We have known for years that the distribution of healthcare expenditures in the United States is skewed, with a small portion of the population consuming a disproportionately high share of resources. In 2010, 1% of the population accounted for 21.4% of the $1.3 trillion spent on healthcare.[1] Growing evidence documents that most of these high‐cost patients are not receiving coordinated care, preventive care, or care in the most appropriate settings.[2] The term superutilizer describes individuals with complex physical, behavioral, and social needs who have frequent emergency department (ED) visits and multiple costly hospital admissions.[3] Not surprisingly, multiple superutilizer programs and new funding opportunities target this population attempting to reduce their healthcare costs while improving their care, as public and private insurers shift to value‐based care.[4]

Beginning in 2006, the Robert Wood Johnson Foundation supported the Camden Coalition[5] with 3 grants to develop a community‐based approach to identify high‐utilizer patients and provide them with coordinated medical and social services.[6] These programs include community‐based teams that focus on the highest utilizers in a specific geographic area and provide intensive outpatient case management. Building on these efforts, the Center for Medicare and Medicaid Innovation (CMMI) awarded 2 Health Care Innovation Awards totaling $17.2 million to target Medicaid superutilizers.[7] Through its State Innovation Models initiative, CMMI also encourages states to pilot superutilizer programs to increase care coordination and support of persons with certain risk factors such as homelessness or mental illness.[8] Additionally, the National Governors Association developed a 1‐year, multistate policy academy to develop state‐level capacity and state action plans that guide how to improve the delivery and financing of care for superutilizers.[9]

With all these ongoing activities in the setting of a paucity of research identifying the most cost‐efficient practices to manage super‐utilizers, we are glad to see the Journal of Hospital Medicine publish an evaluation of a quality‐improvement project targeting superutilizers.[10] Mercer and colleagues at Duke University Hospital show that developing an individualized care plan and integrating it into their electronic health record (EHR) reduced hospital admissions, but not ED visits. Although we applaud the reportedly individualized patient approach and recognize the effort required to refer patients to a more appropriate care setting, we believe the researchers neglected 3 important components for the intervention: (1) patient engagement in developing individualized care plans, (2) care coordination integrated with community collaboration, and (3) feedback on continuum of care relayed back to providers. The managing strategies mentioned in the article seem to have evolved exclusively from the provider's perspective, a common mistake that the Patient‐Centered Outcomes Research Institute emphasizes must be avoided. We are concerned about the lack of clarity regarding the set of management strategies focused on providing high‐quality care while limiting unnecessary admissions reported by them. We fear this strategy was imposed on patients and not developed collaboratively with them. Effective interventions for superutilizers should do more than just guide providers actions, but also connect services to the patient's needs. There should be coordination and continuous improvement of these efforts, which requires engagement of the patient and their community with feedback to the system.

Possibly most important, an individualized approach to superutilizers needs to be patient‐centeredprioritizing patient goals and preferences, selecting interventions and services guided by the needs of the individual, and emphasizing modifiable outcomes that matter to the patient. Such a patient‐centered approach goes beyond the individual patient to incorporate information about social support and family dynamics, highlighting the role of caregivers. Patients and their caregivers must be engaged or activated to ensure adherence to appropriate care and behaviors in any superutilizer programs. Additionally, individualized patient‐centered care plans should be dynamic and bidirectional to accommodate changes in health priorities that may occur over time. Such lack of patient and community engagement may explain why ED‐visit frequency was unchanged in their study.

The approach of having a Complex Care Plan Committee deserves attention as it appropriately included the right people at the academic medical center. However, why is it voluntary? Should not an important, or even essential, committee such as this be supported by the health system? Moreover, although the care plan developed by members of the committee possesses understandable aspects to be considered in a patient's care, why is this not shown to the patient for their input? Instead of being done to the patient, we recommend including patients in this process, believing such patient engagement would improve care further and likely yield sustained changes. We suggest the researchers remember the maxim nothing about me, without me.

Patients who use the most healthcare services typically have complicated social situations that directly impact their ability to improve their health and stay well.[2, 11] Addressing the social determinants of health is not a new concept; however, creating healthy communities as a core responsibility of the healthcare industry is. Contributing to the dizzying state of change in US healthcare are efforts to shift to value‐based purchasing and population health management.[12] This transformation from a fee‐for‐service hospital‐centric industry into one focused on the continuum of care requires outreach into communities where superutilizers live. Ultimately, all healthcare is local, as this is where patients receive the vast majority of their care. Improving quality and reducing costs requires healthcare providers to work together on a collaborative mission that focuses on the needs of patients and community, not just efforts to reduce utilization. Even hospitalists must forge collaborative relationships with skilled nursing facilities and patient‐centered medical homes.

Given the successes of some superutilizer programs,[3] a key issue is how to scale or disseminate such labor‐intensive highly individualized programs. Each patient has very complex and specific medical, behavioral, and social needs that require creativity and flexibility to adequately address these needs. Without question, patients and/or their caregivers should be members of the care team aiming to optimize their care. Unfortunately, our current healthcare system is not designed to address the complexity and uniqueness of each superutilizer. Nonetheless, summarizing patients history into the EHR and integrating recommendations offers an opportunity to share information as originally hoped by the transition from paper‐based records. It additionally offers an opportunity to learn from use of this information as academic medical centers aim to become learning health systems.[13] Future implementation science research in this area should assess how to scale patient‐centered approaches to care, particularly for those with chronic illness and other vulnerabilities. We must eschew efforts that solely focus on reducing utilization by patients without involving them; after all, they are the focus of healthcare.

We have known for years that the distribution of healthcare expenditures in the United States is skewed, with a small portion of the population consuming a disproportionately high share of resources. In 2010, 1% of the population accounted for 21.4% of the $1.3 trillion spent on healthcare.[1] Growing evidence documents that most of these high‐cost patients are not receiving coordinated care, preventive care, or care in the most appropriate settings.[2] The term superutilizer describes individuals with complex physical, behavioral, and social needs who have frequent emergency department (ED) visits and multiple costly hospital admissions.[3] Not surprisingly, multiple superutilizer programs and new funding opportunities target this population attempting to reduce their healthcare costs while improving their care, as public and private insurers shift to value‐based care.[4]

Beginning in 2006, the Robert Wood Johnson Foundation supported the Camden Coalition[5] with 3 grants to develop a community‐based approach to identify high‐utilizer patients and provide them with coordinated medical and social services.[6] These programs include community‐based teams that focus on the highest utilizers in a specific geographic area and provide intensive outpatient case management. Building on these efforts, the Center for Medicare and Medicaid Innovation (CMMI) awarded 2 Health Care Innovation Awards totaling $17.2 million to target Medicaid superutilizers.[7] Through its State Innovation Models initiative, CMMI also encourages states to pilot superutilizer programs to increase care coordination and support of persons with certain risk factors such as homelessness or mental illness.[8] Additionally, the National Governors Association developed a 1‐year, multistate policy academy to develop state‐level capacity and state action plans that guide how to improve the delivery and financing of care for superutilizers.[9]

With all these ongoing activities in the setting of a paucity of research identifying the most cost‐efficient practices to manage super‐utilizers, we are glad to see the Journal of Hospital Medicine publish an evaluation of a quality‐improvement project targeting superutilizers.[10] Mercer and colleagues at Duke University Hospital show that developing an individualized care plan and integrating it into their electronic health record (EHR) reduced hospital admissions, but not ED visits. Although we applaud the reportedly individualized patient approach and recognize the effort required to refer patients to a more appropriate care setting, we believe the researchers neglected 3 important components for the intervention: (1) patient engagement in developing individualized care plans, (2) care coordination integrated with community collaboration, and (3) feedback on continuum of care relayed back to providers. The managing strategies mentioned in the article seem to have evolved exclusively from the provider's perspective, a common mistake that the Patient‐Centered Outcomes Research Institute emphasizes must be avoided. We are concerned about the lack of clarity regarding the set of management strategies focused on providing high‐quality care while limiting unnecessary admissions reported by them. We fear this strategy was imposed on patients and not developed collaboratively with them. Effective interventions for superutilizers should do more than just guide providers actions, but also connect services to the patient's needs. There should be coordination and continuous improvement of these efforts, which requires engagement of the patient and their community with feedback to the system.

Possibly most important, an individualized approach to superutilizers needs to be patient‐centeredprioritizing patient goals and preferences, selecting interventions and services guided by the needs of the individual, and emphasizing modifiable outcomes that matter to the patient. Such a patient‐centered approach goes beyond the individual patient to incorporate information about social support and family dynamics, highlighting the role of caregivers. Patients and their caregivers must be engaged or activated to ensure adherence to appropriate care and behaviors in any superutilizer programs. Additionally, individualized patient‐centered care plans should be dynamic and bidirectional to accommodate changes in health priorities that may occur over time. Such lack of patient and community engagement may explain why ED‐visit frequency was unchanged in their study.

The approach of having a Complex Care Plan Committee deserves attention as it appropriately included the right people at the academic medical center. However, why is it voluntary? Should not an important, or even essential, committee such as this be supported by the health system? Moreover, although the care plan developed by members of the committee possesses understandable aspects to be considered in a patient's care, why is this not shown to the patient for their input? Instead of being done to the patient, we recommend including patients in this process, believing such patient engagement would improve care further and likely yield sustained changes. We suggest the researchers remember the maxim nothing about me, without me.

Patients who use the most healthcare services typically have complicated social situations that directly impact their ability to improve their health and stay well.[2, 11] Addressing the social determinants of health is not a new concept; however, creating healthy communities as a core responsibility of the healthcare industry is. Contributing to the dizzying state of change in US healthcare are efforts to shift to value‐based purchasing and population health management.[12] This transformation from a fee‐for‐service hospital‐centric industry into one focused on the continuum of care requires outreach into communities where superutilizers live. Ultimately, all healthcare is local, as this is where patients receive the vast majority of their care. Improving quality and reducing costs requires healthcare providers to work together on a collaborative mission that focuses on the needs of patients and community, not just efforts to reduce utilization. Even hospitalists must forge collaborative relationships with skilled nursing facilities and patient‐centered medical homes.

Given the successes of some superutilizer programs,[3] a key issue is how to scale or disseminate such labor‐intensive highly individualized programs. Each patient has very complex and specific medical, behavioral, and social needs that require creativity and flexibility to adequately address these needs. Without question, patients and/or their caregivers should be members of the care team aiming to optimize their care. Unfortunately, our current healthcare system is not designed to address the complexity and uniqueness of each superutilizer. Nonetheless, summarizing patients history into the EHR and integrating recommendations offers an opportunity to share information as originally hoped by the transition from paper‐based records. It additionally offers an opportunity to learn from use of this information as academic medical centers aim to become learning health systems.[13] Future implementation science research in this area should assess how to scale patient‐centered approaches to care, particularly for those with chronic illness and other vulnerabilities. We must eschew efforts that solely focus on reducing utilization by patients without involving them; after all, they are the focus of healthcare.

We have known for years that the distribution of healthcare expenditures in the United States is skewed, with a small portion of the population consuming a disproportionately high share of resources. In 2010, 1% of the population accounted for 21.4% of the $1.3 trillion spent on healthcare.[1] Growing evidence documents that most of these high‐cost patients are not receiving coordinated care, preventive care, or care in the most appropriate settings.[2] The term superutilizer describes individuals with complex physical, behavioral, and social needs who have frequent emergency department (ED) visits and multiple costly hospital admissions.[3] Not surprisingly, multiple superutilizer programs and new funding opportunities target this population attempting to reduce their healthcare costs while improving their care, as public and private insurers shift to value‐based care.[4]

Beginning in 2006, the Robert Wood Johnson Foundation supported the Camden Coalition[5] with 3 grants to develop a community‐based approach to identify high‐utilizer patients and provide them with coordinated medical and social services.[6] These programs include community‐based teams that focus on the highest utilizers in a specific geographic area and provide intensive outpatient case management. Building on these efforts, the Center for Medicare and Medicaid Innovation (CMMI) awarded 2 Health Care Innovation Awards totaling $17.2 million to target Medicaid superutilizers.[7] Through its State Innovation Models initiative, CMMI also encourages states to pilot superutilizer programs to increase care coordination and support of persons with certain risk factors such as homelessness or mental illness.[8] Additionally, the National Governors Association developed a 1‐year, multistate policy academy to develop state‐level capacity and state action plans that guide how to improve the delivery and financing of care for superutilizers.[9]

With all these ongoing activities in the setting of a paucity of research identifying the most cost‐efficient practices to manage super‐utilizers, we are glad to see the Journal of Hospital Medicine publish an evaluation of a quality‐improvement project targeting superutilizers.[10] Mercer and colleagues at Duke University Hospital show that developing an individualized care plan and integrating it into their electronic health record (EHR) reduced hospital admissions, but not ED visits. Although we applaud the reportedly individualized patient approach and recognize the effort required to refer patients to a more appropriate care setting, we believe the researchers neglected 3 important components for the intervention: (1) patient engagement in developing individualized care plans, (2) care coordination integrated with community collaboration, and (3) feedback on continuum of care relayed back to providers. The managing strategies mentioned in the article seem to have evolved exclusively from the provider's perspective, a common mistake that the Patient‐Centered Outcomes Research Institute emphasizes must be avoided. We are concerned about the lack of clarity regarding the set of management strategies focused on providing high‐quality care while limiting unnecessary admissions reported by them. We fear this strategy was imposed on patients and not developed collaboratively with them. Effective interventions for superutilizers should do more than just guide providers actions, but also connect services to the patient's needs. There should be coordination and continuous improvement of these efforts, which requires engagement of the patient and their community with feedback to the system.

