community‐acquired pneumonia.4951 Surprisingly, none showed a benefit in clinical efficacy or survival in patients treated with agents active against both atypical and typical organisms when compared to regimens with only typical coverage. In subset analyses, there was a benefit to providing empiric atypical coverage in patients subsequently shown to have Legionella spp. as a causative pathogen. However, this organism was uncommon in all 3 studies. Unfortunately, most studies included in the meta‐analyses compared fluoroquinolone or macrolide monotherapy with third‐generation cephalosporin monotherapy. There have been no high‐quality randomized, controlled trials of the treatment of hospitalized patients with CAP assessing combination therapy covering both typical and atypical organisms with monotherapy targeting typical organisms alone. High‐quality trials are warranted.

Despite the recent articles questioning the importance of atypical coverage, citing the substantial retrospective data and the general inability to identify causative organisms in most cases of CAP, adding a second agent with atypical coverage to a ‐lactam currently appears to be the most efficacious empiric treatment for CAP. Nearly all expert guidelines for the management of community‐acquired pneumonia recommend this empiric approach.1, 14, 27

Table 4 displays our recommendations for the treatment of community‐acquired pneumonia requiring hospitalization. Before implementation of these guidelines, hospitalists should consult with their infectious disease experts and consider local resistance patterns. In general, a typical adult patient with non‐severe CAP without additional risk factors should receive a parenteral extended‐spectrum ‐lactam plus either doxycycline or an advanced macrolide (see Table 4). Extended‐spectrum ‐lactams include cefotaxime, ceftriaxone, ampicillin‐sulbactam, and ertapenem. A respiratory fluoroquinolone as a single agent can be used for non‐ICU patients with CAP, but some agencies, including the Centers for Disease Control (CDC), discourage routine use of these agents in all patients secondary to concerns about cost and increasing gram‐negative rod fluoroquinolone resistance.52, 53

Patients hospitalized with severe CAP who require ICU‐level care are at increased risk of Legionella spp. and drug‐resistant S. pneumoniae, which must be reflected in their initial antibiotic therapy.5 Patients with severe pneumonia should receive an intravenous extended‐spectrum ‐lactam plus either an intravenous macrolide or an intravenous respiratory fluoroquinolone.

All patients with severe CAP who are admitted to the intensive care unit should be routinely screened for risk factors for Pseudomonas aeruginosa. The known risk factors for pseudomonal infection are: bronchiectasis, immunosuppression including more than 10 mg/day of prednisone, malnutrition, and treatment with broad‐spectrum antibiotics in the last month.14 Those at risk for Pseudomonas aeruginosa or other resistant gram‐negative rod infection should be treated with an antipseudomonal ‐lactam plus an antipseudomonal fluoroquinolone. Many patients with severe CAP have risk factors for MRSA infection including recent prolonged hospitalization, recent use of broad‐spectrum antibiotics, and significant underlying lung disease, which should be considered in choosing initial antibiotic therapy.54 In addition, there have been reports of patients without underlying risk factors presenting with severe community‐acquired MRSA pneumonia. Many of these patients were younger and the MRSA pneumonia was associated with a necrotizing or cavitary disease requiring prolonged ICU stays.5558 In such cases or if an institution's rate of methicillin resistance in S. aureus community isolates is high (>15%20%), it may be appropriate to add initial empiric MRSA coverage for patients admitted to the ICU with CAP.55

Some patients will have unique risk factors and clinical presentations, which may require modification of these empiric recommendations. Several studies found 5%15% of cases of community‐acquired pneumonia to be aspiration pneumonia.57 Risk factors for aspiration events include, among others, dysphagia, history of stroke, altered level of consciousness, poor dentition, and tube feeding. Aspiration pneumonia traditionally was believed to be secondary to oral anaerobes, but recent research suggests gram‐positive cocci and gram‐negative rods are the predominant organisms.58 Antibiotic therapy in patients with clear aspiration pneumonia should be directed at these microbes with an extended‐spectrum ‐lactam (eg, ceftriaxone) or a respiratory fluoroquinolone (eg, levofloxacin or moxifloxacin). Anaerobic bacterial coverage can be added in patients with severe periodontal disease, alcoholism, concern for empyema, or evidence of aspiration with pulmonary abscess.58

Patients residing in long‐term care facilities are at high risk of contracting pneumonia. The microbiology of infections acquired in nursing facilities is similar to that in hospital‐acquired cases.59, 60 As a result, patients who develop pneumonia in institutional settings such as nursing homes should be treated with broad‐spectrum antibiotics, including coverage for MRSA.

