Editorial

Fall prevention strategies for hospitalized older adults include environmental factors such as adequate room lighting and patient-specific factors such as medications. In 2008, the Centers for Medicare & Medicaid Services (CMS) implemented a regulatory “shining of the light” on hospital-acquired falls by eliminating hospital payment for fall-related injuries. Shorr et al. found that implementation of the CMS Hospital-Acquired Conditions Initiative was associated with only a modest decline in falls and injurious falls over the first seven years, with the greatest reduction occurring in urban, teaching hospitals.¹ These disappointing findings were mitigated only by the finding that the prevalence of physical restraints decreased over the seven years of observation from 1.6% to 0.6%, suggesting that the modest reductions in falls did not occur at the expense of further restricting the mobility of hospitalized older adults. Shorr et al. concluded that falls may be largely attributable to individual patient risk and may not be prevented through health system quality and safety programs such as those that have achieved successes in never-events, including wrong-side surgery and catheter-associated blood stream infections.² The authors expressed concern that hospital leaders remain in the dark regarding proven fall prevention strategies. They question whether hospital-acquired falls are preventable without restricting the mobility of older adults most at risk for falls.

Hoff et al. found in their 2011 literature review of the first three years following implementation of the 2008 CMS hospital payment polices limited evidence-based approaches to address falls as a spotlighted avoidable hospital-acquired condition.³ Swartzell et al. reported that at some level, every patient admitted to an acute care hospital is at risk for falls. “Patients sick enough to be in the hospital have underlying disease, are receiving physiologically altering medications and treatments, and are likely experiencing pain, fatigue, anxiety, sleep disturbance, and other symptoms that interfere with cognitive and physical functioning. The key to preventing falls among hospitalized patients may lie in addressing how the hospital environment creates risk.”⁴

In 2017, Avanecean et al. published a systematic review of randomized control trials on fall prevention in hospitals.⁵ Three of five studies demonstrated 20%-30% reductions in fall rates, whereas two studies showed no difference in fall rates among control and intervention groups. In the three studies that demonstrated reduced fall rates, standardized fall risk assessments were used to identify patient-specific risks for falls. Individualized care plans addressed gait and balance disorders, delirium and cognitive deficits, vision and hearing impairments, and toileting needs. For example, physical therapists provided instruction on the safe use of walkers for those with gait and balance disorders. Patients with delirium and cognitive deficits received some form of staff alert of unsupervised transfers out of bed, ranging from bed alarms to customized rubber socks that contained pressure alarms. All three successful intervention studies included patient-centered care plans for toileting.

None of the three studies that measured the secondary outcome of fall-related injuries demonstrated impact of interventions, although the rates of injurious falls were low in both the control and intervention groups (2%-5%).^3-5

Since the 2008 CMS policies eliminated hospital payments for complications of falls, patient-centered models of fall risk reduction were widely implemented. The Systems Addressing Frail Elder (SAFE) Care, designed by Ansryan et al. includes nursing, social work, pharmacist, and medical provider assessments.⁶ Team huddles occur daily to establish individualized care plans, although as Shorr et al. highlight, without report of outcomes.² Nurses Improving Care for Healthsystem Elders (NICHE) is an New York University-based nursing education and consultation program that has extended to 566 healthcare organizations.⁷ Factors that promote the adoption of organizational interventions such as NICHE have been identified.⁸ The findings that NICHE is adopted more in larger, urban healthcare systems are consistent with the findings reported by Shorr et al. that fall rate reductions were greater in such hospital settings. Patient-centered care, although time-consuming, may promote staff satisfaction and is associated with reductions in other hospital-acquired conditions such as delirium.⁹

Patient-engaged video surveillance systems are recent technological solutions to reduce falls. One staff monitors multiple patients for behaviors that risk falls such as unsupervised transfers out of bed. Staff can speak to a patient through the monitoring system to request the patient to wait for assistance, while the unit staff are alerted to the fall risk. Bedside caregivers can activate virtual privacy screens during personal patient care.

Shorr et al. appropriately call for studies to further illuminate strategies to reduce hospital-acquired falls. A multihospital report of fall rates before and after the implementation of SAFE Care and NICHE would have sufficient scale to address the impact of these patient-centered interventions on injurious falls. Similarly, patient-engaged video surveillance systems need validation from clinical trials.

Fall prevention strategies for hospitalized older adults include environmental factors such as adequate room lighting and patient-specific factors such as medications. In 2008, the Centers for Medicare & Medicaid Services (CMS) implemented a regulatory “shining of the light” on hospital-acquired falls by eliminating hospital payment for fall-related injuries. Shorr et al. found that implementation of the CMS Hospital-Acquired Conditions Initiative was associated with only a modest decline in falls and injurious falls over the first seven years, with the greatest reduction occurring in urban, teaching hospitals.¹ These disappointing findings were mitigated only by the finding that the prevalence of physical restraints decreased over the seven years of observation from 1.6% to 0.6%, suggesting that the modest reductions in falls did not occur at the expense of further restricting the mobility of hospitalized older adults. Shorr et al. concluded that falls may be largely attributable to individual patient risk and may not be prevented through health system quality and safety programs such as those that have achieved successes in never-events, including wrong-side surgery and catheter-associated blood stream infections.² The authors expressed concern that hospital leaders remain in the dark regarding proven fall prevention strategies. They question whether hospital-acquired falls are preventable without restricting the mobility of older adults most at risk for falls.

Hoff et al. found in their 2011 literature review of the first three years following implementation of the 2008 CMS hospital payment polices limited evidence-based approaches to address falls as a spotlighted avoidable hospital-acquired condition.³ Swartzell et al. reported that at some level, every patient admitted to an acute care hospital is at risk for falls. “Patients sick enough to be in the hospital have underlying disease, are receiving physiologically altering medications and treatments, and are likely experiencing pain, fatigue, anxiety, sleep disturbance, and other symptoms that interfere with cognitive and physical functioning. The key to preventing falls among hospitalized patients may lie in addressing how the hospital environment creates risk.”⁴

In 2017, Avanecean et al. published a systematic review of randomized control trials on fall prevention in hospitals.⁵ Three of five studies demonstrated 20%-30% reductions in fall rates, whereas two studies showed no difference in fall rates among control and intervention groups. In the three studies that demonstrated reduced fall rates, standardized fall risk assessments were used to identify patient-specific risks for falls. Individualized care plans addressed gait and balance disorders, delirium and cognitive deficits, vision and hearing impairments, and toileting needs. For example, physical therapists provided instruction on the safe use of walkers for those with gait and balance disorders. Patients with delirium and cognitive deficits received some form of staff alert of unsupervised transfers out of bed, ranging from bed alarms to customized rubber socks that contained pressure alarms. All three successful intervention studies included patient-centered care plans for toileting.

None of the three studies that measured the secondary outcome of fall-related injuries demonstrated impact of interventions, although the rates of injurious falls were low in both the control and intervention groups (2%-5%).^3-5

Since the 2008 CMS policies eliminated hospital payments for complications of falls, patient-centered models of fall risk reduction were widely implemented. The Systems Addressing Frail Elder (SAFE) Care, designed by Ansryan et al. includes nursing, social work, pharmacist, and medical provider assessments.⁶ Team huddles occur daily to establish individualized care plans, although as Shorr et al. highlight, without report of outcomes.² Nurses Improving Care for Healthsystem Elders (NICHE) is an New York University-based nursing education and consultation program that has extended to 566 healthcare organizations.⁷ Factors that promote the adoption of organizational interventions such as NICHE have been identified.⁸ The findings that NICHE is adopted more in larger, urban healthcare systems are consistent with the findings reported by Shorr et al. that fall rate reductions were greater in such hospital settings. Patient-centered care, although time-consuming, may promote staff satisfaction and is associated with reductions in other hospital-acquired conditions such as delirium.⁹

Patient-engaged video surveillance systems are recent technological solutions to reduce falls. One staff monitors multiple patients for behaviors that risk falls such as unsupervised transfers out of bed. Staff can speak to a patient through the monitoring system to request the patient to wait for assistance, while the unit staff are alerted to the fall risk. Bedside caregivers can activate virtual privacy screens during personal patient care.

Shorr et al. appropriately call for studies to further illuminate strategies to reduce hospital-acquired falls. A multihospital report of fall rates before and after the implementation of SAFE Care and NICHE would have sufficient scale to address the impact of these patient-centered interventions on injurious falls. Similarly, patient-engaged video surveillance systems need validation from clinical trials.

Fall prevention strategies for hospitalized older adults include environmental factors such as adequate room lighting and patient-specific factors such as medications. In 2008, the Centers for Medicare & Medicaid Services (CMS) implemented a regulatory “shining of the light” on hospital-acquired falls by eliminating hospital payment for fall-related injuries. Shorr et al. found that implementation of the CMS Hospital-Acquired Conditions Initiative was associated with only a modest decline in falls and injurious falls over the first seven years, with the greatest reduction occurring in urban, teaching hospitals.¹ These disappointing findings were mitigated only by the finding that the prevalence of physical restraints decreased over the seven years of observation from 1.6% to 0.6%, suggesting that the modest reductions in falls did not occur at the expense of further restricting the mobility of hospitalized older adults. Shorr et al. concluded that falls may be largely attributable to individual patient risk and may not be prevented through health system quality and safety programs such as those that have achieved successes in never-events, including wrong-side surgery and catheter-associated blood stream infections.² The authors expressed concern that hospital leaders remain in the dark regarding proven fall prevention strategies. They question whether hospital-acquired falls are preventable without restricting the mobility of older adults most at risk for falls.

Hoff et al. found in their 2011 literature review of the first three years following implementation of the 2008 CMS hospital payment polices limited evidence-based approaches to address falls as a spotlighted avoidable hospital-acquired condition.³ Swartzell et al. reported that at some level, every patient admitted to an acute care hospital is at risk for falls. “Patients sick enough to be in the hospital have underlying disease, are receiving physiologically altering medications and treatments, and are likely experiencing pain, fatigue, anxiety, sleep disturbance, and other symptoms that interfere with cognitive and physical functioning. The key to preventing falls among hospitalized patients may lie in addressing how the hospital environment creates risk.”⁴

In 2017, Avanecean et al. published a systematic review of randomized control trials on fall prevention in hospitals.⁵ Three of five studies demonstrated 20%-30% reductions in fall rates, whereas two studies showed no difference in fall rates among control and intervention groups. In the three studies that demonstrated reduced fall rates, standardized fall risk assessments were used to identify patient-specific risks for falls. Individualized care plans addressed gait and balance disorders, delirium and cognitive deficits, vision and hearing impairments, and toileting needs. For example, physical therapists provided instruction on the safe use of walkers for those with gait and balance disorders. Patients with delirium and cognitive deficits received some form of staff alert of unsupervised transfers out of bed, ranging from bed alarms to customized rubber socks that contained pressure alarms. All three successful intervention studies included patient-centered care plans for toileting.

