Sarah L. Krein, PhD, RN

Peripherally inserted central catheters (PICCs) are among the most prevalent of venous access devices in hospitalized patients.[1, 2] Although growing use of these devices reflects clinical advantages, such as a reduced risk of complications during insertion and durable venous access, use of PICCs is also likely related to the growth of vascular access nursing.[3, 4] A relatively new specialty, vascular access nurses obtain, maintain, and manage venous access in hospitalized patients.[4, 5] Depending on their scope of practice, these professionals are responsible not only for insertion of devices, such as peripheral intravenous catheters and PICCs, but also nontunneled central venous catheters and arterial catheters in some settings.[6]

Although a growing number of US hospitals have introduced vascular nursing teams,[7] little is known about the experience, practice, knowledge, and beliefs of vascular access nurses. This knowledge gap is relevant for hospitalists and hospital medicine as (1) vascular access nurses increasingly represent a key partner in the care of hospitalized patients; (2) the knowledge and practice of these individuals directly affects patient safety and clinical outcomes; and (3) understanding experience, practice, and beliefs of these specialists can help inform decision making and quality‐improvement efforts related to PICCs. As hospitalists increasingly order the placement of and care for patients with PICCs, they are also well suited to improve PICC practice.

Therefore, we conducted a survey of vascular access nurses employed by hospitals that participate in the Michigan Hospital Medicine Safety (HMS) Consortium, a Blue Cross Blue Shield of Michiganfunded collaborative quality initiative.[6] We aimed to understand experience, practice, knowledge, and beliefs related to PICC care and use.

METHODS

RESULTS

Of 172 vascular nurses who received invitations, 140 completed the survey for a response rate of 81%. Respondents reported working in not‐for‐profit hospitals (36%), academic medical centers (29%), and for‐profit hospitals (21%). Although multiple providers (eg, interventional radiology staff and providers, physicians) placed PICCs, 95% of those surveyed reported that they placed the majority of the PICCs at their institutions. Although most respondents placed PICCs in adult patients (86%), a few also placed PICCs in pediatric populations (17%). Vascular nursing programs were largely housed in their own department, but some reported to general nursing or subspecialties such as interventional radiology, cardiology, and critical care. Most respondents indicated their facilities had written policies regarding standard insertion and care practices (87% and 95%, respectively), but only 30% had policies regarding the necessity or appropriateness of PICCs.

Experience among respondents was variable: approximately a third had placed PICCs for <5 years (28.6%), whereas 58% reported placing PICCs for 5 years Correspondingly, 26% reported having placed 100 to 500 PICCs, whereas 34% had placed 1000 or more PICCs. Only 23% of those surveyed held a dedicated vascular access certification, such as board certified in vascular access or certified registered nurse infusion, whereas 16% indicated that they served as the vascular access lead nurse for their facility. Following placement, 94% of respondents reported that their facilities tracked the number of PICCs inserted, but only 40% indicated that dwell times of devices were also recorded. Only 30% of nurses reported that their hospitals had a written policy to evaluate PICC necessity or appropriateness following placement (Table 1).

The most commonly reported indications for PICC placement included intravenous antibiotics at discharge, difficult venous access, and placement for chemotherapy in patients with cancer. Forty‐six percent of nurses indicated they had placed a PICC in a patient receiving some form of dialysis in the past several months; however, 91% of these respondents reported receiving approval from nephrology prior to placement in these patients. Although almost all nurses (91%) used ultrasound to find a suitable vein for PICC placement, a smaller percentage used ultrasound to estimate the catheter‐to‐vein ratio to prevent thrombosis (79%), and only a few (14%) documented this figure in the medical record. Three‐quarters of those surveyed (76%) indicated they used ECG‐based systems to position PICC tips at the cavoatrial junction to prevent thrombosis. Of those who used this technology, 36% still obtained chest x‐rays to verify the position of the PICC tip. According to 84% of respondents, flushing of PICCs was performed mainly by bedside nurses, whereas scheduled weekly dressing changes were most often performed by vascular access nurses (Table 2).

With respect to complications, catheter occlusion, migration, and DVT were reported as the 3 most prevalent adverse events. Interestingly, respondents did not report central lineassociated bloodstream infection (CLABSI) as a common complication. Additionally, 51% of those surveyed indicated that physicians unnecessarily removed PICCs when CLABSI was suspected but not confirmed. When managing catheter occlusion, 50% of respondents began with normal saline flushes but used tissue‐plasminogen activator if saline failed to resolve occlusion. Management of catheter migration varied based on degree of device movement: when the PICC had migrated <5 cm, most respondents (77%) indicated they would first obtain a chest x‐ray to determine the position of the PICC tip, with few (4%) performing catheter exchange. However, if the PICC had migrated more than 5 cm, a significantly greater proportion of respondents (21%) indicated they would perform a catheter exchange. With regard to managing DVT, most vascular nurses reported they notified nurses and physicians to continue using the PICC but recommended tests to confirm the diagnosis.

To better understand the experiences of vascular nurses, we asked for their perceptions regarding appropriateness of PICC use and relationships with bedside nurses, physicians, and leadership. Over a third of respondents (36%) felt that <5% of all PICCs may be inappropriate in their facility, whereas 1 in 5 indicated that 10% to 24% of PICCs placed in their facilities may be inappropriate or could have been avoided. Almost all (98%) of the nurses stated they were not empowered to remove idle or clinically unnecessary PICCs without physician authorization. Although 51% of nurses described the support received from hospital leadership as excellent, very good, or good, 43% described leadership support as either fair or poor. Conversely, relationships with bedside nurses and physicians were rated as being very good or good by nearly two‐thirds of those surveyed (64% and 65%, respectively) (Table 3).

DISCUSSION

In this survey of 140 vascular access nurses in hospitals across Michigan, new insights regarding the experience, practice, knowledge, and beliefs of this group of providers were obtained. We found that vascular access nurses varied with respect to years in practice, volume of PICCs placed, and certification status, reflecting heterogeneity in this provider group. Variation in insertion techniques, such as use of ultrasound to examine catheter‐to‐vein ratio (a key way to prevent thrombosis) or newer ECG technology to position the PICC, was also noted. Although indications for PICC insertion appeared consistent with published literature, the frequency with which these devices were placed in patients receiving dialysis (reportedly with nephrology approval) was surprising given national calls to avoid such use.[16] Opportunities to improve hospital practices, such as tracking PICC dwell times and PICC necessity, as well as the potential need to better educate physicians on when to remove PICCs for suspected CLABSI, were also identified. Collectively, these data are highly relevant to hospitalists and health systems as they help to identify areas for quality improvement and inform clinical practice regarding the use of PICCs in hospitalized patients. As hospitalists increasingly order PICCs and manage complications associated with these devices, they are well suited to use these data so as to improve patient safety and clinical outcomes.