Subsequent Treatment

Initial empiric antibiotic treatment should be modified based on the results of diagnostic testing. Although the specific etiologic agent is determined in only 25% of cases of CAP,35 when an organism is isolated, antibiotic coverage should be narrowed to cover that particular organism with an antibiotic with adequate lung penetration. Evidence suggests clinicians often do not adjust or narrow antibiotics based on sensitivity results, potentially breeding resistant organisms.61

Patients hospitalized with CAP usually improve quickly if they receive early, appropriate antibiotic therapy and supportive care. Excluding patients with severe CAP requiring intensive care unit admission, most patients resolve their tachycardia, tachypnea, and fever by day 2 or 3.62 Recent practice experience, evidence, and published guidelines14, 27 all indicate that patients can safely be transitioned to oral antibiotic therapy earlier in their hospital course. Table 5 outlines criteria that can be used to identify patients who have had an adequate response to parenteral therapy and can be considered for a switch to oral antibiotics. If these criteria are met, patients have less than a 1% chance of clinical deterioration necessitating admission to an ICU or transitional care unit.62 When an etiologic organism is not identified, oral therapy should reflect a spectrum of coverage to that of the initial intravenous therapy. In some cases, this may require use of more than one oral agent. We have had success, however, transitioning non‐ICU patients initially treated with intravenous ceftriaxone plus oral doxycycline, typically for 4872 hours, to oral doxycycline monotherapy at discharge.45

There have been a limited number of high‐quality randomized trials examining the optimal duration of treatment for community‐acquired pneumonia. Most practice guidelines recommend 710 days for patients with CAP requiring hospitalization, with 14 days for documented Mycoplasma pneumoniae or Chlamydia pneumoniae. One recent randomized trial of patients with mild to severe CAP showed a short course of high‐dose levofloxacin (750 mg daily 5 days) was at least as effective as normal dosing (500 mg daily 10 days).63 Clinical experience with high‐dose levofloxacin is limited, but this regimen can be considered because it may reduce costs and exposure to antibiotics. When diagnosed, Legionella is usually treated for 1021 days, but 14 days is adequate with macrolides because of their long half‐life.27 Patients with more virulent pathogens like Staphylococcus aureus or Pseudomonas aeruginosa or other suppurative complications should be treated for at least 14 days.1, 14, 15, 27 In determining length of therapy, clinicians should use these durations of treatment as guides, and to individualize therapy, they should always consider patient age and frailty, comorbid conditions, severity of illness, and hospital course.

Failure to Respond

Although most patients hospitalized for CAP will improve rapidly and reach clinical stability in 23 days, some patients fail to respond. Some studies have estimated that failure to improve or clinical deterioration occurs in 5%10% of patients in the first 23 days.64 The common reasons for clinical decline or nonresponse to treatment, highlighted in Table 6, are:

• 1

  Incorrect diagnosis: Illnesses such as congestive heart failure, pulmonary embolism, neoplasms, and hypersensitivity pneumonitis can mimick CAP.

• 2

  Inadequate antibiotic selection: The etiologic agent may be resistant to empiric antibiotic selections. Examples would include methicillin‐resistant Staphylococcus aureus (MRSA) or multiresistant gram‐negative bacilli.

• 3

  Unusual pathogen: CAP syndromes can be caused by myriad unusual organisms including Pneumocystis jirovecii, mycobacterium tuberculosis, endemic fungal infections (eg, coccidioidomycosis), and nocardiosis.

• 4

  Complications of pneumonia: Specific complications of CAP include empyema, pulmonary abscess, extrapulmonary spread including meningitis or endocarditis, or other organ dysfunctions such as renal failure or myocardial infarction.

• 5

  Inadequate host response: Despite appropriate antibiotic and supportive therapy, patients with CAP often fail to respond.