None of the three studies that measured the secondary outcome of fall-related injuries demonstrated impact of interventions, although the rates of injurious falls were low in both the control and intervention groups (2%-5%).^3-5

Since the 2008 CMS policies eliminated hospital payments for complications of falls, patient-centered models of fall risk reduction were widely implemented. The Systems Addressing Frail Elder (SAFE) Care, designed by Ansryan et al. includes nursing, social work, pharmacist, and medical provider assessments.⁶ Team huddles occur daily to establish individualized care plans, although as Shorr et al. highlight, without report of outcomes.² Nurses Improving Care for Healthsystem Elders (NICHE) is an New York University-based nursing education and consultation program that has extended to 566 healthcare organizations.⁷ Factors that promote the adoption of organizational interventions such as NICHE have been identified.⁸ The findings that NICHE is adopted more in larger, urban healthcare systems are consistent with the findings reported by Shorr et al. that fall rate reductions were greater in such hospital settings. Patient-centered care, although time-consuming, may promote staff satisfaction and is associated with reductions in other hospital-acquired conditions such as delirium.⁹

Patient-engaged video surveillance systems are recent technological solutions to reduce falls. One staff monitors multiple patients for behaviors that risk falls such as unsupervised transfers out of bed. Staff can speak to a patient through the monitoring system to request the patient to wait for assistance, while the unit staff are alerted to the fall risk. Bedside caregivers can activate virtual privacy screens during personal patient care.

Shorr et al. appropriately call for studies to further illuminate strategies to reduce hospital-acquired falls. A multihospital report of fall rates before and after the implementation of SAFE Care and NICHE would have sufficient scale to address the impact of these patient-centered interventions on injurious falls. Similarly, patient-engaged video surveillance systems need validation from clinical trials.

“Never worry alone” is a common mantra that most of us have heard throughout medical training. The premise is simple and well meaning. If a patient has an issue that concerns you, ask someone for help. As a student, this can be a resident; as a resident, this can be an attending. However, for hospitalists, the answer is often a subspecialty consultation. Asking for help never seems to be wrong, but what happens when our worry delays appropriate care with unnecessary consultations? In this month’s issue of the Journal of Hospital Medicine, authors Bellas et al. have investigated this issue through the lens of subspecialty preoperative consultation for patients admitted to a hospitalist comanagement service with a fragility hip fracture requiring surgery.¹

Morbidity and mortality for patients who experience hip fractures are high, and time to appropriate surgery is one of the few modifiable risk factors that may reduce morbidity and mortality.^2,3 Bellas et al. conducted a retrospective cohort study to test the association between preoperative subspecialty consultation and multiple clinically relevant outcomes in patients admitted with an acute hip fracture.¹ All patients were comanaged by a hospitalist and orthopedic surgery, and “consultation” was defined as any preoperative subspecialty consultation requested by the hospitalist. Outcome measures included time to surgery, length of stay, readmission rate, perioperative complications, and 30-day mortality. In total, 36% (177/491) of patients who underwent surgery received a subspecialty preoperative consultation. Unsurprisingly, these patients were older with higher rates of comorbidity. After controlling for age and Charlson Comorbidity Index, preoperative consultation was associated with dramatic delays and increased rates of time to surgery >24 hours (adjusted odds ratio, 4.2; 95% CI: 2.8-6.6). The authors classified 90% of consultations as appropriate, either because of an active condition (eg, acute coronary syndrome) or because admitting physicians documented a perception that patients were at increased risk. However, 73% of consultations had only minor recommendations, such as ordering an ECG or changing the dose of an existing medication, and only 37% of the time did consultations lead to an identifiable change in management as a result of the consultation.

Although striking, integrating these findings into clinical practice is complex. As a retrospective study, patients who received consultations were obviously different from those who did not. The authors attempted to adjust for this but used only age and Charlson Comorbidity Index. Other factors that are both associated with consultations and known to increase mortality—such as frailty and functional status—were not included in their adjustment. Such unmeasured confounders possibly explain at least some, if not all, of the findings that consultations were associated with a doubling of the likelihood of 30-day mortality. In addition, although the authors assessed the appropriateness of consultation and degree of recommendations, their methods for this deserve scrutiny. Two independent providers adjudicated the consultations with excellent agreement (kappa 0.96 for indication, 0.95 for degree of recommendation), but this reliability assessment was done on previously extracted chart data, probably inflating their agreement statistics. Finally, the adjudication of consultant recommendations into minor, moderate, and major categories may oversimplify the outcome of each consultation. For example, all medication recommendations, regardless of type, were considered as minor, and recommendations were considered as major only if they resulted in invasive testing or procedures. This approach may underrepresent the impact of consultations as in clinical practice not all high-impact recommendations result in invasive testing or procedures. Despite these important limitations, Bellas et al. present a compelling case for preoperative consultation being associated with delays in surgery.

How then should this study change practice? The authors’ findings tell two separate but intertwined stories. The first is that preoperative consultation leads to delays in surgery. As patients who received preoperative consultation were obviously sicker, and because delays caused by consultation may lead to increased morbidity and mortality, perhaps the solution is to simply fix the delays. However, this approach ignores the more compelling story the authors tell. More important than the delays was the surprising lack of impact of preoperative consultations. Bellas et al. found that the majority of consultations resulted in only minor recommendations, and more importantly, hospitalists rarely changed treatment as a result. Although patients who received consultations were more ill, consultation rarely changed their care or decreased the risk posed by surgery. Bellas et al. found that only patients with active medical conditions had consultations, which resulted in moderate or major recommendations. These findings highlight an opportunity to better identify patients for whom consultation might be helpful and to prevent delays by avoiding consultation for those unlikely to benefit. There have been several efforts in the orthopedic literature to use guidelines for preoperative cardiac testing to guide cardiology consultation.^4,5,6 One study using this approach reported findings that were extremely similar to those reported by Bellas et al. in that 71% of preoperative cardiology consultations in their institution did not meet the guideline criteria for invasive cardiac testing.⁷ The primary difference between the findings of Bellas et al. and the studies in the orthopedic literature is the presence of the comanaging hospitalist. As more and more patients receive hospitalist comanagement prior to inpatient surgery, it is well within the scope of the hospitalist to differentiate chronic risk factors from active or decompensated medical disease requiring a subspecialist. This is in fact much of the value that a hospitalist adds. Avoiding consultation for patients with only elevated chronic risk factors is an important first step in avoiding unnecessary delays to surgery and an opportunity for hospitalists to improve the care of the patients they comanage.

The goal of teaching trainees to “never worry alone” is to harness the feelings of uncertainty that all providers face to improve patient care. Knowing when to worry is a valuable lesson, but as with all skills, it should be applied thoughtfully and informed by evidence. Appreciating the risks that surgery poses is quintessential to safe perioperative care, but equally important is understanding that inappropriate consultations can create risks from needless delays and testing. Only in balancing these two concerns, and appreciating when it is appropriate to worry, can we provide the highest quality of care to our patients.

“Never worry alone” is a common mantra that most of us have heard throughout medical training. The premise is simple and well meaning. If a patient has an issue that concerns you, ask someone for help. As a student, this can be a resident; as a resident, this can be an attending. However, for hospitalists, the answer is often a subspecialty consultation. Asking for help never seems to be wrong, but what happens when our worry delays appropriate care with unnecessary consultations? In this month’s issue of the Journal of Hospital Medicine, authors Bellas et al. have investigated this issue through the lens of subspecialty preoperative consultation for patients admitted to a hospitalist comanagement service with a fragility hip fracture requiring surgery.¹

Morbidity and mortality for patients who experience hip fractures are high, and time to appropriate surgery is one of the few modifiable risk factors that may reduce morbidity and mortality.^2,3 Bellas et al. conducted a retrospective cohort study to test the association between preoperative subspecialty consultation and multiple clinically relevant outcomes in patients admitted with an acute hip fracture.¹ All patients were comanaged by a hospitalist and orthopedic surgery, and “consultation” was defined as any preoperative subspecialty consultation requested by the hospitalist. Outcome measures included time to surgery, length of stay, readmission rate, perioperative complications, and 30-day mortality. In total, 36% (177/491) of patients who underwent surgery received a subspecialty preoperative consultation. Unsurprisingly, these patients were older with higher rates of comorbidity. After controlling for age and Charlson Comorbidity Index, preoperative consultation was associated with dramatic delays and increased rates of time to surgery >24 hours (adjusted odds ratio, 4.2; 95% CI: 2.8-6.6). The authors classified 90% of consultations as appropriate, either because of an active condition (eg, acute coronary syndrome) or because admitting physicians documented a perception that patients were at increased risk. However, 73% of consultations had only minor recommendations, such as ordering an ECG or changing the dose of an existing medication, and only 37% of the time did consultations lead to an identifiable change in management as a result of the consultation.

Although striking, integrating these findings into clinical practice is complex. As a retrospective study, patients who received consultations were obviously different from those who did not. The authors attempted to adjust for this but used only age and Charlson Comorbidity Index. Other factors that are both associated with consultations and known to increase mortality—such as frailty and functional status—were not included in their adjustment. Such unmeasured confounders possibly explain at least some, if not all, of the findings that consultations were associated with a doubling of the likelihood of 30-day mortality. In addition, although the authors assessed the appropriateness of consultation and degree of recommendations, their methods for this deserve scrutiny. Two independent providers adjudicated the consultations with excellent agreement (kappa 0.96 for indication, 0.95 for degree of recommendation), but this reliability assessment was done on previously extracted chart data, probably inflating their agreement statistics. Finally, the adjudication of consultant recommendations into minor, moderate, and major categories may oversimplify the outcome of each consultation. For example, all medication recommendations, regardless of type, were considered as minor, and recommendations were considered as major only if they resulted in invasive testing or procedures. This approach may underrepresent the impact of consultations as in clinical practice not all high-impact recommendations result in invasive testing or procedures. Despite these important limitations, Bellas et al. present a compelling case for preoperative consultation being associated with delays in surgery.

How then should this study change practice? The authors’ findings tell two separate but intertwined stories. The first is that preoperative consultation leads to delays in surgery. As patients who received preoperative consultation were obviously sicker, and because delays caused by consultation may lead to increased morbidity and mortality, perhaps the solution is to simply fix the delays. However, this approach ignores the more compelling story the authors tell. More important than the delays was the surprising lack of impact of preoperative consultations. Bellas et al. found that the majority of consultations resulted in only minor recommendations, and more importantly, hospitalists rarely changed treatment as a result. Although patients who received consultations were more ill, consultation rarely changed their care or decreased the risk posed by surgery. Bellas et al. found that only patients with active medical conditions had consultations, which resulted in moderate or major recommendations. These findings highlight an opportunity to better identify patients for whom consultation might be helpful and to prevent delays by avoiding consultation for those unlikely to benefit. There have been several efforts in the orthopedic literature to use guidelines for preoperative cardiac testing to guide cardiology consultation.^4,5,6 One study using this approach reported findings that were extremely similar to those reported by Bellas et al. in that 71% of preoperative cardiology consultations in their institution did not meet the guideline criteria for invasive cardiac testing.⁷ The primary difference between the findings of Bellas et al. and the studies in the orthopedic literature is the presence of the comanaging hospitalist. As more and more patients receive hospitalist comanagement prior to inpatient surgery, it is well within the scope of the hospitalist to differentiate chronic risk factors from active or decompensated medical disease requiring a subspecialist. This is in fact much of the value that a hospitalist adds. Avoiding consultation for patients with only elevated chronic risk factors is an important first step in avoiding unnecessary delays to surgery and an opportunity for hospitalists to improve the care of the patients they comanage.