Venous access is the most common medical procedure performed in hospitalized medical patients. Although a number of devices including peripheral intravenous catheters, central venous catheters, and PICCs are used for this purpose, the growing use of PICCs to secure venous access has been documented in several studies.[17] Such growth, in part, undoubtedly reflects increasing availability of vascular access nurses. Traditionally placed by interventional radiologists, the creation of dedicated vascular nursing teams has resulted in these subspecialists now serving in more of a backup or trouble‐shooting role rather than that of primary operator.[4, 14] This paradigm shift is well illustrated in a recent survey of infection preventionists, where over 60% of respondents reported that they had a vascular nursing team in their facility.[7] The growth of these nursing‐led vascular access teams has produced not only high rates of insertion success and low rates of complications, but also greater cost‐effectiveness when compared to interventional radiologybased insertion.[18]

Nonetheless, our survey also identified a number of important concerns regarding PICC practices and vascular nursing providers. First, we found variation in areas such as insertion practices and management of complications. Such variability highlights the importance of both growing and disseminating the evidence base for consistent practice in vascular nursing. Through their close clinical affiliation with vascular nurses and shared interests in obtaining safe and appropriate venous access for patients, hospitalists are ideally poised to lead this effort. Second, similarities between vascular nurse opinions regarding appropriateness of PICCs and those of hospitalists from a prior survey were noted.[19] Namely, a substantial proportion of both vascular nurses and hospitalists felt that some PICCs were inappropriate and could be avoided. Third, although relationships between vascular access nurses and leadership were reported as being variable, the survey responses suggested relatively good interprovider relationships with bedside nurses and physicians. Such relationships likely reflect the close clinical ties that emerge from being in the trenches of patient care and suggest that interventions to improve care in partnership with these providers are highly viable.

Our study has some limitations. First, despite a high response rate, our study used a survey design and reports findings from a convenience sample of vascular access nurses in a single state. Thus, nonrespondent and selection biases remain threats to our conclusions. Additionally, some respondents did not complete all responses, perhaps due to nonapplicability to practice or other unknown reasons. The pattern of missingness observed, however, suggested that such responses were missing at random. Second, we surveyed vascular nurses in hospitals that are actively engaged in improving PICC practices; our findings may therefore not be representative of vascular nursing professionals as a whole and may instead reflect those of a highly motivated group of individuals. Relatedly, the underlying reasons for adoption of specific practices or techniques cannot be discerned from our study. Third, although we did not find differences based on years in practice or certification status, our sample size was relatively small and likely underpowered for these comparisons. Finally, our study sample consists of vascular nurses who are clustered within hospitals in which they are employed. Therefore, overlap between reported practices and those required by the facility are possible.

Despite these limitations, our study has important strengths. First, this is among the most comprehensive of surveys examining vascular nursing experience, practice, knowledge, and beliefs. The growing presence of these providers across US hospitals, coupled with limited insight regarding their clinical practices, highlight the importance and utility of these data. Second, we noted important differences in experience, practices, and interprovider relationships between vascular providers in this field. Although we are unable to ascertain the drivers or significance of such variation, hospitals and health systems focused on improving patient safety should consider quantifying and exploring these factors. Third, findings from our survey within Michigan suggest the need for similar, larger studies across the country. Partnerships with nursing organizations or larger professional groups that represent vascular nursing specialists may be helpful in this regard.

In conclusion, we found important similarities and differences in vascular nursing experience, practice, knowledge, and beliefs in Michigan. These data are useful as they help provide context regarding the constitution of these teams, current practices, and opportunities for improving care. Hospitalists seeking to improve patient safety may use these data to better inform vascular access practice in hospitalized patients.

Peripherally inserted central catheters (PICCs) are among the most prevalent of venous access devices in hospitalized patients.[1, 2] Although growing use of these devices reflects clinical advantages, such as a reduced risk of complications during insertion and durable venous access, use of PICCs is also likely related to the growth of vascular access nursing.[3, 4] A relatively new specialty, vascular access nurses obtain, maintain, and manage venous access in hospitalized patients.[4, 5] Depending on their scope of practice, these professionals are responsible not only for insertion of devices, such as peripheral intravenous catheters and PICCs, but also nontunneled central venous catheters and arterial catheters in some settings.[6]

Although a growing number of US hospitals have introduced vascular nursing teams,[7] little is known about the experience, practice, knowledge, and beliefs of vascular access nurses. This knowledge gap is relevant for hospitalists and hospital medicine as (1) vascular access nurses increasingly represent a key partner in the care of hospitalized patients; (2) the knowledge and practice of these individuals directly affects patient safety and clinical outcomes; and (3) understanding experience, practice, and beliefs of these specialists can help inform decision making and quality‐improvement efforts related to PICCs. As hospitalists increasingly order the placement of and care for patients with PICCs, they are also well suited to improve PICC practice.

Therefore, we conducted a survey of vascular access nurses employed by hospitals that participate in the Michigan Hospital Medicine Safety (HMS) Consortium, a Blue Cross Blue Shield of Michiganfunded collaborative quality initiative.[6] We aimed to understand experience, practice, knowledge, and beliefs related to PICC care and use.

METHODS

RESULTS

Of 172 vascular nurses who received invitations, 140 completed the survey for a response rate of 81%. Respondents reported working in not‐for‐profit hospitals (36%), academic medical centers (29%), and for‐profit hospitals (21%). Although multiple providers (eg, interventional radiology staff and providers, physicians) placed PICCs, 95% of those surveyed reported that they placed the majority of the PICCs at their institutions. Although most respondents placed PICCs in adult patients (86%), a few also placed PICCs in pediatric populations (17%). Vascular nursing programs were largely housed in their own department, but some reported to general nursing or subspecialties such as interventional radiology, cardiology, and critical care. Most respondents indicated their facilities had written policies regarding standard insertion and care practices (87% and 95%, respectively), but only 30% had policies regarding the necessity or appropriateness of PICCs.

Experience among respondents was variable: approximately a third had placed PICCs for <5 years (28.6%), whereas 58% reported placing PICCs for 5 years Correspondingly, 26% reported having placed 100 to 500 PICCs, whereas 34% had placed 1000 or more PICCs. Only 23% of those surveyed held a dedicated vascular access certification, such as board certified in vascular access or certified registered nurse infusion, whereas 16% indicated that they served as the vascular access lead nurse for their facility. Following placement, 94% of respondents reported that their facilities tracked the number of PICCs inserted, but only 40% indicated that dwell times of devices were also recorded. Only 30% of nurses reported that their hospitals had a written policy to evaluate PICC necessity or appropriateness following placement (Table 1).

The most commonly reported indications for PICC placement included intravenous antibiotics at discharge, difficult venous access, and placement for chemotherapy in patients with cancer. Forty‐six percent of nurses indicated they had placed a PICC in a patient receiving some form of dialysis in the past several months; however, 91% of these respondents reported receiving approval from nephrology prior to placement in these patients. Although almost all nurses (91%) used ultrasound to find a suitable vein for PICC placement, a smaller percentage used ultrasound to estimate the catheter‐to‐vein ratio to prevent thrombosis (79%), and only a few (14%) documented this figure in the medical record. Three‐quarters of those surveyed (76%) indicated they used ECG‐based systems to position PICC tips at the cavoatrial junction to prevent thrombosis. Of those who used this technology, 36% still obtained chest x‐rays to verify the position of the PICC tip. According to 84% of respondents, flushing of PICCs was performed mainly by bedside nurses, whereas scheduled weekly dressing changes were most often performed by vascular access nurses (Table 2).

With respect to complications, catheter occlusion, migration, and DVT were reported as the 3 most prevalent adverse events. Interestingly, respondents did not report central lineassociated bloodstream infection (CLABSI) as a common complication. Additionally, 51% of those surveyed indicated that physicians unnecessarily removed PICCs when CLABSI was suspected but not confirmed. When managing catheter occlusion, 50% of respondents began with normal saline flushes but used tissue‐plasminogen activator if saline failed to resolve occlusion. Management of catheter migration varied based on degree of device movement: when the PICC had migrated <5 cm, most respondents (77%) indicated they would first obtain a chest x‐ray to determine the position of the PICC tip, with few (4%) performing catheter exchange. However, if the PICC had migrated more than 5 cm, a significantly greater proportion of respondents (21%) indicated they would perform a catheter exchange. With regard to managing DVT, most vascular nurses reported they notified nurses and physicians to continue using the PICC but recommended tests to confirm the diagnosis.