Progressive pneumonia despite appropriate therapy and empyema were the most common causes of failure to respond in the first 72 hours in a recent study.64 Risk factors for early failure were older age (>65 years), Pneumonia Severity Index > 90, Legionella pneumonia, gram‐negative pneumonia, and initial antimicrobial therapy discordant with final culture and susceptibility results. The initial evaluation of the nonresponding patient should address these common causes and is likely to include additional imaging (CT), sampling of potential extrapulmonary infection (thoracentesis), and, in some cases, bronchoscopy.

Differential Diagnosis

Given the nonspecific nature of the symptoms and signs associated with CAP, there is no single clinical feature or combination of clinical features that adequately rules in or out the diagnosis of CAP. Consequently, the differential diagnosis to be considered in patients with suspected CAP is broad. Noninfectious diseases can often present with similar clinical syndromes; these include congestive heart failure, exacerbation of chronic obstructive pulmonary disease (COPD), asthma, pulmonary embolism, and hypersensitivity pneumonitis. These diseases can often be distinguished with a thorough history and physical examination.

In addition, other upper‐ and lower‐airway infectious diseases can have similar nonspecific signs and symptoms. In particular, pneumonia must often be differentiated from acute bronchitis, which as a diagnosis accounts for up to 40% of patients evaluated for cough (versus 5% for pneumonia).10 Patients with acute bronchitis frequently do not present with high fevers or hypoxia and in general will not benefit from antibiotic therapy.13 Patients believed to have community‐acquired pneumonia might also be suffering from other pneumonia syndromes including aspiration pneumonia, postobstructive pneumonia, and pneumonia in immunocompromised patients (eg, those with HIV, on steroids, receiving chemotherapy). Determining the correct diagnosis can have implications for therapy and prognosis.

Diagnostic Studies

The diagnosis of community‐acquired pneumonia requires that a patient have both signs and symptoms consistent with pulmonary infection and evidence of a new radiographic infiltrate. Therefore, most guidelines recommend that all patients with a possible diagnosis of CAP be evaluated with chest radiography.1, 14, 15

The specific radiographic findings in community‐acquired pneumonia range from lobar consolidation to hazy focal infiltrate to diffuse bilateral interstitial opacities (see Figure 1). Although chest radiography has traditionally been considered the gold standard for the diagnosis of CAP, its exact performance characteristics are unknown, and it is clearly not 100% sensitive or 100% specific. The utility of the chest radiograph can be limited by patient body habitus, underlying lung disease, or dehydration. Computed tomography (CT) scanning, although not recommended for routine use, can identify pulmonary consolidation in up to 30% of patients with a normal or equivocal chest radiograph in whom pneumonia is suspected and can also identify complications of pneumonia including an empyema or pulmonary abscess.16

Figure 1

Limitations in the performance of the chest radiograph have resulted in an interest in the diagnostic performance of serologic markers of infection such as C‐reactive protein (CRP), procalcitonin, and soluble triggering receptor expressed on myeloid cells (s‐TREM).1719 Preliminary evidence suggests these inflammatory markers may ultimately prove useful in differentiating infectious from noninfectious pulmonary processes, but regular use of these new tests cannot currently be recommended.

Most expert guidelines state that 2 sets of blood cultures should be taken and analyzed prior to antibiotic administration in all patients admitted to the hospital with suspected community‐acquired pneumonia.1, 14, 15 Isolation of bacteria from blood cultures in CAP is a very specific way to identify a causative organism in order to subsequently narrow therapy and also identifies a high‐risk group of patients because bacteremia is associated with increased mortality. Obtaining blood cultures within 24 hours of admission has been associated with 10% lower odds of 30‐day mortality in patients with CAP,20 and as a result, drawing blood cultures prior to antibiotic administration is a national quality indicator for CAP.

There are, however, a number of problems with the routine acquisition of blood cultures in all patients admitted with CAP. Practically, the cultures can be difficult to obtain, can potentially delay the initiation of antibiotics, and are often contaminated, which has been shown to increase both cost and length of stay.21, 22 The yield is generally low: the true‐positive bacteremia rate for admitted patients with CAP ranges from 6% to 9%, and the culture results rarely change management or outcomes.23, 24 Given these limitations, many have argued that blood cultures should be obtained with a more targeted approach. A recent study used a Medicare database to create a decision‐support tool to help maximize the value of blood cultures in CAP.25 The predictors of a positive blood culture are shown in Table 3. Not obtaining cultures on patients who had received prior antibiotics or had no risk factors resulted in about 40% fewer overall cultures while identifying approximately 90% of bacteremias. In their guidelines, the British Thoracic Society (BTS) advocates a similar strategy, recommending blood cultures be omitted in nonsevere pneumonia and in patients without comorbidities.15, 26 Although recommendations vary for non‐severe CAP in hospitalized patients, all professional society guidelines agree that blood cultures should be obtained in critically ill patients, and if cultures are obtained, they should be drawn prior to antibiotics.1, 14, 15, 26