Possibly most important, an individualized approach to superutilizers needs to be patient‐centeredprioritizing patient goals and preferences, selecting interventions and services guided by the needs of the individual, and emphasizing modifiable outcomes that matter to the patient. Such a patient‐centered approach goes beyond the individual patient to incorporate information about social support and family dynamics, highlighting the role of caregivers. Patients and their caregivers must be engaged or activated to ensure adherence to appropriate care and behaviors in any superutilizer programs. Additionally, individualized patient‐centered care plans should be dynamic and bidirectional to accommodate changes in health priorities that may occur over time. Such lack of patient and community engagement may explain why ED‐visit frequency was unchanged in their study.

The approach of having a Complex Care Plan Committee deserves attention as it appropriately included the right people at the academic medical center. However, why is it voluntary? Should not an important, or even essential, committee such as this be supported by the health system? Moreover, although the care plan developed by members of the committee possesses understandable aspects to be considered in a patient's care, why is this not shown to the patient for their input? Instead of being done to the patient, we recommend including patients in this process, believing such patient engagement would improve care further and likely yield sustained changes. We suggest the researchers remember the maxim nothing about me, without me.

Patients who use the most healthcare services typically have complicated social situations that directly impact their ability to improve their health and stay well.[2, 11] Addressing the social determinants of health is not a new concept; however, creating healthy communities as a core responsibility of the healthcare industry is. Contributing to the dizzying state of change in US healthcare are efforts to shift to value‐based purchasing and population health management.[12] This transformation from a fee‐for‐service hospital‐centric industry into one focused on the continuum of care requires outreach into communities where superutilizers live. Ultimately, all healthcare is local, as this is where patients receive the vast majority of their care. Improving quality and reducing costs requires healthcare providers to work together on a collaborative mission that focuses on the needs of patients and community, not just efforts to reduce utilization. Even hospitalists must forge collaborative relationships with skilled nursing facilities and patient‐centered medical homes.

Given the successes of some superutilizer programs,[3] a key issue is how to scale or disseminate such labor‐intensive highly individualized programs. Each patient has very complex and specific medical, behavioral, and social needs that require creativity and flexibility to adequately address these needs. Without question, patients and/or their caregivers should be members of the care team aiming to optimize their care. Unfortunately, our current healthcare system is not designed to address the complexity and uniqueness of each superutilizer. Nonetheless, summarizing patients history into the EHR and integrating recommendations offers an opportunity to share information as originally hoped by the transition from paper‐based records. It additionally offers an opportunity to learn from use of this information as academic medical centers aim to become learning health systems.[13] Future implementation science research in this area should assess how to scale patient‐centered approaches to care, particularly for those with chronic illness and other vulnerabilities. We must eschew efforts that solely focus on reducing utilization by patients without involving them; after all, they are the focus of healthcare.

Clinical management of patients with medical and social comorbidities has become increasingly complex.[1, 2, 3, 4] This complexity stems from lack of data for these groups of patients who are often excluded from clinical trials.[5] There are data demonstrating that older patients and patients with multiple comorbidities including diabetes, renal disease, obesity, limited mobility, and poor access to healthcare have worse outcomes for specific conditions compared to otherwise equal counterparts, and that the cost of care for these patients is more expensive.[3, 4, 6] Moreover, traditional risk assessment of disease severity and outcomes are not accurate when applied to medically and socially complex patients.[7, 8]

Treatment decisions regarding antibiotic use add additional complexity. Specifically, physicians antibiotic prescribing decisions can promote the emergence of multidrug resistant pathogens in a hospital or population.[9, 10] Multidrug resistant organisms (MDROs) are particularly problematic, as their prevalence is increasing while the development of new antimicrobial agents is declining.[11] Infections caused by antibiotic‐resistant pathogens are associated with increased morbidity and mortality and healthcare costs.[12, 13, 14] Antibiotic use, although potentially lifesaving, can also result in severe complications such as Clostridium difficile‐associated diarrhea, acute kidney injury, and anaphylaxis, among other adverse events, particularly in older patients with medical comorbidities.[15, 16, 17, 18] Judicious antibiotic use is critical to halt the epidemic of MDROs and to minimize antibiotic‐associated adverse effects.[19, 20, 21]

Evidence‐based guidelines have the potential to assist physicians in choosing the antibiotic that achieves the best clinical outcome for a specific infection or situation.[11] This includes using the narrowest spectrum agent to minimize selection pressure on microorganisms and avoiding unneeded drugs to minimize adverse drug effects.[9, 11] Importantly, guideline adherence regarding antibiotic selection has been shown to be associated with increased clinical success and decreased mortality.[22, 23] Unfortunately, 30% to 50% of antibiotic use in hospitalized patients is inconsistent with national guidelines.[24, 25, 26] Reasons for physicians ordering of tests and treatments inconsistent with guidelines are not fully understood, and potentially include patient and physician factors, and the cultural and social context of the healthcare system.[27]

To optimize the use of antibiotics, it is important to understand how medical complexities (defined as demographic, comorbid, and limited healthcare access characteristics that are associated with suboptimal patient care and outcomes) influence physicians antibiotic prescribing practices.[28] We created 3 clinical vignettes for common diagnoses (dyspnea with initial concern for pneumonia, skin and soft tissue infection, and asymptomatic bacteriuria) among hospitalized patients. We selected these conditions because of their high prevalence, frequent management by hospitalists, generalist physicians, and noninfectious disease specialists, and because well‐documented evidence suggests either no antibiotics or narrower spectrum antibiotics are usually the treatments of choice. Using the Infectious Diseases Society of America (IDSA) guidelines relevant to each clinical vignette,[29, 30, 31] we assessed physicians recommendations for guideline‐appropriate antibiotic management for patients without and with medical complexities using an electronic multiple‐choice survey.

METHODS

RESULTS

Physician Survey

Of the 874 invited physicians, 255 (29%) responded to the survey. Of these, 8/255 (3.1%) responded that they did not spend 2 or more weeks in the inpatient setting and were thus ineligible for the survey. We analyzed data from the remaining 247 physician respondents. Most respondents (217/233, 93%) reported their primary role was direct clinical care (Table 3). Most (185/241, 77%) reported at least half of their clinical work occurred in the hospital. Approximately three‐quarters (183/241, 76%) of the respondents were residents; the remaining respondents were attending physicians (57/241, 24%). Almost half of attending physicians (46%) were internal medicine subspecialists.

Table 3.Characteristics of Physicians Completing the Survey and Association With Recommending Antibiotics Not Consistent With National Guidelines in the Baseline Vignettes by Physician Characteristic

Physician Characteristic	No. (%) Completing the Survey	% of Physicians Not Adhering to Guidelines in Baseline Scenarios	P Value
NOTE: Abbreviations: UCLA, University of California Los Angeles.
Affiliated medical center, n =241
Ronald Reagan UCLA	47 (20%)	37%	0.37
Harbor‐UCLA	106 (44%)	41%
Cedars‐Sinai	86 (35%)	43%
Primary professional activity, n=233
Direct clinical care/teaching	217 (93%)	42%	0.90
Research/administration	16 (7%)	27%
Percent of clinical duties in the hospital, n=241
1%25%	57 (23%)	41%	0.71
51%75%	93 (39%)	42%
76%100%	92 (38%)	41%
Level of training and subspecialization, n=241
Resident/fellow	183 (76%)	43%	0.05
Attending	58 (24%)	34%
Subspecialist	27 (47%)	34%	0.90
Hospitalist	28 (48%)	33%

Physician recommendation for the use of antibiotics inconsistent with IDSA guidelines was prevalent in the baseline vignettes: 42% (303/729) overall, and 49% (120/246) for the dyspnea, 28% (68/242) for the skin infection, and 48% (115/241) for the asymptomatic bacteriuria cases. When the vignettes were modified to include patient complexities, the proportion of physicians recommending antibiotics increased significantly compared to the baseline vignette (63% (459/728), 54% (393/728), 51% (371/728), and 48% (232/487) for medical comorbidities, poor functional status, older age, and limited follow‐up respectively, P<0.001 for all comparisons) (Figure 1). The increase in the proportion of physicians recommending antibiotics inconsistent with guidelines for patients with medical complexities was the same when stratified by case (data not shown).

Figure 1

We found no association between provider characteristics (medical center, degree of attending physician specialization, percentage of clinical time spent practicing hospital‐based medicine, or percentage of time providing direct clinical care) and prescribing antibiotics in the baseline vignettes (Table 3). However, resident physicians (n=183) were more likely than attending physicians (n=57) to have recommended antibiotics in the baseline vignettes (43% vs 34%, P<0.05) (Table 3) and in all 4 vignettes with patient complexities (data not shown).

DISCUSSION

In our survey, almost half of the physician respondents recommended antibiotics that were inconsistent with national guidelines in the baseline vignettes. One explanation for this finding is that physicians may underestimate the risk associated with antibiotic use, such as the emergence of antimicrobial resistant pathogens and drug‐associated adverse effects. Although physicians generally agree that antibiotic resistance is an important problem, many believe that it is not a prominent issue in their practice or through their antibiotic prescribing practices.[43, 44] Others have shown that physicians underestimate the risk and severity of antibiotic‐associated complications such as C difficile.[45, 46] An accurate assessment and heightened awareness of the risks associated with antibiotics is important in clinical decision making, and should potentially be included not only in antibiotic stewardship educational efforts, but in national guideline recommendations as well.

Our survey also demonstrated that the tendency of physician respondents to recommend antibiotics was amplified for patients with medical complexities. This suggests that patient characteristics related to medical and social complexities play an important role in physicians clinical decision making about prescribing antibiotics. Previous investigations have shown that when physicians are deciding whether or not to prescribe antibiotics, they tend to deprioritize guideline recommendations and give greater weight to the risk of disease progression and complications that might occur if antibiotics are withheld.[47, 48, 49] Physicians are believed to prescribe antibiotics for complex patients more often, in part, because complex patients are more likely to suffer bad outcomes if undertreated.[50, 51] Axiomatically, patients with medical complexities are also at higher risk for antibiotic‐associated adverse effects including polypharmacy, drug‐drug interactions, and more severe side effects.[52, 53, 54]

An additional factor contributing to the overuse of antibiotics in our survey could be the lack of clear guideline recommendations for antibiotic management, especially among patients with complexities. We reviewed 20 national guidelines that addressed the medical decision making relevant to the survey's 3 clinical vignettes (see Supporting Information, Appendix 1, in the online version of this article). Fifteen of the guidelines provided recommendations for antibiotic management in the baseline vignettes, though most of the recommendations were not explicit about stopping or de‐escalating antibiotics. Furthermore, when antibiotic recommendations were present, they often lacked supporting data for the recommendation. Guidelines were even less complete for patients with medical complexities. For the asymptomatic bacteriuria vignette, 4 of 6 guidelines provided recommendations for patients with medical comorbidities described in our survey's modified vignettes.[31, 55, 56, 57] None of the guidelines related to the dyspnea or skin infection vignette provided specific antibiotic recommendations for complex patients with medical comorbidities, poor functional status, older age, or limited follow‐up (see Supporting Information, Appendix 1, in the online version of this article). Given these findings, along with evidence that medically complex patients are more likely to receive antibiotics compared to their less complex counterparts, there is a need for subsequent guidelines to more explicitly recommend best antibiotic practices for patients with medical and social complexities.

We also found that resident physicians were significantly more likely to recommend antibiotics inconsistent with guidelines compared to attending physicians. Although the underlying explanations for this finding were not explored in this study, possibilities include a lack of familiarity with guideline recommendations, less comfort in discontinuing antibiotics in the setting of clinical uncertainty, and/or a preference to accept the risks of overtreatment over the risks of undertreatment.

There are limitations to our study. First, although previous investigations have shown a high degree of correlation between actual antibiotic prescribing practices and antibiotic prescribing decisions self‐reported in clinical vignettes, physician responses in our survey may not reflect actual practices.[58] Second, we were not able to directly measure how physicians knowledge and interpretation of national guidelines influenced their antibiotic management decisions in the survey. Third, while our study was multisite including physicians from 3 different centers, not all invited physicians responded. Because of the confidentiality procedures surrounding the email distribution lists of potential participants, we were unable to obtain additional details about the non‐responders. Finally, because the large majority of respondents were trainees (76%), the generalizability of our findings to attending physicians may be limited. Nevertheless, because residents soon become staff physicians, resident‐reported data supplemented by that from attending physicians seems relevant to identifying opportunities for improving medical care.