To better understand the experiences of vascular nurses, we asked for their perceptions regarding appropriateness of PICC use and relationships with bedside nurses, physicians, and leadership. Over a third of respondents (36%) felt that <5% of all PICCs may be inappropriate in their facility, whereas 1 in 5 indicated that 10% to 24% of PICCs placed in their facilities may be inappropriate or could have been avoided. Almost all (98%) of the nurses stated they were not empowered to remove idle or clinically unnecessary PICCs without physician authorization. Although 51% of nurses described the support received from hospital leadership as excellent, very good, or good, 43% described leadership support as either fair or poor. Conversely, relationships with bedside nurses and physicians were rated as being very good or good by nearly two‐thirds of those surveyed (64% and 65%, respectively) (Table 3).

DISCUSSION

In this survey of 140 vascular access nurses in hospitals across Michigan, new insights regarding the experience, practice, knowledge, and beliefs of this group of providers were obtained. We found that vascular access nurses varied with respect to years in practice, volume of PICCs placed, and certification status, reflecting heterogeneity in this provider group. Variation in insertion techniques, such as use of ultrasound to examine catheter‐to‐vein ratio (a key way to prevent thrombosis) or newer ECG technology to position the PICC, was also noted. Although indications for PICC insertion appeared consistent with published literature, the frequency with which these devices were placed in patients receiving dialysis (reportedly with nephrology approval) was surprising given national calls to avoid such use.[16] Opportunities to improve hospital practices, such as tracking PICC dwell times and PICC necessity, as well as the potential need to better educate physicians on when to remove PICCs for suspected CLABSI, were also identified. Collectively, these data are highly relevant to hospitalists and health systems as they help to identify areas for quality improvement and inform clinical practice regarding the use of PICCs in hospitalized patients. As hospitalists increasingly order PICCs and manage complications associated with these devices, they are well suited to use these data so as to improve patient safety and clinical outcomes.

Venous access is the most common medical procedure performed in hospitalized medical patients. Although a number of devices including peripheral intravenous catheters, central venous catheters, and PICCs are used for this purpose, the growing use of PICCs to secure venous access has been documented in several studies.[17] Such growth, in part, undoubtedly reflects increasing availability of vascular access nurses. Traditionally placed by interventional radiologists, the creation of dedicated vascular nursing teams has resulted in these subspecialists now serving in more of a backup or trouble‐shooting role rather than that of primary operator.[4, 14] This paradigm shift is well illustrated in a recent survey of infection preventionists, where over 60% of respondents reported that they had a vascular nursing team in their facility.[7] The growth of these nursing‐led vascular access teams has produced not only high rates of insertion success and low rates of complications, but also greater cost‐effectiveness when compared to interventional radiologybased insertion.[18]

Nonetheless, our survey also identified a number of important concerns regarding PICC practices and vascular nursing providers. First, we found variation in areas such as insertion practices and management of complications. Such variability highlights the importance of both growing and disseminating the evidence base for consistent practice in vascular nursing. Through their close clinical affiliation with vascular nurses and shared interests in obtaining safe and appropriate venous access for patients, hospitalists are ideally poised to lead this effort. Second, similarities between vascular nurse opinions regarding appropriateness of PICCs and those of hospitalists from a prior survey were noted.[19] Namely, a substantial proportion of both vascular nurses and hospitalists felt that some PICCs were inappropriate and could be avoided. Third, although relationships between vascular access nurses and leadership were reported as being variable, the survey responses suggested relatively good interprovider relationships with bedside nurses and physicians. Such relationships likely reflect the close clinical ties that emerge from being in the trenches of patient care and suggest that interventions to improve care in partnership with these providers are highly viable.

Our study has some limitations. First, despite a high response rate, our study used a survey design and reports findings from a convenience sample of vascular access nurses in a single state. Thus, nonrespondent and selection biases remain threats to our conclusions. Additionally, some respondents did not complete all responses, perhaps due to nonapplicability to practice or other unknown reasons. The pattern of missingness observed, however, suggested that such responses were missing at random. Second, we surveyed vascular nurses in hospitals that are actively engaged in improving PICC practices; our findings may therefore not be representative of vascular nursing professionals as a whole and may instead reflect those of a highly motivated group of individuals. Relatedly, the underlying reasons for adoption of specific practices or techniques cannot be discerned from our study. Third, although we did not find differences based on years in practice or certification status, our sample size was relatively small and likely underpowered for these comparisons. Finally, our study sample consists of vascular nurses who are clustered within hospitals in which they are employed. Therefore, overlap between reported practices and those required by the facility are possible.

Despite these limitations, our study has important strengths. First, this is among the most comprehensive of surveys examining vascular nursing experience, practice, knowledge, and beliefs. The growing presence of these providers across US hospitals, coupled with limited insight regarding their clinical practices, highlight the importance and utility of these data. Second, we noted important differences in experience, practices, and interprovider relationships between vascular providers in this field. Although we are unable to ascertain the drivers or significance of such variation, hospitals and health systems focused on improving patient safety should consider quantifying and exploring these factors. Third, findings from our survey within Michigan suggest the need for similar, larger studies across the country. Partnerships with nursing organizations or larger professional groups that represent vascular nursing specialists may be helpful in this regard.

In conclusion, we found important similarities and differences in vascular nursing experience, practice, knowledge, and beliefs in Michigan. These data are useful as they help provide context regarding the constitution of these teams, current practices, and opportunities for improving care. Hospitalists seeking to improve patient safety may use these data to better inform vascular access practice in hospitalized patients.

Peripherally inserted central catheters (PICCs) are among the most prevalent of venous access devices in hospitalized patients.[1, 2] Although growing use of these devices reflects clinical advantages, such as a reduced risk of complications during insertion and durable venous access, use of PICCs is also likely related to the growth of vascular access nursing.[3, 4] A relatively new specialty, vascular access nurses obtain, maintain, and manage venous access in hospitalized patients.[4, 5] Depending on their scope of practice, these professionals are responsible not only for insertion of devices, such as peripheral intravenous catheters and PICCs, but also nontunneled central venous catheters and arterial catheters in some settings.[6]

Although a growing number of US hospitals have introduced vascular nursing teams,[7] little is known about the experience, practice, knowledge, and beliefs of vascular access nurses. This knowledge gap is relevant for hospitalists and hospital medicine as (1) vascular access nurses increasingly represent a key partner in the care of hospitalized patients; (2) the knowledge and practice of these individuals directly affects patient safety and clinical outcomes; and (3) understanding experience, practice, and beliefs of these specialists can help inform decision making and quality‐improvement efforts related to PICCs. As hospitalists increasingly order the placement of and care for patients with PICCs, they are also well suited to improve PICC practice.

Therefore, we conducted a survey of vascular access nurses employed by hospitals that participate in the Michigan Hospital Medicine Safety (HMS) Consortium, a Blue Cross Blue Shield of Michiganfunded collaborative quality initiative.[6] We aimed to understand experience, practice, knowledge, and beliefs related to PICC care and use.

METHODS

RESULTS

Of 172 vascular nurses who received invitations, 140 completed the survey for a response rate of 81%. Respondents reported working in not‐for‐profit hospitals (36%), academic medical centers (29%), and for‐profit hospitals (21%). Although multiple providers (eg, interventional radiology staff and providers, physicians) placed PICCs, 95% of those surveyed reported that they placed the majority of the PICCs at their institutions. Although most respondents placed PICCs in adult patients (86%), a few also placed PICCs in pediatric populations (17%). Vascular nursing programs were largely housed in their own department, but some reported to general nursing or subspecialties such as interventional radiology, cardiology, and critical care. Most respondents indicated their facilities had written policies regarding standard insertion and care practices (87% and 95%, respectively), but only 30% had policies regarding the necessity or appropriateness of PICCs.