The goal of teaching trainees to “never worry alone” is to harness the feelings of uncertainty that all providers face to improve patient care. Knowing when to worry is a valuable lesson, but as with all skills, it should be applied thoughtfully and informed by evidence. Appreciating the risks that surgery poses is quintessential to safe perioperative care, but equally important is understanding that inappropriate consultations can create risks from needless delays and testing. Only in balancing these two concerns, and appreciating when it is appropriate to worry, can we provide the highest quality of care to our patients.

“Never worry alone” is a common mantra that most of us have heard throughout medical training. The premise is simple and well meaning. If a patient has an issue that concerns you, ask someone for help. As a student, this can be a resident; as a resident, this can be an attending. However, for hospitalists, the answer is often a subspecialty consultation. Asking for help never seems to be wrong, but what happens when our worry delays appropriate care with unnecessary consultations? In this month’s issue of the Journal of Hospital Medicine, authors Bellas et al. have investigated this issue through the lens of subspecialty preoperative consultation for patients admitted to a hospitalist comanagement service with a fragility hip fracture requiring surgery.¹

Morbidity and mortality for patients who experience hip fractures are high, and time to appropriate surgery is one of the few modifiable risk factors that may reduce morbidity and mortality.^2,3 Bellas et al. conducted a retrospective cohort study to test the association between preoperative subspecialty consultation and multiple clinically relevant outcomes in patients admitted with an acute hip fracture.¹ All patients were comanaged by a hospitalist and orthopedic surgery, and “consultation” was defined as any preoperative subspecialty consultation requested by the hospitalist. Outcome measures included time to surgery, length of stay, readmission rate, perioperative complications, and 30-day mortality. In total, 36% (177/491) of patients who underwent surgery received a subspecialty preoperative consultation. Unsurprisingly, these patients were older with higher rates of comorbidity. After controlling for age and Charlson Comorbidity Index, preoperative consultation was associated with dramatic delays and increased rates of time to surgery >24 hours (adjusted odds ratio, 4.2; 95% CI: 2.8-6.6). The authors classified 90% of consultations as appropriate, either because of an active condition (eg, acute coronary syndrome) or because admitting physicians documented a perception that patients were at increased risk. However, 73% of consultations had only minor recommendations, such as ordering an ECG or changing the dose of an existing medication, and only 37% of the time did consultations lead to an identifiable change in management as a result of the consultation.

Although striking, integrating these findings into clinical practice is complex. As a retrospective study, patients who received consultations were obviously different from those who did not. The authors attempted to adjust for this but used only age and Charlson Comorbidity Index. Other factors that are both associated with consultations and known to increase mortality—such as frailty and functional status—were not included in their adjustment. Such unmeasured confounders possibly explain at least some, if not all, of the findings that consultations were associated with a doubling of the likelihood of 30-day mortality. In addition, although the authors assessed the appropriateness of consultation and degree of recommendations, their methods for this deserve scrutiny. Two independent providers adjudicated the consultations with excellent agreement (kappa 0.96 for indication, 0.95 for degree of recommendation), but this reliability assessment was done on previously extracted chart data, probably inflating their agreement statistics. Finally, the adjudication of consultant recommendations into minor, moderate, and major categories may oversimplify the outcome of each consultation. For example, all medication recommendations, regardless of type, were considered as minor, and recommendations were considered as major only if they resulted in invasive testing or procedures. This approach may underrepresent the impact of consultations as in clinical practice not all high-impact recommendations result in invasive testing or procedures. Despite these important limitations, Bellas et al. present a compelling case for preoperative consultation being associated with delays in surgery.

How then should this study change practice? The authors’ findings tell two separate but intertwined stories. The first is that preoperative consultation leads to delays in surgery. As patients who received preoperative consultation were obviously sicker, and because delays caused by consultation may lead to increased morbidity and mortality, perhaps the solution is to simply fix the delays. However, this approach ignores the more compelling story the authors tell. More important than the delays was the surprising lack of impact of preoperative consultations. Bellas et al. found that the majority of consultations resulted in only minor recommendations, and more importantly, hospitalists rarely changed treatment as a result. Although patients who received consultations were more ill, consultation rarely changed their care or decreased the risk posed by surgery. Bellas et al. found that only patients with active medical conditions had consultations, which resulted in moderate or major recommendations. These findings highlight an opportunity to better identify patients for whom consultation might be helpful and to prevent delays by avoiding consultation for those unlikely to benefit. There have been several efforts in the orthopedic literature to use guidelines for preoperative cardiac testing to guide cardiology consultation.^4,5,6 One study using this approach reported findings that were extremely similar to those reported by Bellas et al. in that 71% of preoperative cardiology consultations in their institution did not meet the guideline criteria for invasive cardiac testing.⁷ The primary difference between the findings of Bellas et al. and the studies in the orthopedic literature is the presence of the comanaging hospitalist. As more and more patients receive hospitalist comanagement prior to inpatient surgery, it is well within the scope of the hospitalist to differentiate chronic risk factors from active or decompensated medical disease requiring a subspecialist. This is in fact much of the value that a hospitalist adds. Avoiding consultation for patients with only elevated chronic risk factors is an important first step in avoiding unnecessary delays to surgery and an opportunity for hospitalists to improve the care of the patients they comanage.

The goal of teaching trainees to “never worry alone” is to harness the feelings of uncertainty that all providers face to improve patient care. Knowing when to worry is a valuable lesson, but as with all skills, it should be applied thoughtfully and informed by evidence. Appreciating the risks that surgery poses is quintessential to safe perioperative care, but equally important is understanding that inappropriate consultations can create risks from needless delays and testing. Only in balancing these two concerns, and appreciating when it is appropriate to worry, can we provide the highest quality of care to our patients.

Even though I could not remember her name, I remembered her story, and I would bet that my colleagues did as well. She was someone that we had all cared for at one time or another. She frequently presented to the hospital with chest pain or shortness of breath attributable to a combination of longstanding congestive heart failure, chronic obstructive pulmonary disease, and cocaine abuse. But most tragic of all, she was homeless, which meant that she was frequently hospitalized not only for medical complaints but also for a night’s shelter and a bite of food. Even though she often refused medical treatment and social workers’ efforts to stabilize her housing situation, the staff in the emergency room and observation unit all knew her by name and greeted her like an old friend. And then one day she stopped showing up to the hospital. Sitting in the emergency department (ED), I overheard that she was found outside of a storefront and had passed away. Saddened by her death, which was not unexpected given her medical issues, I still wondered if we had done right by her during the hundreds of times that she had come to our hospital. Clinicians at busy safety-net hospitals face these questions every day, and it would seem beyond doubt that our duty is to address both medical and nonmedical determinants of health of everyone that walks through our door. But is this in fact the right thing to do? Is it possible that we unwittingly expose these vulnerable patients to risks from hospitalization alone?

In this month’s Journal of Hospital Medicine, Sekijima et al. sought to quantify precisely the risks of hospitalization, particularly among the subset of patients whose “severity” of medical problems alone might not have warranted hospital admission, a scenario known colloquially as a “social” admission.¹ In real time, an inhouse triage physician classified patients as being admitted with or without “definite medical acuity.” Investigators retrospectively identified adverse events and illness acuity using standardized instruments, the Institute for Healthcare Improvement Global Trigger Tool and Emergency Severity Index, respectively. Despite the acknowledged differences in the patient population and the inherent subjectivity within the designation process, Sekijima et al. found no statistically significant difference in the percentage of admissions with an adverse event nor in the rate of adverse events per 1,000 patient days. Falls, oversedation/hypotension and code/arrest/rapid response activation were the most frequently encountered adverse events.

Delving deeper into the origin of admissions without definite medical acuity, the authors identified homelessness, lack of outpatient social support, substance use disorder, and lack of outpatient medical support as the most common reasons for “nonmedical” admissions. As healthcare providers, we recognize that these factors are generally long-term, chronic socioeconomic determinants of health. Despite our objective knowledge that we are limited in our ability to fix these problems on a short-term basis, the authors’ observations reflect our compulsion to try and help in any way possible. Patients admitted without definite medical acuity were more vulnerable and had higher rates of public insurance and housing insecurity. However, they were less acutely ill, as indicated by lower Emergency Severity Index scores. These factors were not associated with statistically significant differences in either 48-hour ED readmission or 30-day hospital readmission rates.

The process of appropriately triaging patients to an inpatient setting is challenging because of wide variability in both patients and ED providers. Hospitalists are increasingly recognized as an additional resource to assist in the triage process, as we are uniquely in a position to view the patient’s clinical presentation within the context of their anticipated clinical trajectory, promote effective utilization of inpatient bed availability, and anticipate potential barriers to discharge. Graduate medical education now identifies the triage process as a specific milestone within the transitions of care competency, as it requires mastery of interpersonal communication, professionalism, systems-based thinking, and patient-centered care.² However, many institutions lack a dedicated faculty member to perform the triage role. Our institution recently examined the feasibility of instituting a daily “huddle” between the admitting hospitalist and the ED to facilitate interdepartmental communication to create care plans in patient triage and to promote patient throughput. Available admission beds are valuable commodities, and one challenge is that the ED makes disposition decisions without knowledge of the number of available beds in the hospital. The goal of the huddle was to quickly discuss all patients potentially requiring admission prior to the final disposition decision and to address any modifiable factors to potentially prevent a “social” admission with social work early in the day. Further work is in progress to determine if introducing flexibility within existing provider roles can improve the triage process in a measurable and efficient manner.

Many challenges remain as we balance the medical needs of patients with any potential social drivers that necessitate admission to the inpatient hospital setting. From an ED perspective, social support and community follow-up were “universally considered powerful influences on admission,” and other factors such as time of day, clinical volume, and the four-hour waiting time target also played a significant role in the decision to admit.³ Hunter et al. found that admissions with moderate to low acuity may be shorter or less costly,⁴ which presents an interesting question of cost-effectiveness as an avenue for further study. As clinicians, we are intuitively aware of the subjective risk of hospitalization itself, and this work provides new objective evidence that hospitalization confers specific and quantifiable risks. Though we can undoubtedly use this knowledge to guide internal decisions about admissions and discharges, do we also have an obligation to inform our patients about these risks in real time? Ultimately, hospitalization itself might be viewed as a “procedure” or intervention that has inherent risks for all who receive it, regardless of the individual patient or hospital characteristics. As hospitalists, we should continue to strive to reduce these risks, but we should also initiate a conversation about the risks and benefits of hospitalization similarly to how we discuss other procedures with patients and their families.