Substantial controversy surrounds the utility of routine sputum gram stains and cultures for patients admitted to the hospital with CAP. The Infectious Disease Society of America (IDSA) and the British Thoracic Society (BTS) both recommend that all patients admitted to the hospital with community‐acquired pneumonia should have a gram stain and culture of expectorated sputum.1, 15, 26 Both organizations argue sputum collection is a simple and inexpensive procedure that can potentially identify pathogenic organisms and can affect both initial and long‐term antibiotic therapy. Most notably, they highlight gram stain specificity of greater than 80% for pneumococcal pneumonia. Conversely, the American Thoracic Society (ATS) argues that sputum gram stains and cultures generally have low sensitivity, specificity, and positive predictive value.14 Furthermore, they argue the utility of sputum testing is also limited practically; in one study 30% of patients could not produce an adequate sputum specimen and up to 30% had received prior antibiotic therapy, substantially reducing the yield.27 In another study, good‐quality sputum with a predominant morphotype could be obtained in only 14% of patients admitted with CAP.28 However, targeting sputum analysis to patients who have not received prior antibiotics and are able to produce an adequate sample improved the yield significantly.29 In addition, with increasing rates of antibiotic resistance among common community isolates (ie, S. Pneumoniae) and the increasing prevalence of infecting organisms not targeted by routine empiric therapy (methicillin‐resistant Staphylococcus Aureus [MRSA]), isolation of potential causative pathogens is increasingly important. We believe that severely ill patients with CAP (such as patients admitted to the ICU), as well as patients with identifiable risk factors for uncommon or drug‐resistant pathogens (eg, Pseudomonas aeruginosa, enteric gram‐negative rods, MRSA, etc.) should have sputum sent for gram stain and culture. Ideally, sputum obtained for gram stain and culture should be:

• 1

  Prior to antibiotic therapy,

• 2

  A deep‐cough, expectorated specimen,

• 3

  A purulent specimen (>25 polymorphonucleacytes and less than 10 squamous cells per high‐powered field), and

• 4

  Rapidly transported to the laboratory.

Subsequent gram stain and culture results should be interpreted in the specific clinical context and antibiotic choices targeted appropriately.

Alternative Diagnostic Tests

In recent years, there has been growth in additional diagnostic tests targeting specific organisms. The pneumococcal urinary antigen assay is a relatively sensitive (50%80%) and highly specific (90%) test for the detection of pneumococcal pneumonia, when compared with conventional diagnostic methods.27 The test is simple, convenient, rapid ( 15 min), and, with its high specificity, may allow for more focused antimicrobial therapy early in management. Current limitations include the possibility of false‐positive tests in patients colonized with S. pneumoniae or infected with other streptococcal species, as well as the inability to determine antibiotic sensitivity from positive tests. Updated IDSA and BTS guidelines state pneumococcal urinary antigen testing is an acceptable adjunct to other diagnostic tests, but blood and sputum analyses should still be performed.26, 27 For patients with suspected Legionella pneumonia (primarily critically ill and immunocompromised patients or in association with regional outbreaks), the urinary Legionella antigen assay is the test of choice, which detects 80%95% of community‐acquired cases of Legionnaires' disease with a specificity of 90%.27

During the winter months (typically from October to March), rapid antigen testing for influenza is generally recommended for patients with signs or symptoms consistent with influenza.27 The sensitivity of these tests is approximately 50%70%, so negative test results do not exclude the diagnosis, but results can be important epidemiologically and therapeutically (differentiating influenza A and B strains).27 Diagnostic tests targeting other common CAP pathogens, such as serologic tests for Mycoplasma pneumoniae or Chlamydia spp, should not be routinely performed. Testing for less common causative pathogens such as Mycobacterium tuberculosis should only be employed in the appropriate clinical setting.