In conclusion, we found that a large proportion of physicians recommended antibiotics that were not indicated based on IDSA guidelines for 3 vignettes depicting common hospital‐based clinical scenarios. This pattern of physicians recommending antibiotics inconsistent with guidelines was accentuated with significantly higher reported use for patients with medical comorbidities, poor functional status, older age, and limited healthcare access. Although good clinical judgment requires increased monitoring of patients with medical complexities, it is important for clinicians not to conflate the need for increased patient monitoring with the need for increased antibiotic use. Additional studies and corresponding guideline recommendations for frail, complex patients could be instrumental in reducing frequent use of antibiotics and the resultant cost, adverse effects, and emergence of antibiotic resistant pathogens. Educational efforts, particularly among trainees, regarding appropriate antibiotic use for clinical indications among patients without and with medical complexities would also likely contribute to these aims. Treatment guidelines should consider explicitly addressing medically complex patients in the context of management of infectious syndromes.

Clinical management of patients with medical and social comorbidities has become increasingly complex.[1, 2, 3, 4] This complexity stems from lack of data for these groups of patients who are often excluded from clinical trials.[5] There are data demonstrating that older patients and patients with multiple comorbidities including diabetes, renal disease, obesity, limited mobility, and poor access to healthcare have worse outcomes for specific conditions compared to otherwise equal counterparts, and that the cost of care for these patients is more expensive.[3, 4, 6] Moreover, traditional risk assessment of disease severity and outcomes are not accurate when applied to medically and socially complex patients.[7, 8]

Treatment decisions regarding antibiotic use add additional complexity. Specifically, physicians antibiotic prescribing decisions can promote the emergence of multidrug resistant pathogens in a hospital or population.[9, 10] Multidrug resistant organisms (MDROs) are particularly problematic, as their prevalence is increasing while the development of new antimicrobial agents is declining.[11] Infections caused by antibiotic‐resistant pathogens are associated with increased morbidity and mortality and healthcare costs.[12, 13, 14] Antibiotic use, although potentially lifesaving, can also result in severe complications such as Clostridium difficile‐associated diarrhea, acute kidney injury, and anaphylaxis, among other adverse events, particularly in older patients with medical comorbidities.[15, 16, 17, 18] Judicious antibiotic use is critical to halt the epidemic of MDROs and to minimize antibiotic‐associated adverse effects.[19, 20, 21]

Evidence‐based guidelines have the potential to assist physicians in choosing the antibiotic that achieves the best clinical outcome for a specific infection or situation.[11] This includes using the narrowest spectrum agent to minimize selection pressure on microorganisms and avoiding unneeded drugs to minimize adverse drug effects.[9, 11] Importantly, guideline adherence regarding antibiotic selection has been shown to be associated with increased clinical success and decreased mortality.[22, 23] Unfortunately, 30% to 50% of antibiotic use in hospitalized patients is inconsistent with national guidelines.[24, 25, 26] Reasons for physicians ordering of tests and treatments inconsistent with guidelines are not fully understood, and potentially include patient and physician factors, and the cultural and social context of the healthcare system.[27]

To optimize the use of antibiotics, it is important to understand how medical complexities (defined as demographic, comorbid, and limited healthcare access characteristics that are associated with suboptimal patient care and outcomes) influence physicians antibiotic prescribing practices.[28] We created 3 clinical vignettes for common diagnoses (dyspnea with initial concern for pneumonia, skin and soft tissue infection, and asymptomatic bacteriuria) among hospitalized patients. We selected these conditions because of their high prevalence, frequent management by hospitalists, generalist physicians, and noninfectious disease specialists, and because well‐documented evidence suggests either no antibiotics or narrower spectrum antibiotics are usually the treatments of choice. Using the Infectious Diseases Society of America (IDSA) guidelines relevant to each clinical vignette,[29, 30, 31] we assessed physicians recommendations for guideline‐appropriate antibiotic management for patients without and with medical complexities using an electronic multiple‐choice survey.

METHODS

RESULTS

Physician Survey

Of the 874 invited physicians, 255 (29%) responded to the survey. Of these, 8/255 (3.1%) responded that they did not spend 2 or more weeks in the inpatient setting and were thus ineligible for the survey. We analyzed data from the remaining 247 physician respondents. Most respondents (217/233, 93%) reported their primary role was direct clinical care (Table 3). Most (185/241, 77%) reported at least half of their clinical work occurred in the hospital. Approximately three‐quarters (183/241, 76%) of the respondents were residents; the remaining respondents were attending physicians (57/241, 24%). Almost half of attending physicians (46%) were internal medicine subspecialists.

Table 3.Characteristics of Physicians Completing the Survey and Association With Recommending Antibiotics Not Consistent With National Guidelines in the Baseline Vignettes by Physician Characteristic

Physician Characteristic	No. (%) Completing the Survey	% of Physicians Not Adhering to Guidelines in Baseline Scenarios	P Value
NOTE: Abbreviations: UCLA, University of California Los Angeles.
Affiliated medical center, n =241
Ronald Reagan UCLA	47 (20%)	37%	0.37
Harbor‐UCLA	106 (44%)	41%
Cedars‐Sinai	86 (35%)	43%
Primary professional activity, n=233
Direct clinical care/teaching	217 (93%)	42%	0.90
Research/administration	16 (7%)	27%
Percent of clinical duties in the hospital, n=241
1%25%	57 (23%)	41%	0.71
51%75%	93 (39%)	42%
76%100%	92 (38%)	41%
Level of training and subspecialization, n=241
Resident/fellow	183 (76%)	43%	0.05
Attending	58 (24%)	34%
Subspecialist	27 (47%)	34%	0.90
Hospitalist	28 (48%)	33%

Physician recommendation for the use of antibiotics inconsistent with IDSA guidelines was prevalent in the baseline vignettes: 42% (303/729) overall, and 49% (120/246) for the dyspnea, 28% (68/242) for the skin infection, and 48% (115/241) for the asymptomatic bacteriuria cases. When the vignettes were modified to include patient complexities, the proportion of physicians recommending antibiotics increased significantly compared to the baseline vignette (63% (459/728), 54% (393/728), 51% (371/728), and 48% (232/487) for medical comorbidities, poor functional status, older age, and limited follow‐up respectively, P<0.001 for all comparisons) (Figure 1). The increase in the proportion of physicians recommending antibiotics inconsistent with guidelines for patients with medical complexities was the same when stratified by case (data not shown).

Figure 1

We found no association between provider characteristics (medical center, degree of attending physician specialization, percentage of clinical time spent practicing hospital‐based medicine, or percentage of time providing direct clinical care) and prescribing antibiotics in the baseline vignettes (Table 3). However, resident physicians (n=183) were more likely than attending physicians (n=57) to have recommended antibiotics in the baseline vignettes (43% vs 34%, P<0.05) (Table 3) and in all 4 vignettes with patient complexities (data not shown).

DISCUSSION

In our survey, almost half of the physician respondents recommended antibiotics that were inconsistent with national guidelines in the baseline vignettes. One explanation for this finding is that physicians may underestimate the risk associated with antibiotic use, such as the emergence of antimicrobial resistant pathogens and drug‐associated adverse effects. Although physicians generally agree that antibiotic resistance is an important problem, many believe that it is not a prominent issue in their practice or through their antibiotic prescribing practices.[43, 44] Others have shown that physicians underestimate the risk and severity of antibiotic‐associated complications such as C difficile.[45, 46] An accurate assessment and heightened awareness of the risks associated with antibiotics is important in clinical decision making, and should potentially be included not only in antibiotic stewardship educational efforts, but in national guideline recommendations as well.

Our survey also demonstrated that the tendency of physician respondents to recommend antibiotics was amplified for patients with medical complexities. This suggests that patient characteristics related to medical and social complexities play an important role in physicians clinical decision making about prescribing antibiotics. Previous investigations have shown that when physicians are deciding whether or not to prescribe antibiotics, they tend to deprioritize guideline recommendations and give greater weight to the risk of disease progression and complications that might occur if antibiotics are withheld.[47, 48, 49] Physicians are believed to prescribe antibiotics for complex patients more often, in part, because complex patients are more likely to suffer bad outcomes if undertreated.[50, 51] Axiomatically, patients with medical complexities are also at higher risk for antibiotic‐associated adverse effects including polypharmacy, drug‐drug interactions, and more severe side effects.[52, 53, 54]

An additional factor contributing to the overuse of antibiotics in our survey could be the lack of clear guideline recommendations for antibiotic management, especially among patients with complexities. We reviewed 20 national guidelines that addressed the medical decision making relevant to the survey's 3 clinical vignettes (see Supporting Information, Appendix 1, in the online version of this article). Fifteen of the guidelines provided recommendations for antibiotic management in the baseline vignettes, though most of the recommendations were not explicit about stopping or de‐escalating antibiotics. Furthermore, when antibiotic recommendations were present, they often lacked supporting data for the recommendation. Guidelines were even less complete for patients with medical complexities. For the asymptomatic bacteriuria vignette, 4 of 6 guidelines provided recommendations for patients with medical comorbidities described in our survey's modified vignettes.[31, 55, 56, 57] None of the guidelines related to the dyspnea or skin infection vignette provided specific antibiotic recommendations for complex patients with medical comorbidities, poor functional status, older age, or limited follow‐up (see Supporting Information, Appendix 1, in the online version of this article). Given these findings, along with evidence that medically complex patients are more likely to receive antibiotics compared to their less complex counterparts, there is a need for subsequent guidelines to more explicitly recommend best antibiotic practices for patients with medical and social complexities.

We also found that resident physicians were significantly more likely to recommend antibiotics inconsistent with guidelines compared to attending physicians. Although the underlying explanations for this finding were not explored in this study, possibilities include a lack of familiarity with guideline recommendations, less comfort in discontinuing antibiotics in the setting of clinical uncertainty, and/or a preference to accept the risks of overtreatment over the risks of undertreatment.

There are limitations to our study. First, although previous investigations have shown a high degree of correlation between actual antibiotic prescribing practices and antibiotic prescribing decisions self‐reported in clinical vignettes, physician responses in our survey may not reflect actual practices.[58] Second, we were not able to directly measure how physicians knowledge and interpretation of national guidelines influenced their antibiotic management decisions in the survey. Third, while our study was multisite including physicians from 3 different centers, not all invited physicians responded. Because of the confidentiality procedures surrounding the email distribution lists of potential participants, we were unable to obtain additional details about the non‐responders. Finally, because the large majority of respondents were trainees (76%), the generalizability of our findings to attending physicians may be limited. Nevertheless, because residents soon become staff physicians, resident‐reported data supplemented by that from attending physicians seems relevant to identifying opportunities for improving medical care.

In conclusion, we found that a large proportion of physicians recommended antibiotics that were not indicated based on IDSA guidelines for 3 vignettes depicting common hospital‐based clinical scenarios. This pattern of physicians recommending antibiotics inconsistent with guidelines was accentuated with significantly higher reported use for patients with medical comorbidities, poor functional status, older age, and limited healthcare access. Although good clinical judgment requires increased monitoring of patients with medical complexities, it is important for clinicians not to conflate the need for increased patient monitoring with the need for increased antibiotic use. Additional studies and corresponding guideline recommendations for frail, complex patients could be instrumental in reducing frequent use of antibiotics and the resultant cost, adverse effects, and emergence of antibiotic resistant pathogens. Educational efforts, particularly among trainees, regarding appropriate antibiotic use for clinical indications among patients without and with medical complexities would also likely contribute to these aims. Treatment guidelines should consider explicitly addressing medically complex patients in the context of management of infectious syndromes.