Even though I could not remember her name, I remembered her story, and I would bet that my colleagues did as well. She was someone that we had all cared for at one time or another. She frequently presented to the hospital with chest pain or shortness of breath attributable to a combination of longstanding congestive heart failure, chronic obstructive pulmonary disease, and cocaine abuse. But most tragic of all, she was homeless, which meant that she was frequently hospitalized not only for medical complaints but also for a night’s shelter and a bite of food. Even though she often refused medical treatment and social workers’ efforts to stabilize her housing situation, the staff in the emergency room and observation unit all knew her by name and greeted her like an old friend. And then one day she stopped showing up to the hospital. Sitting in the emergency department (ED), I overheard that she was found outside of a storefront and had passed away. Saddened by her death, which was not unexpected given her medical issues, I still wondered if we had done right by her during the hundreds of times that she had come to our hospital. Clinicians at busy safety-net hospitals face these questions every day, and it would seem beyond doubt that our duty is to address both medical and nonmedical determinants of health of everyone that walks through our door. But is this in fact the right thing to do? Is it possible that we unwittingly expose these vulnerable patients to risks from hospitalization alone?

In this month’s Journal of Hospital Medicine, Sekijima et al. sought to quantify precisely the risks of hospitalization, particularly among the subset of patients whose “severity” of medical problems alone might not have warranted hospital admission, a scenario known colloquially as a “social” admission.¹ In real time, an inhouse triage physician classified patients as being admitted with or without “definite medical acuity.” Investigators retrospectively identified adverse events and illness acuity using standardized instruments, the Institute for Healthcare Improvement Global Trigger Tool and Emergency Severity Index, respectively. Despite the acknowledged differences in the patient population and the inherent subjectivity within the designation process, Sekijima et al. found no statistically significant difference in the percentage of admissions with an adverse event nor in the rate of adverse events per 1,000 patient days. Falls, oversedation/hypotension and code/arrest/rapid response activation were the most frequently encountered adverse events.

Delving deeper into the origin of admissions without definite medical acuity, the authors identified homelessness, lack of outpatient social support, substance use disorder, and lack of outpatient medical support as the most common reasons for “nonmedical” admissions. As healthcare providers, we recognize that these factors are generally long-term, chronic socioeconomic determinants of health. Despite our objective knowledge that we are limited in our ability to fix these problems on a short-term basis, the authors’ observations reflect our compulsion to try and help in any way possible. Patients admitted without definite medical acuity were more vulnerable and had higher rates of public insurance and housing insecurity. However, they were less acutely ill, as indicated by lower Emergency Severity Index scores. These factors were not associated with statistically significant differences in either 48-hour ED readmission or 30-day hospital readmission rates.

The process of appropriately triaging patients to an inpatient setting is challenging because of wide variability in both patients and ED providers. Hospitalists are increasingly recognized as an additional resource to assist in the triage process, as we are uniquely in a position to view the patient’s clinical presentation within the context of their anticipated clinical trajectory, promote effective utilization of inpatient bed availability, and anticipate potential barriers to discharge. Graduate medical education now identifies the triage process as a specific milestone within the transitions of care competency, as it requires mastery of interpersonal communication, professionalism, systems-based thinking, and patient-centered care.² However, many institutions lack a dedicated faculty member to perform the triage role. Our institution recently examined the feasibility of instituting a daily “huddle” between the admitting hospitalist and the ED to facilitate interdepartmental communication to create care plans in patient triage and to promote patient throughput. Available admission beds are valuable commodities, and one challenge is that the ED makes disposition decisions without knowledge of the number of available beds in the hospital. The goal of the huddle was to quickly discuss all patients potentially requiring admission prior to the final disposition decision and to address any modifiable factors to potentially prevent a “social” admission with social work early in the day. Further work is in progress to determine if introducing flexibility within existing provider roles can improve the triage process in a measurable and efficient manner.

Many challenges remain as we balance the medical needs of patients with any potential social drivers that necessitate admission to the inpatient hospital setting. From an ED perspective, social support and community follow-up were “universally considered powerful influences on admission,” and other factors such as time of day, clinical volume, and the four-hour waiting time target also played a significant role in the decision to admit.³ Hunter et al. found that admissions with moderate to low acuity may be shorter or less costly,⁴ which presents an interesting question of cost-effectiveness as an avenue for further study. As clinicians, we are intuitively aware of the subjective risk of hospitalization itself, and this work provides new objective evidence that hospitalization confers specific and quantifiable risks. Though we can undoubtedly use this knowledge to guide internal decisions about admissions and discharges, do we also have an obligation to inform our patients about these risks in real time? Ultimately, hospitalization itself might be viewed as a “procedure” or intervention that has inherent risks for all who receive it, regardless of the individual patient or hospital characteristics. As hospitalists, we should continue to strive to reduce these risks, but we should also initiate a conversation about the risks and benefits of hospitalization similarly to how we discuss other procedures with patients and their families.

Even though I could not remember her name, I remembered her story, and I would bet that my colleagues did as well. She was someone that we had all cared for at one time or another. She frequently presented to the hospital with chest pain or shortness of breath attributable to a combination of longstanding congestive heart failure, chronic obstructive pulmonary disease, and cocaine abuse. But most tragic of all, she was homeless, which meant that she was frequently hospitalized not only for medical complaints but also for a night’s shelter and a bite of food. Even though she often refused medical treatment and social workers’ efforts to stabilize her housing situation, the staff in the emergency room and observation unit all knew her by name and greeted her like an old friend. And then one day she stopped showing up to the hospital. Sitting in the emergency department (ED), I overheard that she was found outside of a storefront and had passed away. Saddened by her death, which was not unexpected given her medical issues, I still wondered if we had done right by her during the hundreds of times that she had come to our hospital. Clinicians at busy safety-net hospitals face these questions every day, and it would seem beyond doubt that our duty is to address both medical and nonmedical determinants of health of everyone that walks through our door. But is this in fact the right thing to do? Is it possible that we unwittingly expose these vulnerable patients to risks from hospitalization alone?

In this month’s Journal of Hospital Medicine, Sekijima et al. sought to quantify precisely the risks of hospitalization, particularly among the subset of patients whose “severity” of medical problems alone might not have warranted hospital admission, a scenario known colloquially as a “social” admission.¹ In real time, an inhouse triage physician classified patients as being admitted with or without “definite medical acuity.” Investigators retrospectively identified adverse events and illness acuity using standardized instruments, the Institute for Healthcare Improvement Global Trigger Tool and Emergency Severity Index, respectively. Despite the acknowledged differences in the patient population and the inherent subjectivity within the designation process, Sekijima et al. found no statistically significant difference in the percentage of admissions with an adverse event nor in the rate of adverse events per 1,000 patient days. Falls, oversedation/hypotension and code/arrest/rapid response activation were the most frequently encountered adverse events.

Delving deeper into the origin of admissions without definite medical acuity, the authors identified homelessness, lack of outpatient social support, substance use disorder, and lack of outpatient medical support as the most common reasons for “nonmedical” admissions. As healthcare providers, we recognize that these factors are generally long-term, chronic socioeconomic determinants of health. Despite our objective knowledge that we are limited in our ability to fix these problems on a short-term basis, the authors’ observations reflect our compulsion to try and help in any way possible. Patients admitted without definite medical acuity were more vulnerable and had higher rates of public insurance and housing insecurity. However, they were less acutely ill, as indicated by lower Emergency Severity Index scores. These factors were not associated with statistically significant differences in either 48-hour ED readmission or 30-day hospital readmission rates.

The process of appropriately triaging patients to an inpatient setting is challenging because of wide variability in both patients and ED providers. Hospitalists are increasingly recognized as an additional resource to assist in the triage process, as we are uniquely in a position to view the patient’s clinical presentation within the context of their anticipated clinical trajectory, promote effective utilization of inpatient bed availability, and anticipate potential barriers to discharge. Graduate medical education now identifies the triage process as a specific milestone within the transitions of care competency, as it requires mastery of interpersonal communication, professionalism, systems-based thinking, and patient-centered care.² However, many institutions lack a dedicated faculty member to perform the triage role. Our institution recently examined the feasibility of instituting a daily “huddle” between the admitting hospitalist and the ED to facilitate interdepartmental communication to create care plans in patient triage and to promote patient throughput. Available admission beds are valuable commodities, and one challenge is that the ED makes disposition decisions without knowledge of the number of available beds in the hospital. The goal of the huddle was to quickly discuss all patients potentially requiring admission prior to the final disposition decision and to address any modifiable factors to potentially prevent a “social” admission with social work early in the day. Further work is in progress to determine if introducing flexibility within existing provider roles can improve the triage process in a measurable and efficient manner.

Many challenges remain as we balance the medical needs of patients with any potential social drivers that necessitate admission to the inpatient hospital setting. From an ED perspective, social support and community follow-up were “universally considered powerful influences on admission,” and other factors such as time of day, clinical volume, and the four-hour waiting time target also played a significant role in the decision to admit.³ Hunter et al. found that admissions with moderate to low acuity may be shorter or less costly,⁴ which presents an interesting question of cost-effectiveness as an avenue for further study. As clinicians, we are intuitively aware of the subjective risk of hospitalization itself, and this work provides new objective evidence that hospitalization confers specific and quantifiable risks. Though we can undoubtedly use this knowledge to guide internal decisions about admissions and discharges, do we also have an obligation to inform our patients about these risks in real time? Ultimately, hospitalization itself might be viewed as a “procedure” or intervention that has inherent risks for all who receive it, regardless of the individual patient or hospital characteristics. As hospitalists, we should continue to strive to reduce these risks, but we should also initiate a conversation about the risks and benefits of hospitalization similarly to how we discuss other procedures with patients and their families.

In an effort to improve healthcare for Americans by linking hospital payments to quality of care, Medicare’s Hospital Readmission Reduction Program (HRRP) began penalizing hospitals with “excess” readmission rates in 2012. The decision sparked widespread debate about the definition of a preventable readmission and whether a patient’s socioeconomic status should be considered for risk adjustment. Although coming back to the hospital after an admission is an undesirable outcome for any patient, the suitability of readmission as a quality measure remains a hot and debated topic. Research on the subject skyrocketed; over 12000 articles about hospital readmissions have been indexed in PubMed since 2000, and the number of publications has steadily increased since 2010 (Figure).

Although the HRRP is a Medicare initiative, there has been a substantial focus on readmissions in pediatrics as well. The National Quality Forum has endorsed three quality measures specific to readmission in children: (1) the rate of unplanned readmissions to the pediatric intensive care unit within 24 hours after discharge or transfer, (2) the pediatric lower respiratory infection readmission measure, defined as the percentage of admissions followed by one or more readmissions within 30 days of hospitalization for lower respiratory infection, and (3) the pediatric all-cause readmission measure, defined as the percentage of admissions followed by one or more readmissions within 30 days. These endorsements were preceded by studies showing that pediatric readmission rates varied substantially across hospitals and clinical conditions, and that children with chronic illnesses were at the highest risk.

Readmission is an attractive pediatric quality measure for a number of reasons. This measure is easy to apply to data at the hospital, health system, and payor levels at relatively low cost. Relatedly, the all-condition measure can be applied to all pediatric hospitalizations, overcoming the very real challenge in pediatric quality measurement of inadequate sample sizes to discern differences in healthcare quality at the hospital level for many disease-specific measures.¹ In addition, this measure moves beyond process measurement to quantify an outcome relevant to families as well as healthcare systems. Finally, the measure is founded on a compelling conceptual framework (albeit one that remains challenging to prove) that efforts to improve a patient’s hospital-to-home transition and discharge readiness will reduce their likelihood of readmission.