Initial Treatment

Once the admission decision is made and the initial diagnostic tests are completed (including blood and sputum cultures), patients with presumed community‐acquired pneumonia should receive necessary supportive care (O₂, intravenous fluids, etc.) and prompt antimicrobial therapy. Antibiotics should be administered within 4 hours of arrival to patients with suspected CAP, as such prompt administration may be associated with shorter in‐hospital stays and decreased 30‐day mortality.34, 35 Regulatory organizations such as the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) and the Center for Medicare Services (CMS) have made delivery of antibiotics in less than 4 hours a hospital quality measure.

Despite diagnostic testing, the specific etiologic agent causing the pneumonia of a patient remains unknown in up to 75% of those admitted to the hospital.14 Most expert guidelines therefore recommend broad‐spectrum empiric therapy targeting both the typical and the atypical organisms that commonly cause CAP (Table 1).

Recommendations for empiric antibiotics are driven by 2 key factors: antibiotic resistance by S. pneumoniae and the results of studies of CAP treatment outcomes. Historically, patients with suspected community‐acquired pneumonia were treated with penicillin with generally good outcomes. Recently, the rate of S. pneumoniae isolates resistant to penicillin has risen dramatically in the United States, ranging between 20% and 30%, with high‐level resistance (MIC 4 mg/L) as high as 5.7%.36, 37 Concurrently, the rates of resistance of S. pneumoniae to many other antibiotics commonly used to treat CAP have also risen.37 Despite increasing resistance overall, most U.S. pneumococcal isolates have low resistance to third‐generation cephalosporins and fluoroquinolones with enhanced activity against S. pneumoniae.3638 In addition, despite increasing resistance by pneumococcal isolates to penicillin, several observational studies have shown that regardless of initial therapy, resistance to penicillin as well as third‐generation cephalosporins is not associated with higher mortality or worse outcomes when controlled for other risk factors for drug resistance.39, 40 An exception to that rule is pneumococcal isolates that are very highly resistant to PCN (MIC 4 mg/dL). At least one study has shown that patients with such isolates may be at higher risk for adverse outcomes and should probably not be treated with penicillins.1, 14, 15, 41 However, nationally, fewer than 6% of pneumococcal isolates have this level of resistance.37

The rationale for empiric broad‐spectrum coverage against both typical and atypical organisms has arisen from many retrospective and observational studies that have suggested that there is clinical benefit and improved outcomes with such regimens. One large retrospective study showed that in elderly patients with CAP, fluoroquinolone monotherapy was associated with lower 30‐day mortality when compared to monotherapy with a third‐generation cephalosporin.34 Adding an extended‐spectrum macrolide (eg, azithromycin) to an extended‐spectrum ‐lactam (eg, ceftriaxone) in the treatment of patients hospitalized with nonsevere CAP also appears to be associated with improved outcomes. Adding a macrolide has resulted in shorter lengths of stay (LOS), less treatment failure, and lower mortality.34, 4244 Similarly, according to unpublished observations, adding doxycycline to a ‐lactam as initial therapy was associated with a benefit of decreased mortality.45 The presumed etiology of the benefit has been the addition of specific coverage of atypical organisms, such as Mycoplasma pneumoniae and Chlamydia pneumoniae, which are common causes of CAP (Table 1). Others have proposed that the benefit of therapy with macrolides may be derived from the inherent anti‐inflammatory properties of macrolides.46 Because research has shown a benefit of dual versus monotherapy across a spectrum of antibiotics, others have proposed the benefit is simply a result of receiving double antibiotic coverage. In particular, 2 studies found a benefit of reduced mortality from combination therapy over monotherapy in bacteremic pneumococcal pneumonia.47, 48

Yet the accumulated evidence for adding coverage of atypical organisms has been only retrospective and observational. Because of this, the recommendation to routinely add antibiotics active against atypical organisms has been questioned by some. Two recent meta‐analyses and a systematic review examined all the available data on the need for atypical coverage in the treatment of patients with community‐acquired pneumonia.4951 Surprisingly, none showed a benefit in clinical efficacy or survival in patients treated with agents active against both atypical and typical organisms when compared to regimens with only typical coverage. In subset analyses, there was a benefit to providing empiric atypical coverage in patients subsequently shown to have Legionella spp. as a causative pathogen. However, this organism was uncommon in all 3 studies. Unfortunately, most studies included in the meta‐analyses compared fluoroquinolone or macrolide monotherapy with third‐generation cephalosporin monotherapy. There have been no high‐quality randomized, controlled trials of the treatment of hospitalized patients with CAP assessing combination therapy covering both typical and atypical organisms with monotherapy targeting typical organisms alone. High‐quality trials are warranted.