Clinical management of patients with medical and social comorbidities has become increasingly complex.[1, 2, 3, 4] This complexity stems from lack of data for these groups of patients who are often excluded from clinical trials.[5] There are data demonstrating that older patients and patients with multiple comorbidities including diabetes, renal disease, obesity, limited mobility, and poor access to healthcare have worse outcomes for specific conditions compared to otherwise equal counterparts, and that the cost of care for these patients is more expensive.[3, 4, 6] Moreover, traditional risk assessment of disease severity and outcomes are not accurate when applied to medically and socially complex patients.[7, 8]

Treatment decisions regarding antibiotic use add additional complexity. Specifically, physicians antibiotic prescribing decisions can promote the emergence of multidrug resistant pathogens in a hospital or population.[9, 10] Multidrug resistant organisms (MDROs) are particularly problematic, as their prevalence is increasing while the development of new antimicrobial agents is declining.[11] Infections caused by antibiotic‐resistant pathogens are associated with increased morbidity and mortality and healthcare costs.[12, 13, 14] Antibiotic use, although potentially lifesaving, can also result in severe complications such as Clostridium difficile‐associated diarrhea, acute kidney injury, and anaphylaxis, among other adverse events, particularly in older patients with medical comorbidities.[15, 16, 17, 18] Judicious antibiotic use is critical to halt the epidemic of MDROs and to minimize antibiotic‐associated adverse effects.[19, 20, 21]

Evidence‐based guidelines have the potential to assist physicians in choosing the antibiotic that achieves the best clinical outcome for a specific infection or situation.[11] This includes using the narrowest spectrum agent to minimize selection pressure on microorganisms and avoiding unneeded drugs to minimize adverse drug effects.[9, 11] Importantly, guideline adherence regarding antibiotic selection has been shown to be associated with increased clinical success and decreased mortality.[22, 23] Unfortunately, 30% to 50% of antibiotic use in hospitalized patients is inconsistent with national guidelines.[24, 25, 26] Reasons for physicians ordering of tests and treatments inconsistent with guidelines are not fully understood, and potentially include patient and physician factors, and the cultural and social context of the healthcare system.[27]

To optimize the use of antibiotics, it is important to understand how medical complexities (defined as demographic, comorbid, and limited healthcare access characteristics that are associated with suboptimal patient care and outcomes) influence physicians antibiotic prescribing practices.[28] We created 3 clinical vignettes for common diagnoses (dyspnea with initial concern for pneumonia, skin and soft tissue infection, and asymptomatic bacteriuria) among hospitalized patients. We selected these conditions because of their high prevalence, frequent management by hospitalists, generalist physicians, and noninfectious disease specialists, and because well‐documented evidence suggests either no antibiotics or narrower spectrum antibiotics are usually the treatments of choice. Using the Infectious Diseases Society of America (IDSA) guidelines relevant to each clinical vignette,[29, 30, 31] we assessed physicians recommendations for guideline‐appropriate antibiotic management for patients without and with medical complexities using an electronic multiple‐choice survey.

METHODS

RESULTS

Physician Survey

Of the 874 invited physicians, 255 (29%) responded to the survey. Of these, 8/255 (3.1%) responded that they did not spend 2 or more weeks in the inpatient setting and were thus ineligible for the survey. We analyzed data from the remaining 247 physician respondents. Most respondents (217/233, 93%) reported their primary role was direct clinical care (Table 3). Most (185/241, 77%) reported at least half of their clinical work occurred in the hospital. Approximately three‐quarters (183/241, 76%) of the respondents were residents; the remaining respondents were attending physicians (57/241, 24%). Almost half of attending physicians (46%) were internal medicine subspecialists.

Table 3.Characteristics of Physicians Completing the Survey and Association With Recommending Antibiotics Not Consistent With National Guidelines in the Baseline Vignettes by Physician Characteristic

Physician Characteristic	No. (%) Completing the Survey	% of Physicians Not Adhering to Guidelines in Baseline Scenarios	P Value
NOTE: Abbreviations: UCLA, University of California Los Angeles.
Affiliated medical center, n =241
Ronald Reagan UCLA	47 (20%)	37%	0.37
Harbor‐UCLA	106 (44%)	41%
Cedars‐Sinai	86 (35%)	43%
Primary professional activity, n=233
Direct clinical care/teaching	217 (93%)	42%	0.90
Research/administration	16 (7%)	27%
Percent of clinical duties in the hospital, n=241
1%25%	57 (23%)	41%	0.71
51%75%	93 (39%)	42%
76%100%	92 (38%)	41%
Level of training and subspecialization, n=241
Resident/fellow	183 (76%)	43%	0.05
Attending	58 (24%)	34%
Subspecialist	27 (47%)	34%	0.90
Hospitalist	28 (48%)	33%

Physician recommendation for the use of antibiotics inconsistent with IDSA guidelines was prevalent in the baseline vignettes: 42% (303/729) overall, and 49% (120/246) for the dyspnea, 28% (68/242) for the skin infection, and 48% (115/241) for the asymptomatic bacteriuria cases. When the vignettes were modified to include patient complexities, the proportion of physicians recommending antibiotics increased significantly compared to the baseline vignette (63% (459/728), 54% (393/728), 51% (371/728), and 48% (232/487) for medical comorbidities, poor functional status, older age, and limited follow‐up respectively, P<0.001 for all comparisons) (Figure 1). The increase in the proportion of physicians recommending antibiotics inconsistent with guidelines for patients with medical complexities was the same when stratified by case (data not shown).

Figure 1

We found no association between provider characteristics (medical center, degree of attending physician specialization, percentage of clinical time spent practicing hospital‐based medicine, or percentage of time providing direct clinical care) and prescribing antibiotics in the baseline vignettes (Table 3). However, resident physicians (n=183) were more likely than attending physicians (n=57) to have recommended antibiotics in the baseline vignettes (43% vs 34%, P<0.05) (Table 3) and in all 4 vignettes with patient complexities (data not shown).

DISCUSSION

In our survey, almost half of the physician respondents recommended antibiotics that were inconsistent with national guidelines in the baseline vignettes. One explanation for this finding is that physicians may underestimate the risk associated with antibiotic use, such as the emergence of antimicrobial resistant pathogens and drug‐associated adverse effects. Although physicians generally agree that antibiotic resistance is an important problem, many believe that it is not a prominent issue in their practice or through their antibiotic prescribing practices.[43, 44] Others have shown that physicians underestimate the risk and severity of antibiotic‐associated complications such as C difficile.[45, 46] An accurate assessment and heightened awareness of the risks associated with antibiotics is important in clinical decision making, and should potentially be included not only in antibiotic stewardship educational efforts, but in national guideline recommendations as well.

Our survey also demonstrated that the tendency of physician respondents to recommend antibiotics was amplified for patients with medical complexities. This suggests that patient characteristics related to medical and social complexities play an important role in physicians clinical decision making about prescribing antibiotics. Previous investigations have shown that when physicians are deciding whether or not to prescribe antibiotics, they tend to deprioritize guideline recommendations and give greater weight to the risk of disease progression and complications that might occur if antibiotics are withheld.[47, 48, 49] Physicians are believed to prescribe antibiotics for complex patients more often, in part, because complex patients are more likely to suffer bad outcomes if undertreated.[50, 51] Axiomatically, patients with medical complexities are also at higher risk for antibiotic‐associated adverse effects including polypharmacy, drug‐drug interactions, and more severe side effects.[52, 53, 54]

An additional factor contributing to the overuse of antibiotics in our survey could be the lack of clear guideline recommendations for antibiotic management, especially among patients with complexities. We reviewed 20 national guidelines that addressed the medical decision making relevant to the survey's 3 clinical vignettes (see Supporting Information, Appendix 1, in the online version of this article). Fifteen of the guidelines provided recommendations for antibiotic management in the baseline vignettes, though most of the recommendations were not explicit about stopping or de‐escalating antibiotics. Furthermore, when antibiotic recommendations were present, they often lacked supporting data for the recommendation. Guidelines were even less complete for patients with medical complexities. For the asymptomatic bacteriuria vignette, 4 of 6 guidelines provided recommendations for patients with medical comorbidities described in our survey's modified vignettes.[31, 55, 56, 57] None of the guidelines related to the dyspnea or skin infection vignette provided specific antibiotic recommendations for complex patients with medical comorbidities, poor functional status, older age, or limited follow‐up (see Supporting Information, Appendix 1, in the online version of this article). Given these findings, along with evidence that medically complex patients are more likely to receive antibiotics compared to their less complex counterparts, there is a need for subsequent guidelines to more explicitly recommend best antibiotic practices for patients with medical and social complexities.

We also found that resident physicians were significantly more likely to recommend antibiotics inconsistent with guidelines compared to attending physicians. Although the underlying explanations for this finding were not explored in this study, possibilities include a lack of familiarity with guideline recommendations, less comfort in discontinuing antibiotics in the setting of clinical uncertainty, and/or a preference to accept the risks of overtreatment over the risks of undertreatment.

There are limitations to our study. First, although previous investigations have shown a high degree of correlation between actual antibiotic prescribing practices and antibiotic prescribing decisions self‐reported in clinical vignettes, physician responses in our survey may not reflect actual practices.[58] Second, we were not able to directly measure how physicians knowledge and interpretation of national guidelines influenced their antibiotic management decisions in the survey. Third, while our study was multisite including physicians from 3 different centers, not all invited physicians responded. Because of the confidentiality procedures surrounding the email distribution lists of potential participants, we were unable to obtain additional details about the non‐responders. Finally, because the large majority of respondents were trainees (76%), the generalizability of our findings to attending physicians may be limited. Nevertheless, because residents soon become staff physicians, resident‐reported data supplemented by that from attending physicians seems relevant to identifying opportunities for improving medical care.

In conclusion, we found that a large proportion of physicians recommended antibiotics that were not indicated based on IDSA guidelines for 3 vignettes depicting common hospital‐based clinical scenarios. This pattern of physicians recommending antibiotics inconsistent with guidelines was accentuated with significantly higher reported use for patients with medical comorbidities, poor functional status, older age, and limited healthcare access. Although good clinical judgment requires increased monitoring of patients with medical complexities, it is important for clinicians not to conflate the need for increased patient monitoring with the need for increased antibiotic use. Additional studies and corresponding guideline recommendations for frail, complex patients could be instrumental in reducing frequent use of antibiotics and the resultant cost, adverse effects, and emergence of antibiotic resistant pathogens. Educational efforts, particularly among trainees, regarding appropriate antibiotic use for clinical indications among patients without and with medical complexities would also likely contribute to these aims. Treatment guidelines should consider explicitly addressing medically complex patients in the context of management of infectious syndromes.

Despite 15 years of national and local investment in improving the safety of hospital care, patient safety remains a leading problem in both adult and pediatric hospitals. A 2010 study found that 180,000 Medicare beneficiaries likely die each year due to harm suffered as a result of medical care,[1] a death toll surpassed only by deaths due to cardiovascular disease and cancer. Even though initial efforts in the field have shown great promise for stemming the tide of healthcare‐associated infections,[2] surgical errors,[3] handoff failures,[4] and errors in the care of adults hospitalized for myocardial infarction and congestive heart failure,[5] much work remains to be done.[6] The root causes of many adverse events are poorly understood and unaddressed. Resultant tragedies remain all too common.

In the current issue of the Journal of Hospital Medicine, Bonafide and colleagues report the results of an innovative observational pilot study designed to assess the role of an inadequately addressed root cause of serious errors: alarm fatigue.[7] Alarm fatigue is the phenomenon of desensitization to alarms, particularly in the context of excessive false alarms. In a videotaped observational assessment of nurse response times to 5070 alarms on a pediatric ward and intensive care unit (ICU), the authors found that nurses responded significantly more slowly as the number of nonactionable alarms in the preceding 2 hours increased. Although a substantial majority of these alarms were technically valid (ie, representing true deviations of vital signs outside of the normal range rather than sensor or equipment problems), the vast majority required no action to be takenapproximately 7 out of 8 in the ICU and an astonishing 99 out of 100 on the ward.

As any hospitalist, intensivist, or nurse knows well, alarms are rampant throughout hospitals. It is impossible to walk down any hallway on a busy hospital wardnever mind an ICUwithout seeing a flashing light or 2 above a doorway, and hearing the incessant beeping of oxygen saturation and cardiovascular/respiratory monitors, a thousand bits of technology forever crying wolf. The problem, of course, is that sometimes there really is a wolf, but it is hard to take the risk seriously when the false alarms happen not just twice before a true threat materializes, as in Aesop's fable, but 7 times in the ICU, or worse, 99 times in the setting where most hospitalists practice. Moreover, even when the threat is real, in most cases it is caught in time one way or another, and no lasting harm results.

So why not simply shut off the unremitting noise? In 1987, outside of Baltimore, Amtrak experienced what at the time was the deadliest rail crash in its history after 1 of its passenger trains collided with a Conrail freight train. A major root cause of the crash was that the crew on the freight train had placed duct tape over an annoying automated signal alarm.[8, 9] Tragically, on this particular day, the suppressed alarm was all too relevant. Identifying the real alarm, however, can be nearly impossible when it sounds the same as 100 irritating sounds constantly emanating from the environment. It is the challenge of identifying the needle in the haystack, after you have developed an allergy to the hay.

What then to do? More research like that conducted by Bonafide and colleagues is needed to better understand how healthcare providers respond to the onslaught of alarms they encounter, and to inform refinement of these systems. Understanding how alarm fatigue plays out in the context of different clinical settings, with different workloads, varying levels of distraction, and different rates of true and false‐positive alarms will be critical. Furthermore, understanding how individuals' physiologic fatigue, circadian misalignment, mood, stress, and cognitive state may play into alarm response is likewise essential, if we are to design appropriate alarm systems that function effectively in the busy 24‐hour environment of healthcare. Ongoing work suggests that smart alarms, using algorithms that integrate data from multiple vital sign readings over time, may reduce the frequency of false alarms and better identify clinically significant events.[10] Replacing existing range‐limit monitors with these types of smart alarms has the potential to greatly improve both the sensitivity and specificity of hospital alarms, but further work in this area is needed.