In this issue of the Journal of Hospital Medicine, Katherine Auger and colleagues present their analysis of pediatric readmission rates from 2010 to 2016 across 66 children’s hospitals.² They found that the median seven-day all-cause pediatric readmission rate was 5.1%, with no change in rates over the seven-year study period. Applying proprietary software to identify potentially preventable readmissions (PPR), they reported that approximately 40% of these readmissions may be preventable, a proportion that was also unchanged over time. Interestingly, 88% of the hospitals represented in their data were participating in the Solutions for Patient Safety national learning collaborative during the study period, making efforts to reduce seven-day readmission rates. Despite this, the figures presented in this paper of all-condition and potentially preventable readmission rates over time are very, very flat.

This work by Auger et al. contributes to our understanding about the preventability, or lack thereof, of pediatric all-condition readmissions. If 40% of these readmissions are indeed preventable, then why did Auger et al. not observe a declining proportion of PPR over time as a result of hospital participation in a national collaborative? Past quantitative and qualitative studies provide important context. First, the 40% rate of readmission preventability is twofold higher than that reported in past studies that relied on physician judgement to determine readmission preventability;^3,4 the authors’ use of proprietary software to categorize the preventability of a readmission limits our ability to explain the differences in these rates. However, in these past studies, the rates of initial agreement between physician reviewers about readmission preventability were poor, highlighting the challenges associated with determining readmission preventability. Moreover, qualitative studies suggest that physicians and families lack a shared understanding of the preventability of readmissions.⁵ Finally, a systematic review of pediatric hospital discharge interventions did not identify any one intervention that was consistently effective in reducing hospital readmission rates.⁶ The following important questions remain: Were hospitals’ efforts to reduce PPR targeting the wrong patients? Were the interventions insufficient or ineffective? Or are readmission measures insufficiently sensitive to improved processes of care?

Recognizing that the majority of research on readmission as well as HRRP penalties focuses on adult populations, perhaps we can apply some lessons learned from the HRRP to pediatrics. Recent analyses by Medicare Payment Advisory Commission (MedPAC) suggest that raw and risk-adjusted readmission rates have declined for conditions covered by the HRRP, with readmission rates for HRRP target conditions declining more quickly than that for nontarget conditions.⁷ Just as the HRRP has focused on target conditions with relatively high readmission rates, analogous efforts to focus pediatric readmission reduction on children at greatest risk may enable measurement of change over time. For example, although children with complex chronic medical conditions represent a small proportion of the pediatric population, they account for 60% of all pediatric readmissions in the United States. However, similar to the above-described meta-analysis of readmission reduction efforts in children, at least one meta-analysis has demonstrated that there is no one intervention or even bundle of interventions that has consistently reduced readmissions in adults.⁸ Although the readmission rates for HRRP target conditions have decreased, the results of clinical trials evaluating readmission reduction efforts are difficult to translate into practice given substantial heterogeneity in study designs, interventions, and patient populations.

Does this study by Auger et al. suggest that pediatric readmission reduction efforts are misguided or futile? No. But it does provide compelling data that efforts to reduce all-cause readmissions for all children may not yield measureable changes using the current measures. A narrowed focus on children with chronic illnesses, who account for approximately half of all pediatric admissions, may be warranted. A number of studies have summarized families’ preferences regarding their hospital-to-home transitions; the results indicate that families of children with chronic illness have unique desires and needs.^9,10 Perhaps it is time to take a step back from pediatric readmission reduction efforts, largely inspired by the HRRP, and redirect our resources to implement and evaluate processes and outcomes most valued by children and their families.

Disclosures

Drs. Lagu and Lindenauer have served as consultants for the Yale Center for Outcomes Research and Evaluation (under contract to the Centers for Medicare and Medicaid Services) providing clinical and methodological expertise and input on the development, reevaluation, and implementation of hospital outcome and efficiency measures.

Funding

Dr. Lagu is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award R01 HL139985-01A1 and 1R01HL146884-01. Dr. Lindenauer was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K24HL132008.

Disclaimer

The views expressed in this manuscript do not necessarily reflect those of the Yale Center for Outcomes Research and Evaluation or the Centers for Medicare and Medicaid Services. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

In an effort to improve healthcare for Americans by linking hospital payments to quality of care, Medicare’s Hospital Readmission Reduction Program (HRRP) began penalizing hospitals with “excess” readmission rates in 2012. The decision sparked widespread debate about the definition of a preventable readmission and whether a patient’s socioeconomic status should be considered for risk adjustment. Although coming back to the hospital after an admission is an undesirable outcome for any patient, the suitability of readmission as a quality measure remains a hot and debated topic. Research on the subject skyrocketed; over 12000 articles about hospital readmissions have been indexed in PubMed since 2000, and the number of publications has steadily increased since 2010 (Figure).

Although the HRRP is a Medicare initiative, there has been a substantial focus on readmissions in pediatrics as well. The National Quality Forum has endorsed three quality measures specific to readmission in children: (1) the rate of unplanned readmissions to the pediatric intensive care unit within 24 hours after discharge or transfer, (2) the pediatric lower respiratory infection readmission measure, defined as the percentage of admissions followed by one or more readmissions within 30 days of hospitalization for lower respiratory infection, and (3) the pediatric all-cause readmission measure, defined as the percentage of admissions followed by one or more readmissions within 30 days. These endorsements were preceded by studies showing that pediatric readmission rates varied substantially across hospitals and clinical conditions, and that children with chronic illnesses were at the highest risk.

Readmission is an attractive pediatric quality measure for a number of reasons. This measure is easy to apply to data at the hospital, health system, and payor levels at relatively low cost. Relatedly, the all-condition measure can be applied to all pediatric hospitalizations, overcoming the very real challenge in pediatric quality measurement of inadequate sample sizes to discern differences in healthcare quality at the hospital level for many disease-specific measures.¹ In addition, this measure moves beyond process measurement to quantify an outcome relevant to families as well as healthcare systems. Finally, the measure is founded on a compelling conceptual framework (albeit one that remains challenging to prove) that efforts to improve a patient’s hospital-to-home transition and discharge readiness will reduce their likelihood of readmission.

In this issue of the Journal of Hospital Medicine, Katherine Auger and colleagues present their analysis of pediatric readmission rates from 2010 to 2016 across 66 children’s hospitals.² They found that the median seven-day all-cause pediatric readmission rate was 5.1%, with no change in rates over the seven-year study period. Applying proprietary software to identify potentially preventable readmissions (PPR), they reported that approximately 40% of these readmissions may be preventable, a proportion that was also unchanged over time. Interestingly, 88% of the hospitals represented in their data were participating in the Solutions for Patient Safety national learning collaborative during the study period, making efforts to reduce seven-day readmission rates. Despite this, the figures presented in this paper of all-condition and potentially preventable readmission rates over time are very, very flat.

This work by Auger et al. contributes to our understanding about the preventability, or lack thereof, of pediatric all-condition readmissions. If 40% of these readmissions are indeed preventable, then why did Auger et al. not observe a declining proportion of PPR over time as a result of hospital participation in a national collaborative? Past quantitative and qualitative studies provide important context. First, the 40% rate of readmission preventability is twofold higher than that reported in past studies that relied on physician judgement to determine readmission preventability;^3,4 the authors’ use of proprietary software to categorize the preventability of a readmission limits our ability to explain the differences in these rates. However, in these past studies, the rates of initial agreement between physician reviewers about readmission preventability were poor, highlighting the challenges associated with determining readmission preventability. Moreover, qualitative studies suggest that physicians and families lack a shared understanding of the preventability of readmissions.⁵ Finally, a systematic review of pediatric hospital discharge interventions did not identify any one intervention that was consistently effective in reducing hospital readmission rates.⁶ The following important questions remain: Were hospitals’ efforts to reduce PPR targeting the wrong patients? Were the interventions insufficient or ineffective? Or are readmission measures insufficiently sensitive to improved processes of care?

Recognizing that the majority of research on readmission as well as HRRP penalties focuses on adult populations, perhaps we can apply some lessons learned from the HRRP to pediatrics. Recent analyses by Medicare Payment Advisory Commission (MedPAC) suggest that raw and risk-adjusted readmission rates have declined for conditions covered by the HRRP, with readmission rates for HRRP target conditions declining more quickly than that for nontarget conditions.⁷ Just as the HRRP has focused on target conditions with relatively high readmission rates, analogous efforts to focus pediatric readmission reduction on children at greatest risk may enable measurement of change over time. For example, although children with complex chronic medical conditions represent a small proportion of the pediatric population, they account for 60% of all pediatric readmissions in the United States. However, similar to the above-described meta-analysis of readmission reduction efforts in children, at least one meta-analysis has demonstrated that there is no one intervention or even bundle of interventions that has consistently reduced readmissions in adults.⁸ Although the readmission rates for HRRP target conditions have decreased, the results of clinical trials evaluating readmission reduction efforts are difficult to translate into practice given substantial heterogeneity in study designs, interventions, and patient populations.

Does this study by Auger et al. suggest that pediatric readmission reduction efforts are misguided or futile? No. But it does provide compelling data that efforts to reduce all-cause readmissions for all children may not yield measureable changes using the current measures. A narrowed focus on children with chronic illnesses, who account for approximately half of all pediatric admissions, may be warranted. A number of studies have summarized families’ preferences regarding their hospital-to-home transitions; the results indicate that families of children with chronic illness have unique desires and needs.^9,10 Perhaps it is time to take a step back from pediatric readmission reduction efforts, largely inspired by the HRRP, and redirect our resources to implement and evaluate processes and outcomes most valued by children and their families.

Disclosures

Drs. Lagu and Lindenauer have served as consultants for the Yale Center for Outcomes Research and Evaluation (under contract to the Centers for Medicare and Medicaid Services) providing clinical and methodological expertise and input on the development, reevaluation, and implementation of hospital outcome and efficiency measures.

Funding

Dr. Lagu is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award R01 HL139985-01A1 and 1R01HL146884-01. Dr. Lindenauer was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K24HL132008.

Disclaimer

The views expressed in this manuscript do not necessarily reflect those of the Yale Center for Outcomes Research and Evaluation or the Centers for Medicare and Medicaid Services. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

In an effort to improve healthcare for Americans by linking hospital payments to quality of care, Medicare’s Hospital Readmission Reduction Program (HRRP) began penalizing hospitals with “excess” readmission rates in 2012. The decision sparked widespread debate about the definition of a preventable readmission and whether a patient’s socioeconomic status should be considered for risk adjustment. Although coming back to the hospital after an admission is an undesirable outcome for any patient, the suitability of readmission as a quality measure remains a hot and debated topic. Research on the subject skyrocketed; over 12000 articles about hospital readmissions have been indexed in PubMed since 2000, and the number of publications has steadily increased since 2010 (Figure).