Despite the recent articles questioning the importance of atypical coverage, citing the substantial retrospective data and the general inability to identify causative organisms in most cases of CAP, adding a second agent with atypical coverage to a ‐lactam currently appears to be the most efficacious empiric treatment for CAP. Nearly all expert guidelines for the management of community‐acquired pneumonia recommend this empiric approach.1, 14, 27

Table 4 displays our recommendations for the treatment of community‐acquired pneumonia requiring hospitalization. Before implementation of these guidelines, hospitalists should consult with their infectious disease experts and consider local resistance patterns. In general, a typical adult patient with non‐severe CAP without additional risk factors should receive a parenteral extended‐spectrum ‐lactam plus either doxycycline or an advanced macrolide (see Table 4). Extended‐spectrum ‐lactams include cefotaxime, ceftriaxone, ampicillin‐sulbactam, and ertapenem. A respiratory fluoroquinolone as a single agent can be used for non‐ICU patients with CAP, but some agencies, including the Centers for Disease Control (CDC), discourage routine use of these agents in all patients secondary to concerns about cost and increasing gram‐negative rod fluoroquinolone resistance.52, 53

Patients hospitalized with severe CAP who require ICU‐level care are at increased risk of Legionella spp. and drug‐resistant S. pneumoniae, which must be reflected in their initial antibiotic therapy.5 Patients with severe pneumonia should receive an intravenous extended‐spectrum ‐lactam plus either an intravenous macrolide or an intravenous respiratory fluoroquinolone.

All patients with severe CAP who are admitted to the intensive care unit should be routinely screened for risk factors for Pseudomonas aeruginosa. The known risk factors for pseudomonal infection are: bronchiectasis, immunosuppression including more than 10 mg/day of prednisone, malnutrition, and treatment with broad‐spectrum antibiotics in the last month.14 Those at risk for Pseudomonas aeruginosa or other resistant gram‐negative rod infection should be treated with an antipseudomonal ‐lactam plus an antipseudomonal fluoroquinolone. Many patients with severe CAP have risk factors for MRSA infection including recent prolonged hospitalization, recent use of broad‐spectrum antibiotics, and significant underlying lung disease, which should be considered in choosing initial antibiotic therapy.54 In addition, there have been reports of patients without underlying risk factors presenting with severe community‐acquired MRSA pneumonia. Many of these patients were younger and the MRSA pneumonia was associated with a necrotizing or cavitary disease requiring prolonged ICU stays.5558 In such cases or if an institution's rate of methicillin resistance in S. aureus community isolates is high (>15%20%), it may be appropriate to add initial empiric MRSA coverage for patients admitted to the ICU with CAP.55

Some patients will have unique risk factors and clinical presentations, which may require modification of these empiric recommendations. Several studies found 5%15% of cases of community‐acquired pneumonia to be aspiration pneumonia.57 Risk factors for aspiration events include, among others, dysphagia, history of stroke, altered level of consciousness, poor dentition, and tube feeding. Aspiration pneumonia traditionally was believed to be secondary to oral anaerobes, but recent research suggests gram‐positive cocci and gram‐negative rods are the predominant organisms.58 Antibiotic therapy in patients with clear aspiration pneumonia should be directed at these microbes with an extended‐spectrum ‐lactam (eg, ceftriaxone) or a respiratory fluoroquinolone (eg, levofloxacin or moxifloxacin). Anaerobic bacterial coverage can be added in patients with severe periodontal disease, alcoholism, concern for empyema, or evidence of aspiration with pulmonary abscess.58

Patients residing in long‐term care facilities are at high risk of contracting pneumonia. The microbiology of infections acquired in nursing facilities is similar to that in hospital‐acquired cases.59, 60 As a result, patients who develop pneumonia in institutional settings such as nursing homes should be treated with broad‐spectrum antibiotics, including coverage for MRSA.