Ultimately, if we can better separate out the signal, we will be better poised to respond to the true emergencies that arise that are currently obscured by the ever‐present noise. Better trust in the alarm systems we have would help all of us focus our energies on the problems that matter most. Doing so, we could better care for our patients, and better identify the system failures that cause them harm in our hospitals.

Disclosures: Dr. Landrigan is supported in part by the Children's Hospital Association for his work as an Executive Council Member of the Pediatric Research in Inpatient Settings network. Dr. Landrigan serves as a consultant to Virgin Pulse regarding sleep, safety, and health. In addition, Dr. Landrigan has received monetary awards, honoraria, and travel reimbursement from multiple academic and professional organizations for delivering lectures on sleep deprivation, physician performance, handoffs, and patient safety, and has served as an expert witness in cases regarding patient safety.

Despite 15 years of national and local investment in improving the safety of hospital care, patient safety remains a leading problem in both adult and pediatric hospitals. A 2010 study found that 180,000 Medicare beneficiaries likely die each year due to harm suffered as a result of medical care,[1] a death toll surpassed only by deaths due to cardiovascular disease and cancer. Even though initial efforts in the field have shown great promise for stemming the tide of healthcare‐associated infections,[2] surgical errors,[3] handoff failures,[4] and errors in the care of adults hospitalized for myocardial infarction and congestive heart failure,[5] much work remains to be done.[6] The root causes of many adverse events are poorly understood and unaddressed. Resultant tragedies remain all too common.

In the current issue of the Journal of Hospital Medicine, Bonafide and colleagues report the results of an innovative observational pilot study designed to assess the role of an inadequately addressed root cause of serious errors: alarm fatigue.[7] Alarm fatigue is the phenomenon of desensitization to alarms, particularly in the context of excessive false alarms. In a videotaped observational assessment of nurse response times to 5070 alarms on a pediatric ward and intensive care unit (ICU), the authors found that nurses responded significantly more slowly as the number of nonactionable alarms in the preceding 2 hours increased. Although a substantial majority of these alarms were technically valid (ie, representing true deviations of vital signs outside of the normal range rather than sensor or equipment problems), the vast majority required no action to be takenapproximately 7 out of 8 in the ICU and an astonishing 99 out of 100 on the ward.

As any hospitalist, intensivist, or nurse knows well, alarms are rampant throughout hospitals. It is impossible to walk down any hallway on a busy hospital wardnever mind an ICUwithout seeing a flashing light or 2 above a doorway, and hearing the incessant beeping of oxygen saturation and cardiovascular/respiratory monitors, a thousand bits of technology forever crying wolf. The problem, of course, is that sometimes there really is a wolf, but it is hard to take the risk seriously when the false alarms happen not just twice before a true threat materializes, as in Aesop's fable, but 7 times in the ICU, or worse, 99 times in the setting where most hospitalists practice. Moreover, even when the threat is real, in most cases it is caught in time one way or another, and no lasting harm results.

So why not simply shut off the unremitting noise? In 1987, outside of Baltimore, Amtrak experienced what at the time was the deadliest rail crash in its history after 1 of its passenger trains collided with a Conrail freight train. A major root cause of the crash was that the crew on the freight train had placed duct tape over an annoying automated signal alarm.[8, 9] Tragically, on this particular day, the suppressed alarm was all too relevant. Identifying the real alarm, however, can be nearly impossible when it sounds the same as 100 irritating sounds constantly emanating from the environment. It is the challenge of identifying the needle in the haystack, after you have developed an allergy to the hay.

What then to do? More research like that conducted by Bonafide and colleagues is needed to better understand how healthcare providers respond to the onslaught of alarms they encounter, and to inform refinement of these systems. Understanding how alarm fatigue plays out in the context of different clinical settings, with different workloads, varying levels of distraction, and different rates of true and false‐positive alarms will be critical. Furthermore, understanding how individuals' physiologic fatigue, circadian misalignment, mood, stress, and cognitive state may play into alarm response is likewise essential, if we are to design appropriate alarm systems that function effectively in the busy 24‐hour environment of healthcare. Ongoing work suggests that smart alarms, using algorithms that integrate data from multiple vital sign readings over time, may reduce the frequency of false alarms and better identify clinically significant events.[10] Replacing existing range‐limit monitors with these types of smart alarms has the potential to greatly improve both the sensitivity and specificity of hospital alarms, but further work in this area is needed.

Ultimately, if we can better separate out the signal, we will be better poised to respond to the true emergencies that arise that are currently obscured by the ever‐present noise. Better trust in the alarm systems we have would help all of us focus our energies on the problems that matter most. Doing so, we could better care for our patients, and better identify the system failures that cause them harm in our hospitals.

Disclosures: Dr. Landrigan is supported in part by the Children's Hospital Association for his work as an Executive Council Member of the Pediatric Research in Inpatient Settings network. Dr. Landrigan serves as a consultant to Virgin Pulse regarding sleep, safety, and health. In addition, Dr. Landrigan has received monetary awards, honoraria, and travel reimbursement from multiple academic and professional organizations for delivering lectures on sleep deprivation, physician performance, handoffs, and patient safety, and has served as an expert witness in cases regarding patient safety.

Despite 15 years of national and local investment in improving the safety of hospital care, patient safety remains a leading problem in both adult and pediatric hospitals. A 2010 study found that 180,000 Medicare beneficiaries likely die each year due to harm suffered as a result of medical care,[1] a death toll surpassed only by deaths due to cardiovascular disease and cancer. Even though initial efforts in the field have shown great promise for stemming the tide of healthcare‐associated infections,[2] surgical errors,[3] handoff failures,[4] and errors in the care of adults hospitalized for myocardial infarction and congestive heart failure,[5] much work remains to be done.[6] The root causes of many adverse events are poorly understood and unaddressed. Resultant tragedies remain all too common.

In the current issue of the Journal of Hospital Medicine, Bonafide and colleagues report the results of an innovative observational pilot study designed to assess the role of an inadequately addressed root cause of serious errors: alarm fatigue.[7] Alarm fatigue is the phenomenon of desensitization to alarms, particularly in the context of excessive false alarms. In a videotaped observational assessment of nurse response times to 5070 alarms on a pediatric ward and intensive care unit (ICU), the authors found that nurses responded significantly more slowly as the number of nonactionable alarms in the preceding 2 hours increased. Although a substantial majority of these alarms were technically valid (ie, representing true deviations of vital signs outside of the normal range rather than sensor or equipment problems), the vast majority required no action to be takenapproximately 7 out of 8 in the ICU and an astonishing 99 out of 100 on the ward.

As any hospitalist, intensivist, or nurse knows well, alarms are rampant throughout hospitals. It is impossible to walk down any hallway on a busy hospital wardnever mind an ICUwithout seeing a flashing light or 2 above a doorway, and hearing the incessant beeping of oxygen saturation and cardiovascular/respiratory monitors, a thousand bits of technology forever crying wolf. The problem, of course, is that sometimes there really is a wolf, but it is hard to take the risk seriously when the false alarms happen not just twice before a true threat materializes, as in Aesop's fable, but 7 times in the ICU, or worse, 99 times in the setting where most hospitalists practice. Moreover, even when the threat is real, in most cases it is caught in time one way or another, and no lasting harm results.

So why not simply shut off the unremitting noise? In 1987, outside of Baltimore, Amtrak experienced what at the time was the deadliest rail crash in its history after 1 of its passenger trains collided with a Conrail freight train. A major root cause of the crash was that the crew on the freight train had placed duct tape over an annoying automated signal alarm.[8, 9] Tragically, on this particular day, the suppressed alarm was all too relevant. Identifying the real alarm, however, can be nearly impossible when it sounds the same as 100 irritating sounds constantly emanating from the environment. It is the challenge of identifying the needle in the haystack, after you have developed an allergy to the hay.

What then to do? More research like that conducted by Bonafide and colleagues is needed to better understand how healthcare providers respond to the onslaught of alarms they encounter, and to inform refinement of these systems. Understanding how alarm fatigue plays out in the context of different clinical settings, with different workloads, varying levels of distraction, and different rates of true and false‐positive alarms will be critical. Furthermore, understanding how individuals' physiologic fatigue, circadian misalignment, mood, stress, and cognitive state may play into alarm response is likewise essential, if we are to design appropriate alarm systems that function effectively in the busy 24‐hour environment of healthcare. Ongoing work suggests that smart alarms, using algorithms that integrate data from multiple vital sign readings over time, may reduce the frequency of false alarms and better identify clinically significant events.[10] Replacing existing range‐limit monitors with these types of smart alarms has the potential to greatly improve both the sensitivity and specificity of hospital alarms, but further work in this area is needed.

Ultimately, if we can better separate out the signal, we will be better poised to respond to the true emergencies that arise that are currently obscured by the ever‐present noise. Better trust in the alarm systems we have would help all of us focus our energies on the problems that matter most. Doing so, we could better care for our patients, and better identify the system failures that cause them harm in our hospitals.

Disclosures: Dr. Landrigan is supported in part by the Children's Hospital Association for his work as an Executive Council Member of the Pediatric Research in Inpatient Settings network. Dr. Landrigan serves as a consultant to Virgin Pulse regarding sleep, safety, and health. In addition, Dr. Landrigan has received monetary awards, honoraria, and travel reimbursement from multiple academic and professional organizations for delivering lectures on sleep deprivation, physician performance, handoffs, and patient safety, and has served as an expert witness in cases regarding patient safety.

Hospital physiologic monitors can alert clinicians to early signs of physiologic deterioration, and thus have great potential to save lives. However, monitors generate frequent alarms,[1, 2, 3, 4, 5, 6, 7, 8] and most are not relevant to the patient's safety (over 90% of pediatric intensive care unit (PICU)[1, 2] and over 70% of adult intensive care alarms).[5, 6] In psychology experiments, humans rapidly learn to ignore or respond more slowly to alarms when exposed to high false‐alarm rates, exhibiting alarm fatigue.[9, 10] In 2013, The Joint Commission named alarm fatigue the most common contributing factor to alarm‐related sentinel events in hospitals.[11, 12]

Although alarm fatigue has been implicated as a major threat to patient safety, little empirical data support its existence in hospitals. In this study, we aimed to determine if there was an association between nurses' recent exposure to nonactionable physiologic monitor alarms and their response time to future alarms for the same patients. This exploratory work was designed to inform future research in this area, acknowledging that the sample size would be too small for multivariable modeling.

METHODS

Study Definitions

The alarm classification scheme is shown in Figure 1. Note that, for clarity, we have intentionally avoided using the terms true and false alarms because their interpretations vary across studies and can be misleading.

Figure 1

Potentially Critical Alarm

A potentially critical alarm is any alarm for a clinical condition for which a timely response is important to determine if the alarm requires intervention to save the patient's life. This is based on the alarm type alone, including alarms for life‐threatening arrhythmias such as asystole and ventricular tachycardia, as well as alarms for vital signs outside the set limits. Supporting Table 1 in the online version of this article lists the breakdown of alarm types that we defined a priori as potentially and not potentially critical.

Table 1.Characteristics of the 2,445 Alarms for Clinical Conditions

	PICU				Ward
Alarm type	No.	% of Total	% Valid	% Actionable	No.	% of Total	% Valid	% Actionable
NOTE: Abbreviations: N/A, not applicable; PICU, pediatric intensive care unit.
Oxygen saturation	197	19.4	82.7	38.6	590	41.2	24.4	1.9
Heart rate	194	19.1	95.4	1.0	266	18.6	87.2	0.0
Respiratory rate	229	22.6	80.8	13.5	316	22.1	48.1	1.0
Blood pressure	259	25.5	83.8	5.8	11	0.8	72.7	0.0
Critical arrhythmia	1	0.1	0.0	0.0	4	0.3	0.0	0.0
Noncritical arrhythmia	71	7.0	2.8	0.0	244	17.1	8.6	0.0
Central venous pressure	49	4.8	0.0	0.0	0	0.0	N/A	N/A
Exhaled carbon dioxide	14	1.4	92.9	50.0	0	0.0	N/A	N/A
Total	1014	100.0	75.6	12.9	1,431	100.0	38.9	1.0

Valid Alarm

A valid alarm is any alarm that correctly identifies the physiologic status of the patient. Validity was based on waveform quality, lead signal strength indicators, and artifact conditions, referencing each monitor's operator's manual.

Actionable Alarm

An actionable alarm is any valid alarm for a clinical condition that either: (1) leads to a clinical intervention; (2) leads to a consultation with another clinician at the bedside (and thus visible on camera); or (3) is a situation that should have led to intervention or consultation, but the alarm was unwitnessed or misinterpreted by the staff at the bedside.

Nonactionable Alarm

An unactionable alarm is any alarm that does not meet the actionable definition above, including invalid alarms such as those caused by motion artifact, equipment/technical alarms, and alarms that are valid but nonactionable (nuisance alarms).[13]

Response Time

The response time is the time elapsed from when the alarm fired at the bedside to when the nurse entered the room or peered through a window or door, measured in seconds.