Experience among respondents was variable: approximately a third had placed PICCs for <5 years (28.6%), whereas 58% reported placing PICCs for 5 years Correspondingly, 26% reported having placed 100 to 500 PICCs, whereas 34% had placed 1000 or more PICCs. Only 23% of those surveyed held a dedicated vascular access certification, such as board certified in vascular access or certified registered nurse infusion, whereas 16% indicated that they served as the vascular access lead nurse for their facility. Following placement, 94% of respondents reported that their facilities tracked the number of PICCs inserted, but only 40% indicated that dwell times of devices were also recorded. Only 30% of nurses reported that their hospitals had a written policy to evaluate PICC necessity or appropriateness following placement (Table 1).

The most commonly reported indications for PICC placement included intravenous antibiotics at discharge, difficult venous access, and placement for chemotherapy in patients with cancer. Forty‐six percent of nurses indicated they had placed a PICC in a patient receiving some form of dialysis in the past several months; however, 91% of these respondents reported receiving approval from nephrology prior to placement in these patients. Although almost all nurses (91%) used ultrasound to find a suitable vein for PICC placement, a smaller percentage used ultrasound to estimate the catheter‐to‐vein ratio to prevent thrombosis (79%), and only a few (14%) documented this figure in the medical record. Three‐quarters of those surveyed (76%) indicated they used ECG‐based systems to position PICC tips at the cavoatrial junction to prevent thrombosis. Of those who used this technology, 36% still obtained chest x‐rays to verify the position of the PICC tip. According to 84% of respondents, flushing of PICCs was performed mainly by bedside nurses, whereas scheduled weekly dressing changes were most often performed by vascular access nurses (Table 2).

With respect to complications, catheter occlusion, migration, and DVT were reported as the 3 most prevalent adverse events. Interestingly, respondents did not report central lineassociated bloodstream infection (CLABSI) as a common complication. Additionally, 51% of those surveyed indicated that physicians unnecessarily removed PICCs when CLABSI was suspected but not confirmed. When managing catheter occlusion, 50% of respondents began with normal saline flushes but used tissue‐plasminogen activator if saline failed to resolve occlusion. Management of catheter migration varied based on degree of device movement: when the PICC had migrated <5 cm, most respondents (77%) indicated they would first obtain a chest x‐ray to determine the position of the PICC tip, with few (4%) performing catheter exchange. However, if the PICC had migrated more than 5 cm, a significantly greater proportion of respondents (21%) indicated they would perform a catheter exchange. With regard to managing DVT, most vascular nurses reported they notified nurses and physicians to continue using the PICC but recommended tests to confirm the diagnosis.

To better understand the experiences of vascular nurses, we asked for their perceptions regarding appropriateness of PICC use and relationships with bedside nurses, physicians, and leadership. Over a third of respondents (36%) felt that <5% of all PICCs may be inappropriate in their facility, whereas 1 in 5 indicated that 10% to 24% of PICCs placed in their facilities may be inappropriate or could have been avoided. Almost all (98%) of the nurses stated they were not empowered to remove idle or clinically unnecessary PICCs without physician authorization. Although 51% of nurses described the support received from hospital leadership as excellent, very good, or good, 43% described leadership support as either fair or poor. Conversely, relationships with bedside nurses and physicians were rated as being very good or good by nearly two‐thirds of those surveyed (64% and 65%, respectively) (Table 3).

DISCUSSION

In this survey of 140 vascular access nurses in hospitals across Michigan, new insights regarding the experience, practice, knowledge, and beliefs of this group of providers were obtained. We found that vascular access nurses varied with respect to years in practice, volume of PICCs placed, and certification status, reflecting heterogeneity in this provider group. Variation in insertion techniques, such as use of ultrasound to examine catheter‐to‐vein ratio (a key way to prevent thrombosis) or newer ECG technology to position the PICC, was also noted. Although indications for PICC insertion appeared consistent with published literature, the frequency with which these devices were placed in patients receiving dialysis (reportedly with nephrology approval) was surprising given national calls to avoid such use.[16] Opportunities to improve hospital practices, such as tracking PICC dwell times and PICC necessity, as well as the potential need to better educate physicians on when to remove PICCs for suspected CLABSI, were also identified. Collectively, these data are highly relevant to hospitalists and health systems as they help to identify areas for quality improvement and inform clinical practice regarding the use of PICCs in hospitalized patients. As hospitalists increasingly order PICCs and manage complications associated with these devices, they are well suited to use these data so as to improve patient safety and clinical outcomes.

Venous access is the most common medical procedure performed in hospitalized medical patients. Although a number of devices including peripheral intravenous catheters, central venous catheters, and PICCs are used for this purpose, the growing use of PICCs to secure venous access has been documented in several studies.[17] Such growth, in part, undoubtedly reflects increasing availability of vascular access nurses. Traditionally placed by interventional radiologists, the creation of dedicated vascular nursing teams has resulted in these subspecialists now serving in more of a backup or trouble‐shooting role rather than that of primary operator.[4, 14] This paradigm shift is well illustrated in a recent survey of infection preventionists, where over 60% of respondents reported that they had a vascular nursing team in their facility.[7] The growth of these nursing‐led vascular access teams has produced not only high rates of insertion success and low rates of complications, but also greater cost‐effectiveness when compared to interventional radiologybased insertion.[18]

Nonetheless, our survey also identified a number of important concerns regarding PICC practices and vascular nursing providers. First, we found variation in areas such as insertion practices and management of complications. Such variability highlights the importance of both growing and disseminating the evidence base for consistent practice in vascular nursing. Through their close clinical affiliation with vascular nurses and shared interests in obtaining safe and appropriate venous access for patients, hospitalists are ideally poised to lead this effort. Second, similarities between vascular nurse opinions regarding appropriateness of PICCs and those of hospitalists from a prior survey were noted.[19] Namely, a substantial proportion of both vascular nurses and hospitalists felt that some PICCs were inappropriate and could be avoided. Third, although relationships between vascular access nurses and leadership were reported as being variable, the survey responses suggested relatively good interprovider relationships with bedside nurses and physicians. Such relationships likely reflect the close clinical ties that emerge from being in the trenches of patient care and suggest that interventions to improve care in partnership with these providers are highly viable.

Our study has some limitations. First, despite a high response rate, our study used a survey design and reports findings from a convenience sample of vascular access nurses in a single state. Thus, nonrespondent and selection biases remain threats to our conclusions. Additionally, some respondents did not complete all responses, perhaps due to nonapplicability to practice or other unknown reasons. The pattern of missingness observed, however, suggested that such responses were missing at random. Second, we surveyed vascular nurses in hospitals that are actively engaged in improving PICC practices; our findings may therefore not be representative of vascular nursing professionals as a whole and may instead reflect those of a highly motivated group of individuals. Relatedly, the underlying reasons for adoption of specific practices or techniques cannot be discerned from our study. Third, although we did not find differences based on years in practice or certification status, our sample size was relatively small and likely underpowered for these comparisons. Finally, our study sample consists of vascular nurses who are clustered within hospitals in which they are employed. Therefore, overlap between reported practices and those required by the facility are possible.