Although the HRRP is a Medicare initiative, there has been a substantial focus on readmissions in pediatrics as well. The National Quality Forum has endorsed three quality measures specific to readmission in children: (1) the rate of unplanned readmissions to the pediatric intensive care unit within 24 hours after discharge or transfer, (2) the pediatric lower respiratory infection readmission measure, defined as the percentage of admissions followed by one or more readmissions within 30 days of hospitalization for lower respiratory infection, and (3) the pediatric all-cause readmission measure, defined as the percentage of admissions followed by one or more readmissions within 30 days. These endorsements were preceded by studies showing that pediatric readmission rates varied substantially across hospitals and clinical conditions, and that children with chronic illnesses were at the highest risk.

Readmission is an attractive pediatric quality measure for a number of reasons. This measure is easy to apply to data at the hospital, health system, and payor levels at relatively low cost. Relatedly, the all-condition measure can be applied to all pediatric hospitalizations, overcoming the very real challenge in pediatric quality measurement of inadequate sample sizes to discern differences in healthcare quality at the hospital level for many disease-specific measures.¹ In addition, this measure moves beyond process measurement to quantify an outcome relevant to families as well as healthcare systems. Finally, the measure is founded on a compelling conceptual framework (albeit one that remains challenging to prove) that efforts to improve a patient’s hospital-to-home transition and discharge readiness will reduce their likelihood of readmission.

In this issue of the Journal of Hospital Medicine, Katherine Auger and colleagues present their analysis of pediatric readmission rates from 2010 to 2016 across 66 children’s hospitals.² They found that the median seven-day all-cause pediatric readmission rate was 5.1%, with no change in rates over the seven-year study period. Applying proprietary software to identify potentially preventable readmissions (PPR), they reported that approximately 40% of these readmissions may be preventable, a proportion that was also unchanged over time. Interestingly, 88% of the hospitals represented in their data were participating in the Solutions for Patient Safety national learning collaborative during the study period, making efforts to reduce seven-day readmission rates. Despite this, the figures presented in this paper of all-condition and potentially preventable readmission rates over time are very, very flat.

This work by Auger et al. contributes to our understanding about the preventability, or lack thereof, of pediatric all-condition readmissions. If 40% of these readmissions are indeed preventable, then why did Auger et al. not observe a declining proportion of PPR over time as a result of hospital participation in a national collaborative? Past quantitative and qualitative studies provide important context. First, the 40% rate of readmission preventability is twofold higher than that reported in past studies that relied on physician judgement to determine readmission preventability;^3,4 the authors’ use of proprietary software to categorize the preventability of a readmission limits our ability to explain the differences in these rates. However, in these past studies, the rates of initial agreement between physician reviewers about readmission preventability were poor, highlighting the challenges associated with determining readmission preventability. Moreover, qualitative studies suggest that physicians and families lack a shared understanding of the preventability of readmissions.⁵ Finally, a systematic review of pediatric hospital discharge interventions did not identify any one intervention that was consistently effective in reducing hospital readmission rates.⁶ The following important questions remain: Were hospitals’ efforts to reduce PPR targeting the wrong patients? Were the interventions insufficient or ineffective? Or are readmission measures insufficiently sensitive to improved processes of care?

Recognizing that the majority of research on readmission as well as HRRP penalties focuses on adult populations, perhaps we can apply some lessons learned from the HRRP to pediatrics. Recent analyses by Medicare Payment Advisory Commission (MedPAC) suggest that raw and risk-adjusted readmission rates have declined for conditions covered by the HRRP, with readmission rates for HRRP target conditions declining more quickly than that for nontarget conditions.⁷ Just as the HRRP has focused on target conditions with relatively high readmission rates, analogous efforts to focus pediatric readmission reduction on children at greatest risk may enable measurement of change over time. For example, although children with complex chronic medical conditions represent a small proportion of the pediatric population, they account for 60% of all pediatric readmissions in the United States. However, similar to the above-described meta-analysis of readmission reduction efforts in children, at least one meta-analysis has demonstrated that there is no one intervention or even bundle of interventions that has consistently reduced readmissions in adults.⁸ Although the readmission rates for HRRP target conditions have decreased, the results of clinical trials evaluating readmission reduction efforts are difficult to translate into practice given substantial heterogeneity in study designs, interventions, and patient populations.

Does this study by Auger et al. suggest that pediatric readmission reduction efforts are misguided or futile? No. But it does provide compelling data that efforts to reduce all-cause readmissions for all children may not yield measureable changes using the current measures. A narrowed focus on children with chronic illnesses, who account for approximately half of all pediatric admissions, may be warranted. A number of studies have summarized families’ preferences regarding their hospital-to-home transitions; the results indicate that families of children with chronic illness have unique desires and needs.^9,10 Perhaps it is time to take a step back from pediatric readmission reduction efforts, largely inspired by the HRRP, and redirect our resources to implement and evaluate processes and outcomes most valued by children and their families.

Disclosures

Drs. Lagu and Lindenauer have served as consultants for the Yale Center for Outcomes Research and Evaluation (under contract to the Centers for Medicare and Medicaid Services) providing clinical and methodological expertise and input on the development, reevaluation, and implementation of hospital outcome and efficiency measures.

Funding

Dr. Lagu is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award R01 HL139985-01A1 and 1R01HL146884-01. Dr. Lindenauer was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number K24HL132008.

Disclaimer

The views expressed in this manuscript do not necessarily reflect those of the Yale Center for Outcomes Research and Evaluation or the Centers for Medicare and Medicaid Services. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Of all the errors that occur in modern healthcare, medication errors are among the most ubiquitous and consequential. Adverse drug events (ADEs) account for approximately 700,000 emergency department visits, 100,000 hospitalizations, and 1.3 million people are injured by medication errors annually.¹ Among the most frequent causes of preventable ADEs are errors on the medication lists when patients are admitted to hospitals.² Therefore, preventing discrepancies between medications the patient is prescribed (and actually taking) inside and outside the hospital—the so-called “medication reconciliation”—is an intense, ongoing area of focus for health systems, pharmacies, and numerous quality and safety organizations seeking to reduce ADEs.

Past studies of medication reconciliation interventions have suggested benefit from restricting medication reconciliation to admission or discharge, pharmacist or pharmacy technician-led medication reconciliation, and pharmacy-led interventions (ie, telephone follow-up/home visit, patient counseling) for ensuring an accurate medication list.^3-5 Recent evidence suggests that pharmacist discharge medication reconciliation is associated with decreased readmission rates, decreased medication discrepancies, and adverse events associated with drug therapy issues.⁴ The successful interventions were promising, but disseminating such interventions can often be very complex.⁶

In this issue of the Journal of Hospital Medicine, Mixon et al. report the results of a subanalysis of the MARQUIS trial,⁷ wherein they individually examined the on-protocol effects of the interventions that MARQUIS recommended, comparing hospitals to their own running baseline data at the implementation of each intervention to data following the implementation. The authors found that only three of the nine interventions were associated with reducing potentially harmful discrepancies in the medication list—training existing staff to perform discharge medication reconciliation, hiring additional staff for this purpose, and defining roles and responsibilities and roles clearly—and that two were actually associated with harm—training existing staff to take best possible medication histories (BPMHs) and implementing a new Electronic Medical Record (EMR). MARQUIS is unique in not just attempting but in reporting “best case” real-world implementation using available literature to design mentored, practical approaches to those same interventions at sites not involved in their initial setup and validation.

EMR implementation should in theory improve accuracy (or at least legibility), but it can also contribute to new types of inaccuracy or, as the authors propose, deprioritize quality and safety as organizational goals during the rigors of digitization. Similarly, training staff to take a BPMH might create false confidence in the results or interact with medication reconciliation in other complex ways. Opting to add more work instead of hiring additional staff may have increased the burden of medication review and thus contributed to its inaccuracy.

On the contrary, certain interventions, such as having clear accountability for the medication list, hiring additional staff to construct that list, and clearly defining the roles of those involved in the reconciliation process, were associated with improved medication reconciliation. All these strategies require resource allocation, but at least the current study provides evidence that such resource allocation can be effective in new settings as they were in their original ones.

The study has important acknowledged limitations. The on-protocol analysis limited the authors to reporting associations rather than causality. Moreover, the original trial ran from 2011 to 2014, which was a time of rapid EMR implementation and new recognition of the problems posed by the same; several organizations are in a far more mature EMR context today. Conversely, newer technologies such as patient-facing medication reconciliation applications, cross-organization medication lists available from some EMR vendors, and health platforms that collect data from multiple EMRs were not evaluated because they did not exist at the time of the original trial. Another important trend in healthcare, the rise of Accountable Care Organizations and their focus on integration and defragmentation, may have an important part to play in medication list accuracy. All the above-mentioned aspects will be important avenues for ongoing research in real-world medication reconciliation.

Mixon’s findings come at a time when medication reconciliation is again a national health informatics priority, a key component of the Medicare Access and CHIP reauthorization Act of 2015 and Merit-based Incentive Payments System⁸ since 2019, with hospitals reporting medication reconciliation rates for financial in addition to quality and safety reasons. Hopefully, this study and others, in combination with the abovementioned incentives, will stimulate further research into impactful strategies for medication reconciliation and ideal ways to implement them. With luck, the end result will be more generalizable interventions, with a track record of success, that would help ensure that patients are prescribed, are reporting, are taking, and are noted to be taking the medications that they and their providers intended, both on presentation to the hospital and on discharge home.

Disclosures

Vicki Jue has no conflicts of interest to report. Raman Khanna reports developing CareWeb, a communication platform that has been licensed to Voalte, Inc. This work is unrelated to the current editorial. No other conflicts of interest to report.

Of all the errors that occur in modern healthcare, medication errors are among the most ubiquitous and consequential. Adverse drug events (ADEs) account for approximately 700,000 emergency department visits, 100,000 hospitalizations, and 1.3 million people are injured by medication errors annually.¹ Among the most frequent causes of preventable ADEs are errors on the medication lists when patients are admitted to hospitals.² Therefore, preventing discrepancies between medications the patient is prescribed (and actually taking) inside and outside the hospital—the so-called “medication reconciliation”—is an intense, ongoing area of focus for health systems, pharmacies, and numerous quality and safety organizations seeking to reduce ADEs.

Past studies of medication reconciliation interventions have suggested benefit from restricting medication reconciliation to admission or discharge, pharmacist or pharmacy technician-led medication reconciliation, and pharmacy-led interventions (ie, telephone follow-up/home visit, patient counseling) for ensuring an accurate medication list.^3-5 Recent evidence suggests that pharmacist discharge medication reconciliation is associated with decreased readmission rates, decreased medication discrepancies, and adverse events associated with drug therapy issues.⁴ The successful interventions were promising, but disseminating such interventions can often be very complex.⁶

In this issue of the Journal of Hospital Medicine, Mixon et al. report the results of a subanalysis of the MARQUIS trial,⁷ wherein they individually examined the on-protocol effects of the interventions that MARQUIS recommended, comparing hospitals to their own running baseline data at the implementation of each intervention to data following the implementation. The authors found that only three of the nine interventions were associated with reducing potentially harmful discrepancies in the medication list—training existing staff to perform discharge medication reconciliation, hiring additional staff for this purpose, and defining roles and responsibilities and roles clearly—and that two were actually associated with harm—training existing staff to take best possible medication histories (BPMHs) and implementing a new Electronic Medical Record (EMR). MARQUIS is unique in not just attempting but in reporting “best case” real-world implementation using available literature to design mentored, practical approaches to those same interventions at sites not involved in their initial setup and validation.