RESULTS

We performed 40 sessions among 40 patients and 36 nurses over 210 hours. We performed 20 sessions in children with heart and/or lung failure in the PICU and 20 sessions in children on a general ward. Sessions took place on weekdays between 9:00 _am and 6:00 _pm. There were 3 occasions when we filmed 2 patients cared for by the same nurse at the same time.

Nurses were mostly female (94.4%) and had between 2 months and 28 years of experience (median, 4.8 years). Patients on the ward ranged from 5 days to 5.4 years old (median, 6 months). Patients in the PICU ranged from 5 months to 16 years old (median, 2.5 years). Among the PICU patients, 14 (70%) were receiving mechanical ventilation only, 3 (15%) were receiving vasopressors only, and 3 (15%) were receiving mechanical ventilation and vasopressors.

We observed 5070 alarms during the 40 sessions. We excluded 108 (2.1%) that occurred at the end of video recording sessions with the nurse absent from the room because the nurse's response could not be determined. Alarms per session ranged from 10 to 1430 (median, 75; interquartile range [IQR], 35138). We excluded the outlier PICU patient with 1430 alarms in 5 hours from the analysis to avoid the potential for biasing the results. Figure 2 depicts the data flow.

Figure 2

Following the 5 training sessions, research assistants independently reviewed and made preliminary assessments on 4674 alarms; these alarms were all secondarily reviewed by a physician. Using the physician reviewer as the gold standard, the research assistant's sensitivity (assess alarm as actionable when physician also assesses as actionable) was 96.8% and specificity (assess alarm as nonactionable when physician also assesses as nonactionable) was 96.9%. We had to review 54 of 4674 alarms (1.2%) with an additional physician or nurse investigator to achieve consensus.

Characteristics of the 2445 alarms for clinical conditions are shown in Table 1. Only 12.9% of alarms in heart‐ and/or lung‐failure patients in the PICU were actionable, and only 1.0% of alarms in medical patients on a general inpatient ward were actionable.

Response‐Time Association With Nonactionable Alarm Exposure

Next, we analyzed the association between response time to potentially critical alarms that occurred when the nurse was not in the patient's room and the number of nonactionable alarms occurring over the preceding 120‐minute window. This required excluding the alarms that occurred in the first 120 minutes of each session, leaving 647 alarms with eligible response times to evaluate the exposure between prior nonactionable alarm exposure and response time: 219 in the PICU and 428 on the ward. Kaplan‐Meier plots and tabulated response times demonstrated the incremental relationships between each nonactionable alarm exposure category in the observed data, with the effects most prominent as the Kaplan‐Meier plots diverged beyond the median (Figure 3 and Table 2). Excluding the extreme outlier patient had no effect on the results, because 1378 of the 1430 alarms occurred with the nurse present at the bedside, and only 2 of the remaining alarms were potentially critical.

Figure 3

Table 2.Association Between Nonactionable Alarm Exposure in Preceding 120 Minutes and Response Time to Potentially Critical Alarms Based on Observed Data and With Response Time Modeled Using Weibull Accelerated Failure‐Time Regression

	Observed Data					Accelerated Failure‐Time Model
	Number of Potentially Critical Alarms	Minutes Elapsed Until This Percentage of Alarms Was Responded to				Modeled Response Time, min	95% CI, min	P Value*
50% (Median)		75%	90%	95%
NOTE: Abbreviations: CI, confidence interval; PICU, pediatric intensive care unit. *The critical P value used as the cut point between significant and nonsignificant, accounting for multiple comparisons, is 0.0167.
PICU
029 nonactionable alarms	70	1.6	8.0	18.6	25.1	2.8	1.9‐3.8	Reference
3079 nonactionable alarms	122	6.3	17.8	22.5	26.0	5.3	4.06.7	0.001 (vs 029)
80+ nonactionable alarms	27	16.0	28.4	32.0	33.1	8.5	4.312.7	0.009 (vs 029), 0.15 (vs 3079)
Ward
029 nonactionable alarms	159	9.8	17.8	25.0	28.9	7.7	6.39.1	Reference
3079 nonactionable alarms	211	11.6	22.4	44.6	63.2	11.5	9.613.3	0.001 (vs 029)
80+ nonactionable alarms	58	8.3	57.6	63.8	69.5	15.6	11.020.1	0.001 (vs 029), 0.09 (vs 3079)

Accelerated failure‐time regressions revealed significant incremental increases in the modeled response time as the number of preceding nonactionable alarms increased in both the PICU and ward settings (Table 2).

DISCUSSION

Alarm fatigue has recently generated interest among nurses,[22] physicians,[23] regulatory bodies,[24] patient safety organizations,[25] and even attorneys,[26] despite a lack of prior evidence linking nonactionable alarm exposure to response time or other adverse patient‐relevant outcomes. This study's main findings were that (1) the vast majority of alarms were nonactionable, (2) response time to alarms occurring while the nurse was out of the room increased as the number of nonactionable alarms over the preceding 120 minutes increased. These findings may be explained by alarm fatigue.

Our results build upon the findings of other related studies. The nonactionable alarm proportions we found were similar to other pediatric studies, reporting greater than 90% nonactionable alarms.[1, 2] One other study has reported a relationship between alarm exposure and response time. In that study, Voepel‐Lewis and colleagues evaluated nurse responses to pulse oximetry desaturation alarms in adult orthopedic surgery patients using time‐stamp data from their monitor notification system.[27] They found that alarm response time was significantly longer for patients in the highest quartile of alarms compared to those in lower quartiles. Our study provides new data suggesting a relationship between nonactionable alarm exposure and nurse response time.

Our study has several limitations. First, as a preliminary study to investigate feasibility and possible association, the sample of patients and nurses was necessarily limited and did not permit adjustment for nurse‐ or patient‐level covariates. A multivariable analysis with a larger sample might provide insight into alternate explanations for these findings other than alarm fatigue, including measures of nurse workload and patient factors (such as age and illness severity). Additional factors that are not as easily measured can also contribute to the complex decision of when and how to respond to alarms.[28, 29] Second, nurses were aware that they were being video recorded as part of a study of nonactionable alarms, although they did not know the specific details of measurement. Although this lack of blinding might lead to a Hawthorne‐like effect, our positive results suggest that this effect, if present, did not fully obscure the association. Third, all sessions took place on weekdays during daytime hours, but effects of nonactionable alarms might vary by time and day. Finally, we suspect that when nurses experience critical alarms that require them to intervene and rescue a patient, their response times to that patient's alarms that occur later in their shift will be quicker due to a heightened concern for the alarm being actionable. We were unable to explore that relationship in this analysis because the number of critical alarms requiring intervention was very small. This is a topic of future study.

CONCLUSIONS

We identified an association between a nurse's prior exposure to nonactionable alarms and response time to future alarms. This finding is consistent with alarm fatigue, but requires further study to more clearly delineate other factors that might confound or modify that relationship.

Hospital physiologic monitors can alert clinicians to early signs of physiologic deterioration, and thus have great potential to save lives. However, monitors generate frequent alarms,[1, 2, 3, 4, 5, 6, 7, 8] and most are not relevant to the patient's safety (over 90% of pediatric intensive care unit (PICU)[1, 2] and over 70% of adult intensive care alarms).[5, 6] In psychology experiments, humans rapidly learn to ignore or respond more slowly to alarms when exposed to high false‐alarm rates, exhibiting alarm fatigue.[9, 10] In 2013, The Joint Commission named alarm fatigue the most common contributing factor to alarm‐related sentinel events in hospitals.[11, 12]

Although alarm fatigue has been implicated as a major threat to patient safety, little empirical data support its existence in hospitals. In this study, we aimed to determine if there was an association between nurses' recent exposure to nonactionable physiologic monitor alarms and their response time to future alarms for the same patients. This exploratory work was designed to inform future research in this area, acknowledging that the sample size would be too small for multivariable modeling.

METHODS

Study Definitions

The alarm classification scheme is shown in Figure 1. Note that, for clarity, we have intentionally avoided using the terms true and false alarms because their interpretations vary across studies and can be misleading.

Figure 1

Potentially Critical Alarm

A potentially critical alarm is any alarm for a clinical condition for which a timely response is important to determine if the alarm requires intervention to save the patient's life. This is based on the alarm type alone, including alarms for life‐threatening arrhythmias such as asystole and ventricular tachycardia, as well as alarms for vital signs outside the set limits. Supporting Table 1 in the online version of this article lists the breakdown of alarm types that we defined a priori as potentially and not potentially critical.

Table 1.Characteristics of the 2,445 Alarms for Clinical Conditions

	PICU				Ward
Alarm type	No.	% of Total	% Valid	% Actionable	No.	% of Total	% Valid	% Actionable
NOTE: Abbreviations: N/A, not applicable; PICU, pediatric intensive care unit.
Oxygen saturation	197	19.4	82.7	38.6	590	41.2	24.4	1.9
Heart rate	194	19.1	95.4	1.0	266	18.6	87.2	0.0
Respiratory rate	229	22.6	80.8	13.5	316	22.1	48.1	1.0
Blood pressure	259	25.5	83.8	5.8	11	0.8	72.7	0.0
Critical arrhythmia	1	0.1	0.0	0.0	4	0.3	0.0	0.0
Noncritical arrhythmia	71	7.0	2.8	0.0	244	17.1	8.6	0.0
Central venous pressure	49	4.8	0.0	0.0	0	0.0	N/A	N/A
Exhaled carbon dioxide	14	1.4	92.9	50.0	0	0.0	N/A	N/A
Total	1014	100.0	75.6	12.9	1,431	100.0	38.9	1.0

Valid Alarm

A valid alarm is any alarm that correctly identifies the physiologic status of the patient. Validity was based on waveform quality, lead signal strength indicators, and artifact conditions, referencing each monitor's operator's manual.

Actionable Alarm

An actionable alarm is any valid alarm for a clinical condition that either: (1) leads to a clinical intervention; (2) leads to a consultation with another clinician at the bedside (and thus visible on camera); or (3) is a situation that should have led to intervention or consultation, but the alarm was unwitnessed or misinterpreted by the staff at the bedside.

Nonactionable Alarm

An unactionable alarm is any alarm that does not meet the actionable definition above, including invalid alarms such as those caused by motion artifact, equipment/technical alarms, and alarms that are valid but nonactionable (nuisance alarms).[13]

Response Time

The response time is the time elapsed from when the alarm fired at the bedside to when the nurse entered the room or peered through a window or door, measured in seconds.

RESULTS

We performed 40 sessions among 40 patients and 36 nurses over 210 hours. We performed 20 sessions in children with heart and/or lung failure in the PICU and 20 sessions in children on a general ward. Sessions took place on weekdays between 9:00 _am and 6:00 _pm. There were 3 occasions when we filmed 2 patients cared for by the same nurse at the same time.

Nurses were mostly female (94.4%) and had between 2 months and 28 years of experience (median, 4.8 years). Patients on the ward ranged from 5 days to 5.4 years old (median, 6 months). Patients in the PICU ranged from 5 months to 16 years old (median, 2.5 years). Among the PICU patients, 14 (70%) were receiving mechanical ventilation only, 3 (15%) were receiving vasopressors only, and 3 (15%) were receiving mechanical ventilation and vasopressors.

We observed 5070 alarms during the 40 sessions. We excluded 108 (2.1%) that occurred at the end of video recording sessions with the nurse absent from the room because the nurse's response could not be determined. Alarms per session ranged from 10 to 1430 (median, 75; interquartile range [IQR], 35138). We excluded the outlier PICU patient with 1430 alarms in 5 hours from the analysis to avoid the potential for biasing the results. Figure 2 depicts the data flow.

Figure 2

Following the 5 training sessions, research assistants independently reviewed and made preliminary assessments on 4674 alarms; these alarms were all secondarily reviewed by a physician. Using the physician reviewer as the gold standard, the research assistant's sensitivity (assess alarm as actionable when physician also assesses as actionable) was 96.8% and specificity (assess alarm as nonactionable when physician also assesses as nonactionable) was 96.9%. We had to review 54 of 4674 alarms (1.2%) with an additional physician or nurse investigator to achieve consensus.

Characteristics of the 2445 alarms for clinical conditions are shown in Table 1. Only 12.9% of alarms in heart‐ and/or lung‐failure patients in the PICU were actionable, and only 1.0% of alarms in medical patients on a general inpatient ward were actionable.

Response‐Time Association With Nonactionable Alarm Exposure

Next, we analyzed the association between response time to potentially critical alarms that occurred when the nurse was not in the patient's room and the number of nonactionable alarms occurring over the preceding 120‐minute window. This required excluding the alarms that occurred in the first 120 minutes of each session, leaving 647 alarms with eligible response times to evaluate the exposure between prior nonactionable alarm exposure and response time: 219 in the PICU and 428 on the ward. Kaplan‐Meier plots and tabulated response times demonstrated the incremental relationships between each nonactionable alarm exposure category in the observed data, with the effects most prominent as the Kaplan‐Meier plots diverged beyond the median (Figure 3 and Table 2). Excluding the extreme outlier patient had no effect on the results, because 1378 of the 1430 alarms occurred with the nurse present at the bedside, and only 2 of the remaining alarms were potentially critical.