Subsequent Treatment

Initial empiric antibiotic treatment should be modified based on the results of diagnostic testing. Although the specific etiologic agent is determined in only 25% of cases of CAP,35 when an organism is isolated, antibiotic coverage should be narrowed to cover that particular organism with an antibiotic with adequate lung penetration. Evidence suggests clinicians often do not adjust or narrow antibiotics based on sensitivity results, potentially breeding resistant organisms.61

Patients hospitalized with CAP usually improve quickly if they receive early, appropriate antibiotic therapy and supportive care. Excluding patients with severe CAP requiring intensive care unit admission, most patients resolve their tachycardia, tachypnea, and fever by day 2 or 3.62 Recent practice experience, evidence, and published guidelines14, 27 all indicate that patients can safely be transitioned to oral antibiotic therapy earlier in their hospital course. Table 5 outlines criteria that can be used to identify patients who have had an adequate response to parenteral therapy and can be considered for a switch to oral antibiotics. If these criteria are met, patients have less than a 1% chance of clinical deterioration necessitating admission to an ICU or transitional care unit.62 When an etiologic organism is not identified, oral therapy should reflect a spectrum of coverage to that of the initial intravenous therapy. In some cases, this may require use of more than one oral agent. We have had success, however, transitioning non‐ICU patients initially treated with intravenous ceftriaxone plus oral doxycycline, typically for 4872 hours, to oral doxycycline monotherapy at discharge.45

There have been a limited number of high‐quality randomized trials examining the optimal duration of treatment for community‐acquired pneumonia. Most practice guidelines recommend 710 days for patients with CAP requiring hospitalization, with 14 days for documented Mycoplasma pneumoniae or Chlamydia pneumoniae. One recent randomized trial of patients with mild to severe CAP showed a short course of high‐dose levofloxacin (750 mg daily 5 days) was at least as effective as normal dosing (500 mg daily 10 days).63 Clinical experience with high‐dose levofloxacin is limited, but this regimen can be considered because it may reduce costs and exposure to antibiotics. When diagnosed, Legionella is usually treated for 1021 days, but 14 days is adequate with macrolides because of their long half‐life.27 Patients with more virulent pathogens like Staphylococcus aureus or Pseudomonas aeruginosa or other suppurative complications should be treated for at least 14 days.1, 14, 15, 27 In determining length of therapy, clinicians should use these durations of treatment as guides, and to individualize therapy, they should always consider patient age and frailty, comorbid conditions, severity of illness, and hospital course.

Failure to Respond

Although most patients hospitalized for CAP will improve rapidly and reach clinical stability in 23 days, some patients fail to respond. Some studies have estimated that failure to improve or clinical deterioration occurs in 5%10% of patients in the first 23 days.64 The common reasons for clinical decline or nonresponse to treatment, highlighted in Table 6, are:

• 1

  Incorrect diagnosis: Illnesses such as congestive heart failure, pulmonary embolism, neoplasms, and hypersensitivity pneumonitis can mimick CAP.

• 2

  Inadequate antibiotic selection: The etiologic agent may be resistant to empiric antibiotic selections. Examples would include methicillin‐resistant Staphylococcus aureus (MRSA) or multiresistant gram‐negative bacilli.

• 3

  Unusual pathogen: CAP syndromes can be caused by myriad unusual organisms including Pneumocystis jirovecii, mycobacterium tuberculosis, endemic fungal infections (eg, coccidioidomycosis), and nocardiosis.

• 4

  Complications of pneumonia: Specific complications of CAP include empyema, pulmonary abscess, extrapulmonary spread including meningitis or endocarditis, or other organ dysfunctions such as renal failure or myocardial infarction.

• 5

  Inadequate host response: Despite appropriate antibiotic and supportive therapy, patients with CAP often fail to respond.

Progressive pneumonia despite appropriate therapy and empyema were the most common causes of failure to respond in the first 72 hours in a recent study.64 Risk factors for early failure were older age (>65 years), Pneumonia Severity Index > 90, Legionella pneumonia, gram‐negative pneumonia, and initial antimicrobial therapy discordant with final culture and susceptibility results. The initial evaluation of the nonresponding patient should address these common causes and is likely to include additional imaging (CT), sampling of potential extrapulmonary infection (thoracentesis), and, in some cases, bronchoscopy.