EMR implementation should in theory improve accuracy (or at least legibility), but it can also contribute to new types of inaccuracy or, as the authors propose, deprioritize quality and safety as organizational goals during the rigors of digitization. Similarly, training staff to take a BPMH might create false confidence in the results or interact with medication reconciliation in other complex ways. Opting to add more work instead of hiring additional staff may have increased the burden of medication review and thus contributed to its inaccuracy.

On the contrary, certain interventions, such as having clear accountability for the medication list, hiring additional staff to construct that list, and clearly defining the roles of those involved in the reconciliation process, were associated with improved medication reconciliation. All these strategies require resource allocation, but at least the current study provides evidence that such resource allocation can be effective in new settings as they were in their original ones.

The study has important acknowledged limitations. The on-protocol analysis limited the authors to reporting associations rather than causality. Moreover, the original trial ran from 2011 to 2014, which was a time of rapid EMR implementation and new recognition of the problems posed by the same; several organizations are in a far more mature EMR context today. Conversely, newer technologies such as patient-facing medication reconciliation applications, cross-organization medication lists available from some EMR vendors, and health platforms that collect data from multiple EMRs were not evaluated because they did not exist at the time of the original trial. Another important trend in healthcare, the rise of Accountable Care Organizations and their focus on integration and defragmentation, may have an important part to play in medication list accuracy. All the above-mentioned aspects will be important avenues for ongoing research in real-world medication reconciliation.

Mixon’s findings come at a time when medication reconciliation is again a national health informatics priority, a key component of the Medicare Access and CHIP reauthorization Act of 2015 and Merit-based Incentive Payments System⁸ since 2019, with hospitals reporting medication reconciliation rates for financial in addition to quality and safety reasons. Hopefully, this study and others, in combination with the abovementioned incentives, will stimulate further research into impactful strategies for medication reconciliation and ideal ways to implement them. With luck, the end result will be more generalizable interventions, with a track record of success, that would help ensure that patients are prescribed, are reporting, are taking, and are noted to be taking the medications that they and their providers intended, both on presentation to the hospital and on discharge home.

Disclosures

Vicki Jue has no conflicts of interest to report. Raman Khanna reports developing CareWeb, a communication platform that has been licensed to Voalte, Inc. This work is unrelated to the current editorial. No other conflicts of interest to report.

Of all the errors that occur in modern healthcare, medication errors are among the most ubiquitous and consequential. Adverse drug events (ADEs) account for approximately 700,000 emergency department visits, 100,000 hospitalizations, and 1.3 million people are injured by medication errors annually.¹ Among the most frequent causes of preventable ADEs are errors on the medication lists when patients are admitted to hospitals.² Therefore, preventing discrepancies between medications the patient is prescribed (and actually taking) inside and outside the hospital—the so-called “medication reconciliation”—is an intense, ongoing area of focus for health systems, pharmacies, and numerous quality and safety organizations seeking to reduce ADEs.

Past studies of medication reconciliation interventions have suggested benefit from restricting medication reconciliation to admission or discharge, pharmacist or pharmacy technician-led medication reconciliation, and pharmacy-led interventions (ie, telephone follow-up/home visit, patient counseling) for ensuring an accurate medication list.^3-5 Recent evidence suggests that pharmacist discharge medication reconciliation is associated with decreased readmission rates, decreased medication discrepancies, and adverse events associated with drug therapy issues.⁴ The successful interventions were promising, but disseminating such interventions can often be very complex.⁶

In this issue of the Journal of Hospital Medicine, Mixon et al. report the results of a subanalysis of the MARQUIS trial,⁷ wherein they individually examined the on-protocol effects of the interventions that MARQUIS recommended, comparing hospitals to their own running baseline data at the implementation of each intervention to data following the implementation. The authors found that only three of the nine interventions were associated with reducing potentially harmful discrepancies in the medication list—training existing staff to perform discharge medication reconciliation, hiring additional staff for this purpose, and defining roles and responsibilities and roles clearly—and that two were actually associated with harm—training existing staff to take best possible medication histories (BPMHs) and implementing a new Electronic Medical Record (EMR). MARQUIS is unique in not just attempting but in reporting “best case” real-world implementation using available literature to design mentored, practical approaches to those same interventions at sites not involved in their initial setup and validation.

EMR implementation should in theory improve accuracy (or at least legibility), but it can also contribute to new types of inaccuracy or, as the authors propose, deprioritize quality and safety as organizational goals during the rigors of digitization. Similarly, training staff to take a BPMH might create false confidence in the results or interact with medication reconciliation in other complex ways. Opting to add more work instead of hiring additional staff may have increased the burden of medication review and thus contributed to its inaccuracy.

On the contrary, certain interventions, such as having clear accountability for the medication list, hiring additional staff to construct that list, and clearly defining the roles of those involved in the reconciliation process, were associated with improved medication reconciliation. All these strategies require resource allocation, but at least the current study provides evidence that such resource allocation can be effective in new settings as they were in their original ones.

The study has important acknowledged limitations. The on-protocol analysis limited the authors to reporting associations rather than causality. Moreover, the original trial ran from 2011 to 2014, which was a time of rapid EMR implementation and new recognition of the problems posed by the same; several organizations are in a far more mature EMR context today. Conversely, newer technologies such as patient-facing medication reconciliation applications, cross-organization medication lists available from some EMR vendors, and health platforms that collect data from multiple EMRs were not evaluated because they did not exist at the time of the original trial. Another important trend in healthcare, the rise of Accountable Care Organizations and their focus on integration and defragmentation, may have an important part to play in medication list accuracy. All the above-mentioned aspects will be important avenues for ongoing research in real-world medication reconciliation.

Mixon’s findings come at a time when medication reconciliation is again a national health informatics priority, a key component of the Medicare Access and CHIP reauthorization Act of 2015 and Merit-based Incentive Payments System⁸ since 2019, with hospitals reporting medication reconciliation rates for financial in addition to quality and safety reasons. Hopefully, this study and others, in combination with the abovementioned incentives, will stimulate further research into impactful strategies for medication reconciliation and ideal ways to implement them. With luck, the end result will be more generalizable interventions, with a track record of success, that would help ensure that patients are prescribed, are reporting, are taking, and are noted to be taking the medications that they and their providers intended, both on presentation to the hospital and on discharge home.

Disclosures

Vicki Jue has no conflicts of interest to report. Raman Khanna reports developing CareWeb, a communication platform that has been licensed to Voalte, Inc. This work is unrelated to the current editorial. No other conflicts of interest to report.

The 25 million people in the United States with limited English proficiency (LEP), which is defined as speaking English less than “very well”, are at increased risk for healthcare disparities that result in preventable harm and poor patient experiences compared with English-proficient patients.^1,2 The use of trained professional interpreters is associated with improved communication, healthcare outcomes, safety, and experiences for LEP patients.³ However, underuse of professional interpreters remains common.⁴ Healthcare staff frequently use family members, friends, or minor children as interpreters or try to “get by” with the patient’s limited English skills or staff’s limited non-English skills.⁵ These practices regularly compromise patient safety and quality for LEP patients and their families.

In the article “Inpatient Communication Barriers and Drivers when Caring for Limited English Proficiency Children,” Dr. Choe and colleagues approach the problem of interpreter underuse by studying the barriers and facilitators that exist at their children’s hospital.⁶ The group conducted four sessions using Group Level Assessment, a structured, interactive approach to understanding a problem and identifying potential solutions. Sixty-four pediatric hospitalists and residents, bedside nurses, and staff interpreters participated. Participants identified four primary barriers to communicating effectively with LEP families: difficulty accessing interpreter services, uncertainty in communicating with LEP families, unclear roles and expectations of different team members, and unmet expectations related to family engagement. They also identified four drivers of effective communication: collaborative problem-solving between providers and interpreters, greater attention to cultural context, practicing empathy for patients and families, and using family centered communication strategies.

This study reinforces that myriad challenges remain in accessing and using an interpreter. The barriers identified fall into two major categories: systems for accessing interpretation and communication involving an interpreter. Both ultimately must be addressed to achieve equitable communication for LEP patients/families. As interpreter use is contingent upon access, optimizing delivery systems is an essential foundation. At this study site, key barriers were the opaque scheduling processes and inconsistent access to and unfamiliarity with interpreter-related technology (eg, for telephone or video interpretation). These barriers are likely generalizable to many other hospitals. Priority should be given to developing transparent, consistent, and reliable processes for interpreter access. Interventions to improve interpreter access, such as one-touch interpreter telephones at every hospital bedside, have been more successful in improving interpreter use than provider education or regulatory mandates.⁴

The challenges identified around communicating with LEP families via interpreter are also likely generalizable. In the current study, participants described a clear tension around the interpreters’ optimal role, in which the care team might want the interpreter to intervene or participate in the discussion more, while interpreter standards require that they remain a neutral conduit for information. This neutral-party approach, when taken to the extreme, can limit the bidirectional communication between clinical teams and interpreters necessary to address communication challenges. Fostering collaborative problem-solving between interpreters and clinicians, in both formal and informal settings, is critically needed to improve the quality of communication during encounters. In addition to the proposed presession meeting between the clinician and interpreter, incorporating a debriefing after an interpreter-mediated encounter could offer an opportunity for bidirectional feedback. Unfortunately, interpreter scheduling constraints, fueled by the lack of reimbursement for interpretation in most states, frequently limit the feasibility of such proposals.

Participating providers also reported decreased engagement with LEP families and that they spent less time with them. These observations also merit attention if we are to achieve equitable outcomes for LEP patients. A conversation via interpreter requires more time for the same content, given the time needed to interpret the message. The fact that participants reported spending less time with LEP families means that less communication occurs with those families, compared with others. There are well-established links between good communication and improved clinical outcomes, including everything from decreased glycosylated hemoglobin levels to lower inpatient narcotic use.⁷ Thus, it is not surprising that patients with fewer opportunities to communicate fully have worse clinical outcomes.⁸ Addressing this will require changing hospital culture and provider expectations. Healthcare systems could support this effort with interventions such as decreased nursing assignments, longer allocated rounding times, longer outpatient clinic visits, and additional “points” in resident patient caps, if they exist, for LEP patients. Such steps would be an important investment in improving outcomes and decreasing costs for these vulnerable patients.

For all the barriers identified by Choe and colleagues, solutions are needed. Some may be generalizable, some may be location-specific, and most will be somewhere in between, requiring context-specific tailoring. We recommend a quality improvement (QI) approach, as the evidence-based best practice for communicating with LEP patients and families is well-known, but the gap is in delivering care that meets that standard. Leveraging the growing QI expertise at many institutions to devise approaches that go beyond provider education to change the systems and culture around communicating with LEP patients holds our best promise for improving the safety and effectiveness of care for this population.