Figure 3

Table 2.Association Between Nonactionable Alarm Exposure in Preceding 120 Minutes and Response Time to Potentially Critical Alarms Based on Observed Data and With Response Time Modeled Using Weibull Accelerated Failure‐Time Regression

	Observed Data					Accelerated Failure‐Time Model
	Number of Potentially Critical Alarms	Minutes Elapsed Until This Percentage of Alarms Was Responded to				Modeled Response Time, min	95% CI, min	P Value*
50% (Median)		75%	90%	95%
NOTE: Abbreviations: CI, confidence interval; PICU, pediatric intensive care unit. *The critical P value used as the cut point between significant and nonsignificant, accounting for multiple comparisons, is 0.0167.
PICU
029 nonactionable alarms	70	1.6	8.0	18.6	25.1	2.8	1.9‐3.8	Reference
3079 nonactionable alarms	122	6.3	17.8	22.5	26.0	5.3	4.06.7	0.001 (vs 029)
80+ nonactionable alarms	27	16.0	28.4	32.0	33.1	8.5	4.312.7	0.009 (vs 029), 0.15 (vs 3079)
Ward
029 nonactionable alarms	159	9.8	17.8	25.0	28.9	7.7	6.39.1	Reference
3079 nonactionable alarms	211	11.6	22.4	44.6	63.2	11.5	9.613.3	0.001 (vs 029)
80+ nonactionable alarms	58	8.3	57.6	63.8	69.5	15.6	11.020.1	0.001 (vs 029), 0.09 (vs 3079)

Accelerated failure‐time regressions revealed significant incremental increases in the modeled response time as the number of preceding nonactionable alarms increased in both the PICU and ward settings (Table 2).

DISCUSSION

Alarm fatigue has recently generated interest among nurses,[22] physicians,[23] regulatory bodies,[24] patient safety organizations,[25] and even attorneys,[26] despite a lack of prior evidence linking nonactionable alarm exposure to response time or other adverse patient‐relevant outcomes. This study's main findings were that (1) the vast majority of alarms were nonactionable, (2) response time to alarms occurring while the nurse was out of the room increased as the number of nonactionable alarms over the preceding 120 minutes increased. These findings may be explained by alarm fatigue.

Our results build upon the findings of other related studies. The nonactionable alarm proportions we found were similar to other pediatric studies, reporting greater than 90% nonactionable alarms.[1, 2] One other study has reported a relationship between alarm exposure and response time. In that study, Voepel‐Lewis and colleagues evaluated nurse responses to pulse oximetry desaturation alarms in adult orthopedic surgery patients using time‐stamp data from their monitor notification system.[27] They found that alarm response time was significantly longer for patients in the highest quartile of alarms compared to those in lower quartiles. Our study provides new data suggesting a relationship between nonactionable alarm exposure and nurse response time.

Our study has several limitations. First, as a preliminary study to investigate feasibility and possible association, the sample of patients and nurses was necessarily limited and did not permit adjustment for nurse‐ or patient‐level covariates. A multivariable analysis with a larger sample might provide insight into alternate explanations for these findings other than alarm fatigue, including measures of nurse workload and patient factors (such as age and illness severity). Additional factors that are not as easily measured can also contribute to the complex decision of when and how to respond to alarms.[28, 29] Second, nurses were aware that they were being video recorded as part of a study of nonactionable alarms, although they did not know the specific details of measurement. Although this lack of blinding might lead to a Hawthorne‐like effect, our positive results suggest that this effect, if present, did not fully obscure the association. Third, all sessions took place on weekdays during daytime hours, but effects of nonactionable alarms might vary by time and day. Finally, we suspect that when nurses experience critical alarms that require them to intervene and rescue a patient, their response times to that patient's alarms that occur later in their shift will be quicker due to a heightened concern for the alarm being actionable. We were unable to explore that relationship in this analysis because the number of critical alarms requiring intervention was very small. This is a topic of future study.

CONCLUSIONS

We identified an association between a nurse's prior exposure to nonactionable alarms and response time to future alarms. This finding is consistent with alarm fatigue, but requires further study to more clearly delineate other factors that might confound or modify that relationship.

Hospital physiologic monitors can alert clinicians to early signs of physiologic deterioration, and thus have great potential to save lives. However, monitors generate frequent alarms,[1, 2, 3, 4, 5, 6, 7, 8] and most are not relevant to the patient's safety (over 90% of pediatric intensive care unit (PICU)[1, 2] and over 70% of adult intensive care alarms).[5, 6] In psychology experiments, humans rapidly learn to ignore or respond more slowly to alarms when exposed to high false‐alarm rates, exhibiting alarm fatigue.[9, 10] In 2013, The Joint Commission named alarm fatigue the most common contributing factor to alarm‐related sentinel events in hospitals.[11, 12]

Although alarm fatigue has been implicated as a major threat to patient safety, little empirical data support its existence in hospitals. In this study, we aimed to determine if there was an association between nurses' recent exposure to nonactionable physiologic monitor alarms and their response time to future alarms for the same patients. This exploratory work was designed to inform future research in this area, acknowledging that the sample size would be too small for multivariable modeling.

METHODS

Study Definitions

The alarm classification scheme is shown in Figure 1. Note that, for clarity, we have intentionally avoided using the terms true and false alarms because their interpretations vary across studies and can be misleading.

Figure 1

Potentially Critical Alarm

A potentially critical alarm is any alarm for a clinical condition for which a timely response is important to determine if the alarm requires intervention to save the patient's life. This is based on the alarm type alone, including alarms for life‐threatening arrhythmias such as asystole and ventricular tachycardia, as well as alarms for vital signs outside the set limits. Supporting Table 1 in the online version of this article lists the breakdown of alarm types that we defined a priori as potentially and not potentially critical.

Table 1.Characteristics of the 2,445 Alarms for Clinical Conditions

	PICU				Ward
Alarm type	No.	% of Total	% Valid	% Actionable	No.	% of Total	% Valid	% Actionable
NOTE: Abbreviations: N/A, not applicable; PICU, pediatric intensive care unit.
Oxygen saturation	197	19.4	82.7	38.6	590	41.2	24.4	1.9
Heart rate	194	19.1	95.4	1.0	266	18.6	87.2	0.0
Respiratory rate	229	22.6	80.8	13.5	316	22.1	48.1	1.0
Blood pressure	259	25.5	83.8	5.8	11	0.8	72.7	0.0
Critical arrhythmia	1	0.1	0.0	0.0	4	0.3	0.0	0.0
Noncritical arrhythmia	71	7.0	2.8	0.0	244	17.1	8.6	0.0
Central venous pressure	49	4.8	0.0	0.0	0	0.0	N/A	N/A
Exhaled carbon dioxide	14	1.4	92.9	50.0	0	0.0	N/A	N/A
Total	1014	100.0	75.6	12.9	1,431	100.0	38.9	1.0

Valid Alarm

A valid alarm is any alarm that correctly identifies the physiologic status of the patient. Validity was based on waveform quality, lead signal strength indicators, and artifact conditions, referencing each monitor's operator's manual.

Actionable Alarm

An actionable alarm is any valid alarm for a clinical condition that either: (1) leads to a clinical intervention; (2) leads to a consultation with another clinician at the bedside (and thus visible on camera); or (3) is a situation that should have led to intervention or consultation, but the alarm was unwitnessed or misinterpreted by the staff at the bedside.

Nonactionable Alarm

An unactionable alarm is any alarm that does not meet the actionable definition above, including invalid alarms such as those caused by motion artifact, equipment/technical alarms, and alarms that are valid but nonactionable (nuisance alarms).[13]

Response Time

The response time is the time elapsed from when the alarm fired at the bedside to when the nurse entered the room or peered through a window or door, measured in seconds.

RESULTS

We performed 40 sessions among 40 patients and 36 nurses over 210 hours. We performed 20 sessions in children with heart and/or lung failure in the PICU and 20 sessions in children on a general ward. Sessions took place on weekdays between 9:00 _am and 6:00 _pm. There were 3 occasions when we filmed 2 patients cared for by the same nurse at the same time.

Nurses were mostly female (94.4%) and had between 2 months and 28 years of experience (median, 4.8 years). Patients on the ward ranged from 5 days to 5.4 years old (median, 6 months). Patients in the PICU ranged from 5 months to 16 years old (median, 2.5 years). Among the PICU patients, 14 (70%) were receiving mechanical ventilation only, 3 (15%) were receiving vasopressors only, and 3 (15%) were receiving mechanical ventilation and vasopressors.

We observed 5070 alarms during the 40 sessions. We excluded 108 (2.1%) that occurred at the end of video recording sessions with the nurse absent from the room because the nurse's response could not be determined. Alarms per session ranged from 10 to 1430 (median, 75; interquartile range [IQR], 35138). We excluded the outlier PICU patient with 1430 alarms in 5 hours from the analysis to avoid the potential for biasing the results. Figure 2 depicts the data flow.

Figure 2

Following the 5 training sessions, research assistants independently reviewed and made preliminary assessments on 4674 alarms; these alarms were all secondarily reviewed by a physician. Using the physician reviewer as the gold standard, the research assistant's sensitivity (assess alarm as actionable when physician also assesses as actionable) was 96.8% and specificity (assess alarm as nonactionable when physician also assesses as nonactionable) was 96.9%. We had to review 54 of 4674 alarms (1.2%) with an additional physician or nurse investigator to achieve consensus.

Characteristics of the 2445 alarms for clinical conditions are shown in Table 1. Only 12.9% of alarms in heart‐ and/or lung‐failure patients in the PICU were actionable, and only 1.0% of alarms in medical patients on a general inpatient ward were actionable.

Response‐Time Association With Nonactionable Alarm Exposure

Next, we analyzed the association between response time to potentially critical alarms that occurred when the nurse was not in the patient's room and the number of nonactionable alarms occurring over the preceding 120‐minute window. This required excluding the alarms that occurred in the first 120 minutes of each session, leaving 647 alarms with eligible response times to evaluate the exposure between prior nonactionable alarm exposure and response time: 219 in the PICU and 428 on the ward. Kaplan‐Meier plots and tabulated response times demonstrated the incremental relationships between each nonactionable alarm exposure category in the observed data, with the effects most prominent as the Kaplan‐Meier plots diverged beyond the median (Figure 3 and Table 2). Excluding the extreme outlier patient had no effect on the results, because 1378 of the 1430 alarms occurred with the nurse present at the bedside, and only 2 of the remaining alarms were potentially critical.

Figure 3

Table 2.Association Between Nonactionable Alarm Exposure in Preceding 120 Minutes and Response Time to Potentially Critical Alarms Based on Observed Data and With Response Time Modeled Using Weibull Accelerated Failure‐Time Regression

	Observed Data					Accelerated Failure‐Time Model
	Number of Potentially Critical Alarms	Minutes Elapsed Until This Percentage of Alarms Was Responded to				Modeled Response Time, min	95% CI, min	P Value*
50% (Median)		75%	90%	95%
NOTE: Abbreviations: CI, confidence interval; PICU, pediatric intensive care unit. *The critical P value used as the cut point between significant and nonsignificant, accounting for multiple comparisons, is 0.0167.
PICU
029 nonactionable alarms	70	1.6	8.0	18.6	25.1	2.8	1.9‐3.8	Reference
3079 nonactionable alarms	122	6.3	17.8	22.5	26.0	5.3	4.06.7	0.001 (vs 029)
80+ nonactionable alarms	27	16.0	28.4	32.0	33.1	8.5	4.312.7	0.009 (vs 029), 0.15 (vs 3079)
Ward
029 nonactionable alarms	159	9.8	17.8	25.0	28.9	7.7	6.39.1	Reference
3079 nonactionable alarms	211	11.6	22.4	44.6	63.2	11.5	9.613.3	0.001 (vs 029)
80+ nonactionable alarms	58	8.3	57.6	63.8	69.5	15.6	11.020.1	0.001 (vs 029), 0.09 (vs 3079)

Accelerated failure‐time regressions revealed significant incremental increases in the modeled response time as the number of preceding nonactionable alarms increased in both the PICU and ward settings (Table 2).

DISCUSSION

Alarm fatigue has recently generated interest among nurses,[22] physicians,[23] regulatory bodies,[24] patient safety organizations,[25] and even attorneys,[26] despite a lack of prior evidence linking nonactionable alarm exposure to response time or other adverse patient‐relevant outcomes. This study's main findings were that (1) the vast majority of alarms were nonactionable, (2) response time to alarms occurring while the nurse was out of the room increased as the number of nonactionable alarms over the preceding 120 minutes increased. These findings may be explained by alarm fatigue.