Despite these limitations, our study has important strengths. First, this is among the most comprehensive of surveys examining vascular nursing experience, practice, knowledge, and beliefs. The growing presence of these providers across US hospitals, coupled with limited insight regarding their clinical practices, highlight the importance and utility of these data. Second, we noted important differences in experience, practices, and interprovider relationships between vascular providers in this field. Although we are unable to ascertain the drivers or significance of such variation, hospitals and health systems focused on improving patient safety should consider quantifying and exploring these factors. Third, findings from our survey within Michigan suggest the need for similar, larger studies across the country. Partnerships with nursing organizations or larger professional groups that represent vascular nursing specialists may be helpful in this regard.

In conclusion, we found important similarities and differences in vascular nursing experience, practice, knowledge, and beliefs in Michigan. These data are useful as they help provide context regarding the constitution of these teams, current practices, and opportunities for improving care. Hospitalists seeking to improve patient safety may use these data to better inform vascular access practice in hospitalized patients.

Improving quality while reducing costs remains important for hospitals across the United States, including the approximately 150 hospitals that are part of the Veterans Affairs (VA) healthcare system.[1, 2] The field of hospital medicine has grown rapidly, leading to predictions that the majority of inpatient care in the United States eventually will be delivered by hospitalists.[3, 4] In 2010, 57% of US hospitals had hospitalists on staff, including 87% of hospitals with 200 beds,[5] and nearly 80% of VA hospitals.[6]

The demand for hospitalists within teaching hospitals has grown in part as a response to the mandate to reduce residency work hours.[7] Furthermore, previous research has found that hospitalist care is associated with modest reductions in length of stay (LOS) and weak but inconsistent differences in quality.[8] The educational effect of hospitalists has been far less examined. The limited number of studies published to date suggests that hospitalists may improve resident learning and house‐officer satisfaction in academic medical centers and community teaching hospitals[9, 10, 11] and provide positive experiences for medical students^12,13; however, Wachter et al reported no significant changes in clinical outcomes or patient, faculty, and house‐staff satisfaction in a newly designed hospital medicine service in San Francisco.[14] Additionally, whether using hospitalists influences nurse‐physician communication[15] is unknown.

Recognizing the limited and sometimes conflicting evidence about the hospitalist model, we report the results of a 3‐year quasi‐experimental evaluation of the experience at our medical center with academic hospitalists. As part of a VA Systems Redesign Improvement Capability Grantknown as the Hospital Outcomes Program of Excellence (HOPE) Initiativewe created a hospitalist‐based medicine team focused on quality improvement, medical education, and patient outcomes.

METHODS

RESULTS

Clinical Outcomes

Patient Outcomes

Our multivariable linear regression analysis, controlling for time, showed a significant reduction in LOS of approximately 0.3 days on all teams after the HOPE Initiative began (P=0.004). There were no significant differences between the Gold and non‐Gold teams in the multivariate models when controlling for time for any of the patient‐outcome measures. The number of admissions increased for all 4 medical teams (Figure 1), but, as shown in Figures 2 and 3, the readmission rates for all teams remained relatively stable over this same period of time.

Figure 1

Figure 2

Figure 3

Clinical Pharmacist on Gold Team Rounds

The inpatient clinical pharmacist changed the management plan for 22% of the patients seen on rounds. Contributions from the clinical pharmacist included adjusting the dosing of ordered medication and correcting medication reconciliation. Education and pharmaceutical information was provided to the team in another 6% of the 170 consecutive patient encounters evaluated.

Perception of Circle of Concern Rounds

Circle of Concern rounds were generally well‐received by both nurses and physicians. In a healthcare‐worker communication survey, completed by 38 physicians (62% response rate) and 48 nurses (54% response rate), the majority of both physicians (83%) and nurses (68%) felt Circle of Concern rounds improved communication.

Nurse Perception of Communication

The healthcare‐worker communication survey asked inpatient nurses to rate communication between nurses and physicians on each of the 4 medicine teams. Significantly more nurses were satisfied with communication with the Gold team (71%) compared with the other 3 medicine teams (53%; P=0.02) (Figure 4).

Figure 4

Perception of the Clinical Care Coordinator

In total, 20 physicians (87% response rate) and 10 nurses (56% response rate) completed the Clinical Care Coordinator survey. The physician results were overwhelmingly positive: 100% were satisfied or very satisfied with the role; 100% felt each team should have a Clinical Care Coordinator; and 100% agreed or strongly agreed that the Clinical Care Coordinator ensures that appropriate follow‐up is arranged, provides continuity of care, assists with interdisciplinary communication, and helps facilitate discharge. The majority of nurses was also satisfied or very satisfied with the Clinical Care Coordinator role and felt each team should have one.

Educational Outcomes

House Officer Evaluation of Attendings

Monthly evaluations of attending physicians by house officers (Figure 5) revealed that prior to the HOPE Initiative, little differences were observed between teams, as would be expected because attending assignment was largely random. After the intervention date of July 2009, however, significant differences were noted, with Gold team attendings receiving significantly higher teaching evaluations immediately after the introduction of the HOPE Initiative. Although ratings for Gold attendings remained more favorable, the difference was no longer statistically significant in the second and third year of the initiative, likely due to Gold attendings serving on other medicine teams, which contributed to an improvement in ratings of all attendings.

Figure 5

Medical Student Evaluation of Attendings

Monthly evaluations of attending physicians by third‐year medical students (Figure 6) revealed differences between the Gold attendings and all others, with the attendings that joined the Gold team in 2009 receiving higher teaching evaluations even before the HOPE Initiative started. However, this difference remained statistically significant in years 2 and 3 postinitiative, despite the addition of 4 new junior attendings.

Figure 6

Medical Student Medicine Shelf Scores

The national average on the shelf exam, which reflects learning after the internal medicine third‐year clerkship, has ranged from 75 to 78 for the past several years, with University of Michigan students averaging significantly higher scores prior to and after the HOPE Initiative. However, following the HOPE Initiative, third‐year medical students on the Gold team scored significantly higher on the shelf exam compared with their colleagues on the non‐Gold teams (84 vs 82; P=0.006). This difference in the shelf exam scores, although small, is statistically significant. It represents a measurable improvement in shelf scores in our system and demonstrates the potential educational benefit for the students. Over this same time period, scores on the United States Medical Licensing Exam, given to medical students at the beginning of their third year, remained stable (233 preHOPE Initiative; 234 postHOPE Initiative).

DISCUSSION

Within academic medical centers, hospitalists are expected to care for patients, teach, and help improve the quality and efficiency of hospital‐based care.[7] The Department of Veterans Affairs runs the largest integrated healthcare system in the United States, with approximately 80% of VA hospitals having hospital medicine programs. Overall, one‐third of US residents perform part of their residency training at a VA hospital.[22, 23] Thus, the effects of a system‐wide change at a VA hospital may have implications throughout the country. We studied one such intervention. Our primary findings are that we were able to improve communication and learner education with minimal effects on patient outcomes. While overall LOS decreased slightly postintervention, after taking into account secular trends, readmission rates did not.

We are not the first to evaluate a hospital medicine team using a quasi‐experimental design. For example, Meltzer and colleagues evaluated a hospitalist program at the University of Chicago Medical Center and found that, by the second year of operation, hospitalist care was associated with significantly shorter LOS (0.49 days), reduced costs, and decreased mortality.[24] Auerbach also evaluated a newly created hospital medicine service, finding decreased LOS (0.61 days), lower costs, and lower risk of mortality by the second year of the program.[25]

Improving nurse‐physician communication is considered important for avoiding medical error,[26] yet there has been limited empirical study of methods to improve communication within the medical profession.[27] Based both on our surveys and qualitative interviews, healthcare‐worker communication appeared to improve on the Gold team during the study. A key component of this improvement is likely related to instituting Circle of Concern rounds, in which nurses joined the medical team during attending rounds. Such an intervention likely helped to address organizational silence[28] and enhance the psychological safety of the nursing staff, because the attending physician was proactive about soliciting the input of nurses during rounds.[29] Such leader inclusivenesswords and deeds exhibited by leaders that invite and appreciate others' contributionscan aid interdisciplinary teams in overcoming the negative effects of status differences, thereby promoting collaboration.[29] The inclusion of nurses on rounds is also relationship‐building, which Gotlib Conn and colleagues found was important to improved interprofessional communication and collaboration.[30] In the future, using a tool such as the Teamwork Effectiveness Assessment Module (TEAM) developed by the American Board of Internal Medicine[31] could provide further evaluation of the impact on interprofessional teamwork and communication.