Our results build upon the findings of other related studies. The nonactionable alarm proportions we found were similar to other pediatric studies, reporting greater than 90% nonactionable alarms.[1, 2] One other study has reported a relationship between alarm exposure and response time. In that study, Voepel‐Lewis and colleagues evaluated nurse responses to pulse oximetry desaturation alarms in adult orthopedic surgery patients using time‐stamp data from their monitor notification system.[27] They found that alarm response time was significantly longer for patients in the highest quartile of alarms compared to those in lower quartiles. Our study provides new data suggesting a relationship between nonactionable alarm exposure and nurse response time.

Our study has several limitations. First, as a preliminary study to investigate feasibility and possible association, the sample of patients and nurses was necessarily limited and did not permit adjustment for nurse‐ or patient‐level covariates. A multivariable analysis with a larger sample might provide insight into alternate explanations for these findings other than alarm fatigue, including measures of nurse workload and patient factors (such as age and illness severity). Additional factors that are not as easily measured can also contribute to the complex decision of when and how to respond to alarms.[28, 29] Second, nurses were aware that they were being video recorded as part of a study of nonactionable alarms, although they did not know the specific details of measurement. Although this lack of blinding might lead to a Hawthorne‐like effect, our positive results suggest that this effect, if present, did not fully obscure the association. Third, all sessions took place on weekdays during daytime hours, but effects of nonactionable alarms might vary by time and day. Finally, we suspect that when nurses experience critical alarms that require them to intervene and rescue a patient, their response times to that patient's alarms that occur later in their shift will be quicker due to a heightened concern for the alarm being actionable. We were unable to explore that relationship in this analysis because the number of critical alarms requiring intervention was very small. This is a topic of future study.

CONCLUSIONS

We identified an association between a nurse's prior exposure to nonactionable alarms and response time to future alarms. This finding is consistent with alarm fatigue, but requires further study to more clearly delineate other factors that might confound or modify that relationship.

In the hospital, cardiopulmonary resuscitation (CPR) is the default treatment for a patient who suffers a cardiac arrest. Clinician assessment of patient preferences regarding resuscitation, with appropriate documentation in the medical record, is therefore essential for patients who do not wish to be resuscitated.[1] In addition, given frequent patient handoffs between physicians, consistent documentation of patient preferences is critical.[2] Unfortunately, multiple deficiencies in the quality of code status documentation have been identified in prior work.[3, 4] In this issue of the Journal of Hospital Medicine, Weinerman and colleagues[5] build on this literature by not only evaluating the completeness of code status documentation in multiple documentation sites, but also its consistency.

In this Canadian multihospital study, the authors found that only 38 of the 187 patients (20%) admitted to 1 of 4 medicine services had complete and consistent documentation of code status. Even more worrisome is that two‐thirds of the patients had inconsistent code status documentation. Although most of these inconsistencies involved missing information in 1 of the 5 sites of documentation (progress note, physician order, electronic resident sign‐out lists, nursing‐care plan, and do‐not‐resuscitate [DNR] face sheet), 31% were deemed clinically significant (eg, DNR in 1 source and full code in another). Such inconsistent documentation represents a serious threat to patient safety, and highlights the need for interventions aimed at improving the quality and reliability of code status documentation.

The authors identified 71 cases where code status documentation in progress notes was missing or inconsistent with documentation in other sites. Sixty of these notes lacked mention of a preference for full code status, 10 lacked documentation of DNR status, and 1 note incorrectly documented full code rather than DNR status. Interpretation of these findings requires consensus on whether the progress note is an appropriate location for code status documentation. With the evolution of the electronic medical record, the role of the progress note has changed, and unfortunately, these notes have become a lengthy chronicle of a patient's hospital course that includes all clinical data, medical problems, and an array of bottom‐of‐the‐list items such as code status. Information is easily added, but rarely removed, and what remains often goes unedited even for high‐stakes issues such as code status. Given the potential for copying and pasting of progress notes day after day, it is critical that clinicians periodically review the code status documented in the patient's notes and update this information as those preferences change. One solution that may minimize the potential for inaccurate documentation in progress notes is for institutions to utilize a separate note for code status documentation that the clinician fills out following any code status discussion. Having this note clearly labeled (eg, Code Status Note) and in a universal place within the electronic record may provide a reliable and efficient way for both physicians and nurses to identify a patient's preferences, while minimizing the inclusion of repetitive information in daily notes. Furthermore, if entered into a discrete field within the electronic record, this information could then autopopulate other sites (eg, sign‐out, nursing forms), thereby maintaining consistency. Use of note templates can provide a way to then help standardize the quality of information that is included in this type of code status note.

An alternate solution that may minimize the potential for inaccurate implementation of code status preferences is to focus on the fact that they are orders. As this study highlights, there is a need to improve both the completeness and consistency of code status documentation and, to this end, orders such as the Medical Orders for Life‐Sustaining Treatment (MOLST) or Physician Orders for Life‐Sustaining Treatment (POLST) may help.[6] Not only do these orders expand upon resuscitation preferences to include broader preferences for treatment in the context of serious illness, but they are also meant to serve as a standard way to document patient care preferences across healthcare settings. Although the MOLST and POLST primarily aim to translate patient preferences into medical orders to be followed outside of the hospital, their implementation into the electronic medical record may provide a more consistent way to document patient preferences in the hospital as well.

Although many studies have identified the need to improve the quality of code status discussions,[7, 8, 9, 10] the work by Weinerman and colleagues reminds us that attention to documentation is also critical. Ensuring that the electronic medical record contains documentation of the patient's resuscitation preferences and overall goals of care, and that this information can be found easily and reliably by physicians and nurses, should drive future quality improvement and research in this area.

In the hospital, cardiopulmonary resuscitation (CPR) is the default treatment for a patient who suffers a cardiac arrest. Clinician assessment of patient preferences regarding resuscitation, with appropriate documentation in the medical record, is therefore essential for patients who do not wish to be resuscitated.[1] In addition, given frequent patient handoffs between physicians, consistent documentation of patient preferences is critical.[2] Unfortunately, multiple deficiencies in the quality of code status documentation have been identified in prior work.[3, 4] In this issue of the Journal of Hospital Medicine, Weinerman and colleagues[5] build on this literature by not only evaluating the completeness of code status documentation in multiple documentation sites, but also its consistency.

In this Canadian multihospital study, the authors found that only 38 of the 187 patients (20%) admitted to 1 of 4 medicine services had complete and consistent documentation of code status. Even more worrisome is that two‐thirds of the patients had inconsistent code status documentation. Although most of these inconsistencies involved missing information in 1 of the 5 sites of documentation (progress note, physician order, electronic resident sign‐out lists, nursing‐care plan, and do‐not‐resuscitate [DNR] face sheet), 31% were deemed clinically significant (eg, DNR in 1 source and full code in another). Such inconsistent documentation represents a serious threat to patient safety, and highlights the need for interventions aimed at improving the quality and reliability of code status documentation.

The authors identified 71 cases where code status documentation in progress notes was missing or inconsistent with documentation in other sites. Sixty of these notes lacked mention of a preference for full code status, 10 lacked documentation of DNR status, and 1 note incorrectly documented full code rather than DNR status. Interpretation of these findings requires consensus on whether the progress note is an appropriate location for code status documentation. With the evolution of the electronic medical record, the role of the progress note has changed, and unfortunately, these notes have become a lengthy chronicle of a patient's hospital course that includes all clinical data, medical problems, and an array of bottom‐of‐the‐list items such as code status. Information is easily added, but rarely removed, and what remains often goes unedited even for high‐stakes issues such as code status. Given the potential for copying and pasting of progress notes day after day, it is critical that clinicians periodically review the code status documented in the patient's notes and update this information as those preferences change. One solution that may minimize the potential for inaccurate documentation in progress notes is for institutions to utilize a separate note for code status documentation that the clinician fills out following any code status discussion. Having this note clearly labeled (eg, Code Status Note) and in a universal place within the electronic record may provide a reliable and efficient way for both physicians and nurses to identify a patient's preferences, while minimizing the inclusion of repetitive information in daily notes. Furthermore, if entered into a discrete field within the electronic record, this information could then autopopulate other sites (eg, sign‐out, nursing forms), thereby maintaining consistency. Use of note templates can provide a way to then help standardize the quality of information that is included in this type of code status note.

An alternate solution that may minimize the potential for inaccurate implementation of code status preferences is to focus on the fact that they are orders. As this study highlights, there is a need to improve both the completeness and consistency of code status documentation and, to this end, orders such as the Medical Orders for Life‐Sustaining Treatment (MOLST) or Physician Orders for Life‐Sustaining Treatment (POLST) may help.[6] Not only do these orders expand upon resuscitation preferences to include broader preferences for treatment in the context of serious illness, but they are also meant to serve as a standard way to document patient care preferences across healthcare settings. Although the MOLST and POLST primarily aim to translate patient preferences into medical orders to be followed outside of the hospital, their implementation into the electronic medical record may provide a more consistent way to document patient preferences in the hospital as well.

Although many studies have identified the need to improve the quality of code status discussions,[7, 8, 9, 10] the work by Weinerman and colleagues reminds us that attention to documentation is also critical. Ensuring that the electronic medical record contains documentation of the patient's resuscitation preferences and overall goals of care, and that this information can be found easily and reliably by physicians and nurses, should drive future quality improvement and research in this area.

In the hospital, cardiopulmonary resuscitation (CPR) is the default treatment for a patient who suffers a cardiac arrest. Clinician assessment of patient preferences regarding resuscitation, with appropriate documentation in the medical record, is therefore essential for patients who do not wish to be resuscitated.[1] In addition, given frequent patient handoffs between physicians, consistent documentation of patient preferences is critical.[2] Unfortunately, multiple deficiencies in the quality of code status documentation have been identified in prior work.[3, 4] In this issue of the Journal of Hospital Medicine, Weinerman and colleagues[5] build on this literature by not only evaluating the completeness of code status documentation in multiple documentation sites, but also its consistency.

In this Canadian multihospital study, the authors found that only 38 of the 187 patients (20%) admitted to 1 of 4 medicine services had complete and consistent documentation of code status. Even more worrisome is that two‐thirds of the patients had inconsistent code status documentation. Although most of these inconsistencies involved missing information in 1 of the 5 sites of documentation (progress note, physician order, electronic resident sign‐out lists, nursing‐care plan, and do‐not‐resuscitate [DNR] face sheet), 31% were deemed clinically significant (eg, DNR in 1 source and full code in another). Such inconsistent documentation represents a serious threat to patient safety, and highlights the need for interventions aimed at improving the quality and reliability of code status documentation.

The authors identified 71 cases where code status documentation in progress notes was missing or inconsistent with documentation in other sites. Sixty of these notes lacked mention of a preference for full code status, 10 lacked documentation of DNR status, and 1 note incorrectly documented full code rather than DNR status. Interpretation of these findings requires consensus on whether the progress note is an appropriate location for code status documentation. With the evolution of the electronic medical record, the role of the progress note has changed, and unfortunately, these notes have become a lengthy chronicle of a patient's hospital course that includes all clinical data, medical problems, and an array of bottom‐of‐the‐list items such as code status. Information is easily added, but rarely removed, and what remains often goes unedited even for high‐stakes issues such as code status. Given the potential for copying and pasting of progress notes day after day, it is critical that clinicians periodically review the code status documented in the patient's notes and update this information as those preferences change. One solution that may minimize the potential for inaccurate documentation in progress notes is for institutions to utilize a separate note for code status documentation that the clinician fills out following any code status discussion. Having this note clearly labeled (eg, Code Status Note) and in a universal place within the electronic record may provide a reliable and efficient way for both physicians and nurses to identify a patient's preferences, while minimizing the inclusion of repetitive information in daily notes. Furthermore, if entered into a discrete field within the electronic record, this information could then autopopulate other sites (eg, sign‐out, nursing forms), thereby maintaining consistency. Use of note templates can provide a way to then help standardize the quality of information that is included in this type of code status note.

An alternate solution that may minimize the potential for inaccurate implementation of code status preferences is to focus on the fact that they are orders. As this study highlights, there is a need to improve both the completeness and consistency of code status documentation and, to this end, orders such as the Medical Orders for Life‐Sustaining Treatment (MOLST) or Physician Orders for Life‐Sustaining Treatment (POLST) may help.[6] Not only do these orders expand upon resuscitation preferences to include broader preferences for treatment in the context of serious illness, but they are also meant to serve as a standard way to document patient care preferences across healthcare settings. Although the MOLST and POLST primarily aim to translate patient preferences into medical orders to be followed outside of the hospital, their implementation into the electronic medical record may provide a more consistent way to document patient preferences in the hospital as well.

Although many studies have identified the need to improve the quality of code status discussions,[7, 8, 9, 10] the work by Weinerman and colleagues reminds us that attention to documentation is also critical. Ensuring that the electronic medical record contains documentation of the patient's resuscitation preferences and overall goals of care, and that this information can be found easily and reliably by physicians and nurses, should drive future quality improvement and research in this area.