The focus on learner education, though evaluated in prior studies, is also novel. One previous survey of medical students showed that engaging students in substantive discussions is associated with greater student satisfaction.[32] Another survey of medical students found that attendings who were enthusiastic about teaching, inspired confidence in knowledge and skills, provided useful feedback, and encouraged increased student responsibility were viewed as more effective teachers.[33] No previous study that we are aware of, however, has looked at actual educational outcomes, such as shelf scores. The National Board of Medical Examiners reports that the Medicine subject exam is scaled to have a mean of 70 and a standard deviation of 8.[34] Thus, a mean increase in score of 2 points is small, but not trivial. This shows improvement in a hard educational outcome. Additionally, 2 points, although small in the context of total score and standard deviation, may make a substantial difference to an individual student in terms of overall grade, and, thus, residency applications. Our finding that third‐year medical students on the Gold team performed significantly better than University of Michigan third‐year medical students on other teams is an intriguing finding that warrants confirmation. On the other hand, this finding is consistent with a previous report evaluating learner satisfaction in which Bodnar et al found improved ratings of quantity and quality of teaching on teams with a nontraditional structure (Gold team).[21] Moreover, despite relatively few studies, the reason underlying the educational benefit of hospitalists should surprise few. The hospitalist model ensures that learners are supervised by physicians who are experts in the care of hospitalized patients.[35] Hospitalists hired at teaching hospitals to work on services with learners are generally chosen because they possess superior educational skills.[7]

Our findings should be interpreted in the context of the following limitations. First, our study focused on a single academically affiliated VA hospital. As other VA hospitals are pursuing a similar approach (eg, the Houston and Detroit VA medical centers), replicating our results will be important. Second, the VA system, although the largest integrated healthcare system in the United States, has unique characteristicssuch as an integrated electronic health record and predominantly male patient populationthat may make generalizations to the larger US healthcare system challenging. Third, there was a slightly lower response rate among nurses on a few of the surveys to evaluate our efforts; however, this rate of response is standard at our facility. Finally, our evaluation lacks an empirical measure of healthcare‐worker communication, such as incident reports.

Despite these limitations, our results have important implications. Using both quantitative and qualitative assessment, we found that academic hospitalists have the ability to improve healthcare‐worker communication and enhance learner education without increasing LOS. These findings are directly applicable to VA medical centers and potentially applicable to other academic medical centers.

Improving quality while reducing costs remains important for hospitals across the United States, including the approximately 150 hospitals that are part of the Veterans Affairs (VA) healthcare system.[1, 2] The field of hospital medicine has grown rapidly, leading to predictions that the majority of inpatient care in the United States eventually will be delivered by hospitalists.[3, 4] In 2010, 57% of US hospitals had hospitalists on staff, including 87% of hospitals with 200 beds,[5] and nearly 80% of VA hospitals.[6]

The demand for hospitalists within teaching hospitals has grown in part as a response to the mandate to reduce residency work hours.[7] Furthermore, previous research has found that hospitalist care is associated with modest reductions in length of stay (LOS) and weak but inconsistent differences in quality.[8] The educational effect of hospitalists has been far less examined. The limited number of studies published to date suggests that hospitalists may improve resident learning and house‐officer satisfaction in academic medical centers and community teaching hospitals[9, 10, 11] and provide positive experiences for medical students^12,13; however, Wachter et al reported no significant changes in clinical outcomes or patient, faculty, and house‐staff satisfaction in a newly designed hospital medicine service in San Francisco.[14] Additionally, whether using hospitalists influences nurse‐physician communication[15] is unknown.

Recognizing the limited and sometimes conflicting evidence about the hospitalist model, we report the results of a 3‐year quasi‐experimental evaluation of the experience at our medical center with academic hospitalists. As part of a VA Systems Redesign Improvement Capability Grantknown as the Hospital Outcomes Program of Excellence (HOPE) Initiativewe created a hospitalist‐based medicine team focused on quality improvement, medical education, and patient outcomes.

METHODS

RESULTS

Clinical Outcomes

Patient Outcomes

Our multivariable linear regression analysis, controlling for time, showed a significant reduction in LOS of approximately 0.3 days on all teams after the HOPE Initiative began (P=0.004). There were no significant differences between the Gold and non‐Gold teams in the multivariate models when controlling for time for any of the patient‐outcome measures. The number of admissions increased for all 4 medical teams (Figure 1), but, as shown in Figures 2 and 3, the readmission rates for all teams remained relatively stable over this same period of time.

Figure 1

Figure 2

Figure 3

Clinical Pharmacist on Gold Team Rounds

The inpatient clinical pharmacist changed the management plan for 22% of the patients seen on rounds. Contributions from the clinical pharmacist included adjusting the dosing of ordered medication and correcting medication reconciliation. Education and pharmaceutical information was provided to the team in another 6% of the 170 consecutive patient encounters evaluated.

Perception of Circle of Concern Rounds

Circle of Concern rounds were generally well‐received by both nurses and physicians. In a healthcare‐worker communication survey, completed by 38 physicians (62% response rate) and 48 nurses (54% response rate), the majority of both physicians (83%) and nurses (68%) felt Circle of Concern rounds improved communication.

Nurse Perception of Communication

The healthcare‐worker communication survey asked inpatient nurses to rate communication between nurses and physicians on each of the 4 medicine teams. Significantly more nurses were satisfied with communication with the Gold team (71%) compared with the other 3 medicine teams (53%; P=0.02) (Figure 4).

Figure 4

Perception of the Clinical Care Coordinator

In total, 20 physicians (87% response rate) and 10 nurses (56% response rate) completed the Clinical Care Coordinator survey. The physician results were overwhelmingly positive: 100% were satisfied or very satisfied with the role; 100% felt each team should have a Clinical Care Coordinator; and 100% agreed or strongly agreed that the Clinical Care Coordinator ensures that appropriate follow‐up is arranged, provides continuity of care, assists with interdisciplinary communication, and helps facilitate discharge. The majority of nurses was also satisfied or very satisfied with the Clinical Care Coordinator role and felt each team should have one.

Educational Outcomes

House Officer Evaluation of Attendings

Monthly evaluations of attending physicians by house officers (Figure 5) revealed that prior to the HOPE Initiative, little differences were observed between teams, as would be expected because attending assignment was largely random. After the intervention date of July 2009, however, significant differences were noted, with Gold team attendings receiving significantly higher teaching evaluations immediately after the introduction of the HOPE Initiative. Although ratings for Gold attendings remained more favorable, the difference was no longer statistically significant in the second and third year of the initiative, likely due to Gold attendings serving on other medicine teams, which contributed to an improvement in ratings of all attendings.

Figure 5

Medical Student Evaluation of Attendings

Monthly evaluations of attending physicians by third‐year medical students (Figure 6) revealed differences between the Gold attendings and all others, with the attendings that joined the Gold team in 2009 receiving higher teaching evaluations even before the HOPE Initiative started. However, this difference remained statistically significant in years 2 and 3 postinitiative, despite the addition of 4 new junior attendings.

Figure 6

Medical Student Medicine Shelf Scores

The national average on the shelf exam, which reflects learning after the internal medicine third‐year clerkship, has ranged from 75 to 78 for the past several years, with University of Michigan students averaging significantly higher scores prior to and after the HOPE Initiative. However, following the HOPE Initiative, third‐year medical students on the Gold team scored significantly higher on the shelf exam compared with their colleagues on the non‐Gold teams (84 vs 82; P=0.006). This difference in the shelf exam scores, although small, is statistically significant. It represents a measurable improvement in shelf scores in our system and demonstrates the potential educational benefit for the students. Over this same time period, scores on the United States Medical Licensing Exam, given to medical students at the beginning of their third year, remained stable (233 preHOPE Initiative; 234 postHOPE Initiative).

DISCUSSION

Within academic medical centers, hospitalists are expected to care for patients, teach, and help improve the quality and efficiency of hospital‐based care.[7] The Department of Veterans Affairs runs the largest integrated healthcare system in the United States, with approximately 80% of VA hospitals having hospital medicine programs. Overall, one‐third of US residents perform part of their residency training at a VA hospital.[22, 23] Thus, the effects of a system‐wide change at a VA hospital may have implications throughout the country. We studied one such intervention. Our primary findings are that we were able to improve communication and learner education with minimal effects on patient outcomes. While overall LOS decreased slightly postintervention, after taking into account secular trends, readmission rates did not.

We are not the first to evaluate a hospital medicine team using a quasi‐experimental design. For example, Meltzer and colleagues evaluated a hospitalist program at the University of Chicago Medical Center and found that, by the second year of operation, hospitalist care was associated with significantly shorter LOS (0.49 days), reduced costs, and decreased mortality.[24] Auerbach also evaluated a newly created hospital medicine service, finding decreased LOS (0.61 days), lower costs, and lower risk of mortality by the second year of the program.[25]

Improving nurse‐physician communication is considered important for avoiding medical error,[26] yet there has been limited empirical study of methods to improve communication within the medical profession.[27] Based both on our surveys and qualitative interviews, healthcare‐worker communication appeared to improve on the Gold team during the study. A key component of this improvement is likely related to instituting Circle of Concern rounds, in which nurses joined the medical team during attending rounds. Such an intervention likely helped to address organizational silence[28] and enhance the psychological safety of the nursing staff, because the attending physician was proactive about soliciting the input of nurses during rounds.[29] Such leader inclusivenesswords and deeds exhibited by leaders that invite and appreciate others' contributionscan aid interdisciplinary teams in overcoming the negative effects of status differences, thereby promoting collaboration.[29] The inclusion of nurses on rounds is also relationship‐building, which Gotlib Conn and colleagues found was important to improved interprofessional communication and collaboration.[30] In the future, using a tool such as the Teamwork Effectiveness Assessment Module (TEAM) developed by the American Board of Internal Medicine[31] could provide further evaluation of the impact on interprofessional teamwork and communication.

The focus on learner education, though evaluated in prior studies, is also novel. One previous survey of medical students showed that engaging students in substantive discussions is associated with greater student satisfaction.[32] Another survey of medical students found that attendings who were enthusiastic about teaching, inspired confidence in knowledge and skills, provided useful feedback, and encouraged increased student responsibility were viewed as more effective teachers.[33] No previous study that we are aware of, however, has looked at actual educational outcomes, such as shelf scores. The National Board of Medical Examiners reports that the Medicine subject exam is scaled to have a mean of 70 and a standard deviation of 8.[34] Thus, a mean increase in score of 2 points is small, but not trivial. This shows improvement in a hard educational outcome. Additionally, 2 points, although small in the context of total score and standard deviation, may make a substantial difference to an individual student in terms of overall grade, and, thus, residency applications. Our finding that third‐year medical students on the Gold team performed significantly better than University of Michigan third‐year medical students on other teams is an intriguing finding that warrants confirmation. On the other hand, this finding is consistent with a previous report evaluating learner satisfaction in which Bodnar et al found improved ratings of quantity and quality of teaching on teams with a nontraditional structure (Gold team).[21] Moreover, despite relatively few studies, the reason underlying the educational benefit of hospitalists should surprise few. The hospitalist model ensures that learners are supervised by physicians who are experts in the care of hospitalized patients.[35] Hospitalists hired at teaching hospitals to work on services with learners are generally chosen because they possess superior educational skills.[7]

Our findings should be interpreted in the context of the following limitations. First, our study focused on a single academically affiliated VA hospital. As other VA hospitals are pursuing a similar approach (eg, the Houston and Detroit VA medical centers), replicating our results will be important. Second, the VA system, although the largest integrated healthcare system in the United States, has unique characteristicssuch as an integrated electronic health record and predominantly male patient populationthat may make generalizations to the larger US healthcare system challenging. Third, there was a slightly lower response rate among nurses on a few of the surveys to evaluate our efforts; however, this rate of response is standard at our facility. Finally, our evaluation lacks an empirical measure of healthcare‐worker communication, such as incident reports.

Despite these limitations, our results have important implications. Using both quantitative and qualitative assessment, we found that academic hospitalists have the ability to improve healthcare‐worker communication and enhance learner education without increasing LOS. These findings are directly applicable to VA medical centers and potentially applicable to other academic medical centers.

Improving quality while reducing costs remains important for hospitals across the United States, including the approximately 150 hospitals that are part of the Veterans Affairs (VA) healthcare system.[1, 2] The field of hospital medicine has grown rapidly, leading to predictions that the majority of inpatient care in the United States eventually will be delivered by hospitalists.[3, 4] In 2010, 57% of US hospitals had hospitalists on staff, including 87% of hospitals with 200 beds,[5] and nearly 80% of VA hospitals.[6]

The demand for hospitalists within teaching hospitals has grown in part as a response to the mandate to reduce residency work hours.[7] Furthermore, previous research has found that hospitalist care is associated with modest reductions in length of stay (LOS) and weak but inconsistent differences in quality.[8] The educational effect of hospitalists has been far less examined. The limited number of studies published to date suggests that hospitalists may improve resident learning and house‐officer satisfaction in academic medical centers and community teaching hospitals[9, 10, 11] and provide positive experiences for medical students^12,13; however, Wachter et al reported no significant changes in clinical outcomes or patient, faculty, and house‐staff satisfaction in a newly designed hospital medicine service in San Francisco.[14] Additionally, whether using hospitalists influences nurse‐physician communication[15] is unknown.

Recognizing the limited and sometimes conflicting evidence about the hospitalist model, we report the results of a 3‐year quasi‐experimental evaluation of the experience at our medical center with academic hospitalists. As part of a VA Systems Redesign Improvement Capability Grantknown as the Hospital Outcomes Program of Excellence (HOPE) Initiativewe created a hospitalist‐based medicine team focused on quality improvement, medical education, and patient outcomes.

METHODS

RESULTS

Clinical Outcomes

Patient Outcomes

Our multivariable linear regression analysis, controlling for time, showed a significant reduction in LOS of approximately 0.3 days on all teams after the HOPE Initiative began (P=0.004). There were no significant differences between the Gold and non‐Gold teams in the multivariate models when controlling for time for any of the patient‐outcome measures. The number of admissions increased for all 4 medical teams (Figure 1), but, as shown in Figures 2 and 3, the readmission rates for all teams remained relatively stable over this same period of time.

Figure 1

Figure 2

Figure 3

Clinical Pharmacist on Gold Team Rounds

The inpatient clinical pharmacist changed the management plan for 22% of the patients seen on rounds. Contributions from the clinical pharmacist included adjusting the dosing of ordered medication and correcting medication reconciliation. Education and pharmaceutical information was provided to the team in another 6% of the 170 consecutive patient encounters evaluated.

Perception of Circle of Concern Rounds

Circle of Concern rounds were generally well‐received by both nurses and physicians. In a healthcare‐worker communication survey, completed by 38 physicians (62% response rate) and 48 nurses (54% response rate), the majority of both physicians (83%) and nurses (68%) felt Circle of Concern rounds improved communication.

Nurse Perception of Communication

The healthcare‐worker communication survey asked inpatient nurses to rate communication between nurses and physicians on each of the 4 medicine teams. Significantly more nurses were satisfied with communication with the Gold team (71%) compared with the other 3 medicine teams (53%; P=0.02) (Figure 4).

Figure 4

Perception of the Clinical Care Coordinator

In total, 20 physicians (87% response rate) and 10 nurses (56% response rate) completed the Clinical Care Coordinator survey. The physician results were overwhelmingly positive: 100% were satisfied or very satisfied with the role; 100% felt each team should have a Clinical Care Coordinator; and 100% agreed or strongly agreed that the Clinical Care Coordinator ensures that appropriate follow‐up is arranged, provides continuity of care, assists with interdisciplinary communication, and helps facilitate discharge. The majority of nurses was also satisfied or very satisfied with the Clinical Care Coordinator role and felt each team should have one.

Educational Outcomes

House Officer Evaluation of Attendings

Monthly evaluations of attending physicians by house officers (Figure 5) revealed that prior to the HOPE Initiative, little differences were observed between teams, as would be expected because attending assignment was largely random. After the intervention date of July 2009, however, significant differences were noted, with Gold team attendings receiving significantly higher teaching evaluations immediately after the introduction of the HOPE Initiative. Although ratings for Gold attendings remained more favorable, the difference was no longer statistically significant in the second and third year of the initiative, likely due to Gold attendings serving on other medicine teams, which contributed to an improvement in ratings of all attendings.

Figure 5

Medical Student Evaluation of Attendings

Monthly evaluations of attending physicians by third‐year medical students (Figure 6) revealed differences between the Gold attendings and all others, with the attendings that joined the Gold team in 2009 receiving higher teaching evaluations even before the HOPE Initiative started. However, this difference remained statistically significant in years 2 and 3 postinitiative, despite the addition of 4 new junior attendings.

Figure 6

Medical Student Medicine Shelf Scores

The national average on the shelf exam, which reflects learning after the internal medicine third‐year clerkship, has ranged from 75 to 78 for the past several years, with University of Michigan students averaging significantly higher scores prior to and after the HOPE Initiative. However, following the HOPE Initiative, third‐year medical students on the Gold team scored significantly higher on the shelf exam compared with their colleagues on the non‐Gold teams (84 vs 82; P=0.006). This difference in the shelf exam scores, although small, is statistically significant. It represents a measurable improvement in shelf scores in our system and demonstrates the potential educational benefit for the students. Over this same time period, scores on the United States Medical Licensing Exam, given to medical students at the beginning of their third year, remained stable (233 preHOPE Initiative; 234 postHOPE Initiative).

DISCUSSION

Within academic medical centers, hospitalists are expected to care for patients, teach, and help improve the quality and efficiency of hospital‐based care.[7] The Department of Veterans Affairs runs the largest integrated healthcare system in the United States, with approximately 80% of VA hospitals having hospital medicine programs. Overall, one‐third of US residents perform part of their residency training at a VA hospital.[22, 23] Thus, the effects of a system‐wide change at a VA hospital may have implications throughout the country. We studied one such intervention. Our primary findings are that we were able to improve communication and learner education with minimal effects on patient outcomes. While overall LOS decreased slightly postintervention, after taking into account secular trends, readmission rates did not.

We are not the first to evaluate a hospital medicine team using a quasi‐experimental design. For example, Meltzer and colleagues evaluated a hospitalist program at the University of Chicago Medical Center and found that, by the second year of operation, hospitalist care was associated with significantly shorter LOS (0.49 days), reduced costs, and decreased mortality.[24] Auerbach also evaluated a newly created hospital medicine service, finding decreased LOS (0.61 days), lower costs, and lower risk of mortality by the second year of the program.[25]

Improving nurse‐physician communication is considered important for avoiding medical error,[26] yet there has been limited empirical study of methods to improve communication within the medical profession.[27] Based both on our surveys and qualitative interviews, healthcare‐worker communication appeared to improve on the Gold team during the study. A key component of this improvement is likely related to instituting Circle of Concern rounds, in which nurses joined the medical team during attending rounds. Such an intervention likely helped to address organizational silence[28] and enhance the psychological safety of the nursing staff, because the attending physician was proactive about soliciting the input of nurses during rounds.[29] Such leader inclusivenesswords and deeds exhibited by leaders that invite and appreciate others' contributionscan aid interdisciplinary teams in overcoming the negative effects of status differences, thereby promoting collaboration.[29] The inclusion of nurses on rounds is also relationship‐building, which Gotlib Conn and colleagues found was important to improved interprofessional communication and collaboration.[30] In the future, using a tool such as the Teamwork Effectiveness Assessment Module (TEAM) developed by the American Board of Internal Medicine[31] could provide further evaluation of the impact on interprofessional teamwork and communication.

The focus on learner education, though evaluated in prior studies, is also novel. One previous survey of medical students showed that engaging students in substantive discussions is associated with greater student satisfaction.[32] Another survey of medical students found that attendings who were enthusiastic about teaching, inspired confidence in knowledge and skills, provided useful feedback, and encouraged increased student responsibility were viewed as more effective teachers.[33] No previous study that we are aware of, however, has looked at actual educational outcomes, such as shelf scores. The National Board of Medical Examiners reports that the Medicine subject exam is scaled to have a mean of 70 and a standard deviation of 8.[34] Thus, a mean increase in score of 2 points is small, but not trivial. This shows improvement in a hard educational outcome. Additionally, 2 points, although small in the context of total score and standard deviation, may make a substantial difference to an individual student in terms of overall grade, and, thus, residency applications. Our finding that third‐year medical students on the Gold team performed significantly better than University of Michigan third‐year medical students on other teams is an intriguing finding that warrants confirmation. On the other hand, this finding is consistent with a previous report evaluating learner satisfaction in which Bodnar et al found improved ratings of quantity and quality of teaching on teams with a nontraditional structure (Gold team).[21] Moreover, despite relatively few studies, the reason underlying the educational benefit of hospitalists should surprise few. The hospitalist model ensures that learners are supervised by physicians who are experts in the care of hospitalized patients.[35] Hospitalists hired at teaching hospitals to work on services with learners are generally chosen because they possess superior educational skills.[7]

Our findings should be interpreted in the context of the following limitations. First, our study focused on a single academically affiliated VA hospital. As other VA hospitals are pursuing a similar approach (eg, the Houston and Detroit VA medical centers), replicating our results will be important. Second, the VA system, although the largest integrated healthcare system in the United States, has unique characteristicssuch as an integrated electronic health record and predominantly male patient populationthat may make generalizations to the larger US healthcare system challenging. Third, there was a slightly lower response rate among nurses on a few of the surveys to evaluate our efforts; however, this rate of response is standard at our facility. Finally, our evaluation lacks an empirical measure of healthcare‐worker communication, such as incident reports.

Despite these limitations, our results have important implications. Using both quantitative and qualitative assessment, we found that academic hospitalists have the ability to improve healthcare‐worker communication and enhance learner education without increasing LOS. These findings are directly applicable to VA medical centers and potentially applicable to other academic medical centers.