Disclosures

The authors have no financial relationships relevant to this article to disclose nor do they have any conflicts of interest relevant to this article to disclose.

Funding

Dr. Lion’s time was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, grant K23 HD078507 (PI Lion).

The 25 million people in the United States with limited English proficiency (LEP), which is defined as speaking English less than “very well”, are at increased risk for healthcare disparities that result in preventable harm and poor patient experiences compared with English-proficient patients.^1,2 The use of trained professional interpreters is associated with improved communication, healthcare outcomes, safety, and experiences for LEP patients.³ However, underuse of professional interpreters remains common.⁴ Healthcare staff frequently use family members, friends, or minor children as interpreters or try to “get by” with the patient’s limited English skills or staff’s limited non-English skills.⁵ These practices regularly compromise patient safety and quality for LEP patients and their families.

In the article “Inpatient Communication Barriers and Drivers when Caring for Limited English Proficiency Children,” Dr. Choe and colleagues approach the problem of interpreter underuse by studying the barriers and facilitators that exist at their children’s hospital.⁶ The group conducted four sessions using Group Level Assessment, a structured, interactive approach to understanding a problem and identifying potential solutions. Sixty-four pediatric hospitalists and residents, bedside nurses, and staff interpreters participated. Participants identified four primary barriers to communicating effectively with LEP families: difficulty accessing interpreter services, uncertainty in communicating with LEP families, unclear roles and expectations of different team members, and unmet expectations related to family engagement. They also identified four drivers of effective communication: collaborative problem-solving between providers and interpreters, greater attention to cultural context, practicing empathy for patients and families, and using family centered communication strategies.

This study reinforces that myriad challenges remain in accessing and using an interpreter. The barriers identified fall into two major categories: systems for accessing interpretation and communication involving an interpreter. Both ultimately must be addressed to achieve equitable communication for LEP patients/families. As interpreter use is contingent upon access, optimizing delivery systems is an essential foundation. At this study site, key barriers were the opaque scheduling processes and inconsistent access to and unfamiliarity with interpreter-related technology (eg, for telephone or video interpretation). These barriers are likely generalizable to many other hospitals. Priority should be given to developing transparent, consistent, and reliable processes for interpreter access. Interventions to improve interpreter access, such as one-touch interpreter telephones at every hospital bedside, have been more successful in improving interpreter use than provider education or regulatory mandates.⁴

The challenges identified around communicating with LEP families via interpreter are also likely generalizable. In the current study, participants described a clear tension around the interpreters’ optimal role, in which the care team might want the interpreter to intervene or participate in the discussion more, while interpreter standards require that they remain a neutral conduit for information. This neutral-party approach, when taken to the extreme, can limit the bidirectional communication between clinical teams and interpreters necessary to address communication challenges. Fostering collaborative problem-solving between interpreters and clinicians, in both formal and informal settings, is critically needed to improve the quality of communication during encounters. In addition to the proposed presession meeting between the clinician and interpreter, incorporating a debriefing after an interpreter-mediated encounter could offer an opportunity for bidirectional feedback. Unfortunately, interpreter scheduling constraints, fueled by the lack of reimbursement for interpretation in most states, frequently limit the feasibility of such proposals.

Participating providers also reported decreased engagement with LEP families and that they spent less time with them. These observations also merit attention if we are to achieve equitable outcomes for LEP patients. A conversation via interpreter requires more time for the same content, given the time needed to interpret the message. The fact that participants reported spending less time with LEP families means that less communication occurs with those families, compared with others. There are well-established links between good communication and improved clinical outcomes, including everything from decreased glycosylated hemoglobin levels to lower inpatient narcotic use.⁷ Thus, it is not surprising that patients with fewer opportunities to communicate fully have worse clinical outcomes.⁸ Addressing this will require changing hospital culture and provider expectations. Healthcare systems could support this effort with interventions such as decreased nursing assignments, longer allocated rounding times, longer outpatient clinic visits, and additional “points” in resident patient caps, if they exist, for LEP patients. Such steps would be an important investment in improving outcomes and decreasing costs for these vulnerable patients.

For all the barriers identified by Choe and colleagues, solutions are needed. Some may be generalizable, some may be location-specific, and most will be somewhere in between, requiring context-specific tailoring. We recommend a quality improvement (QI) approach, as the evidence-based best practice for communicating with LEP patients and families is well-known, but the gap is in delivering care that meets that standard. Leveraging the growing QI expertise at many institutions to devise approaches that go beyond provider education to change the systems and culture around communicating with LEP patients holds our best promise for improving the safety and effectiveness of care for this population.

Disclosures

The authors have no financial relationships relevant to this article to disclose nor do they have any conflicts of interest relevant to this article to disclose.

Funding

Dr. Lion’s time was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, grant K23 HD078507 (PI Lion).

The 25 million people in the United States with limited English proficiency (LEP), which is defined as speaking English less than “very well”, are at increased risk for healthcare disparities that result in preventable harm and poor patient experiences compared with English-proficient patients.^1,2 The use of trained professional interpreters is associated with improved communication, healthcare outcomes, safety, and experiences for LEP patients.³ However, underuse of professional interpreters remains common.⁴ Healthcare staff frequently use family members, friends, or minor children as interpreters or try to “get by” with the patient’s limited English skills or staff’s limited non-English skills.⁵ These practices regularly compromise patient safety and quality for LEP patients and their families.

In the article “Inpatient Communication Barriers and Drivers when Caring for Limited English Proficiency Children,” Dr. Choe and colleagues approach the problem of interpreter underuse by studying the barriers and facilitators that exist at their children’s hospital.⁶ The group conducted four sessions using Group Level Assessment, a structured, interactive approach to understanding a problem and identifying potential solutions. Sixty-four pediatric hospitalists and residents, bedside nurses, and staff interpreters participated. Participants identified four primary barriers to communicating effectively with LEP families: difficulty accessing interpreter services, uncertainty in communicating with LEP families, unclear roles and expectations of different team members, and unmet expectations related to family engagement. They also identified four drivers of effective communication: collaborative problem-solving between providers and interpreters, greater attention to cultural context, practicing empathy for patients and families, and using family centered communication strategies.

This study reinforces that myriad challenges remain in accessing and using an interpreter. The barriers identified fall into two major categories: systems for accessing interpretation and communication involving an interpreter. Both ultimately must be addressed to achieve equitable communication for LEP patients/families. As interpreter use is contingent upon access, optimizing delivery systems is an essential foundation. At this study site, key barriers were the opaque scheduling processes and inconsistent access to and unfamiliarity with interpreter-related technology (eg, for telephone or video interpretation). These barriers are likely generalizable to many other hospitals. Priority should be given to developing transparent, consistent, and reliable processes for interpreter access. Interventions to improve interpreter access, such as one-touch interpreter telephones at every hospital bedside, have been more successful in improving interpreter use than provider education or regulatory mandates.⁴

The challenges identified around communicating with LEP families via interpreter are also likely generalizable. In the current study, participants described a clear tension around the interpreters’ optimal role, in which the care team might want the interpreter to intervene or participate in the discussion more, while interpreter standards require that they remain a neutral conduit for information. This neutral-party approach, when taken to the extreme, can limit the bidirectional communication between clinical teams and interpreters necessary to address communication challenges. Fostering collaborative problem-solving between interpreters and clinicians, in both formal and informal settings, is critically needed to improve the quality of communication during encounters. In addition to the proposed presession meeting between the clinician and interpreter, incorporating a debriefing after an interpreter-mediated encounter could offer an opportunity for bidirectional feedback. Unfortunately, interpreter scheduling constraints, fueled by the lack of reimbursement for interpretation in most states, frequently limit the feasibility of such proposals.

Participating providers also reported decreased engagement with LEP families and that they spent less time with them. These observations also merit attention if we are to achieve equitable outcomes for LEP patients. A conversation via interpreter requires more time for the same content, given the time needed to interpret the message. The fact that participants reported spending less time with LEP families means that less communication occurs with those families, compared with others. There are well-established links between good communication and improved clinical outcomes, including everything from decreased glycosylated hemoglobin levels to lower inpatient narcotic use.⁷ Thus, it is not surprising that patients with fewer opportunities to communicate fully have worse clinical outcomes.⁸ Addressing this will require changing hospital culture and provider expectations. Healthcare systems could support this effort with interventions such as decreased nursing assignments, longer allocated rounding times, longer outpatient clinic visits, and additional “points” in resident patient caps, if they exist, for LEP patients. Such steps would be an important investment in improving outcomes and decreasing costs for these vulnerable patients.

For all the barriers identified by Choe and colleagues, solutions are needed. Some may be generalizable, some may be location-specific, and most will be somewhere in between, requiring context-specific tailoring. We recommend a quality improvement (QI) approach, as the evidence-based best practice for communicating with LEP patients and families is well-known, but the gap is in delivering care that meets that standard. Leveraging the growing QI expertise at many institutions to devise approaches that go beyond provider education to change the systems and culture around communicating with LEP patients holds our best promise for improving the safety and effectiveness of care for this population.

Disclosures

The authors have no financial relationships relevant to this article to disclose nor do they have any conflicts of interest relevant to this article to disclose.

Funding

Dr. Lion’s time was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, grant K23 HD078507 (PI Lion).

In the past five years, it has become increasingly apparent that hospital physiologic monitoring systems are not functioning optimally for children. On pediatric wards, 26%-48% of children are continuously monitored, and these children generate between 42 and 155 alarms per day.¹ Just 1% or fewer are considered actionable or informative, slowing nurses’ response times and placing patients at risk of delayed recognition of life-threatening events.^2,3 While some factors associated with alarm response times have been elucidated,³ in order to design safe and effective monitoring systems, further work is needed to understand the complex decision-making process that nurses face when encountering alarms outside a patient’s room. It is in this area that Schondelmeyer and colleagues strive to enhance our understanding in this issue of the Journal of Hospital Medicine.⁴

Schondelmeyer et al. conducted a single-center, observational study using mixed methods in a general pediatric unit. Trained observers shadowed nine nurses one to four times each, during which nurses were asked to “think aloud” as they managed physiologic monitor alarms, rationalizing their decisions about how and why they might respond for the observer to document. Observers accumulated 61 patient-hours of observation before investigators halted data collection because new insights about alarm responses were no longer emerging from the data (thematic saturation).

Nurses thought aloud about 207 alarms during the study, which the investigators estimated comprised about one third of the alarms that occurred during observation periods. Most of the 207 occurred while the nurse was already in the patient’s room, where a response decision is uncomplicated. More interesting were the 45 alarms heard while outside the patient’s room, where nurses face the complex decision of whether to interrupt their current tasks and respond or delay their response and assume the associated risk of nonresponse to a potentially deteriorating patient. Of the 45 alarms, nurses went into the room to evaluate the patient 15 times and, after doing so, reported that five of the 15 warranted in-person responses to address technical issues with the monitor, clinical issues, or patients’ comfort. Reassuring clinical contexts—such as presence of the medical team or family in the room and recent patient assessments—were the reasons most commonly provided to explain alarm nonresponse.

This study has two key limitations. First, the authors designed the study to observe nurses’ responses until thematic saturation was achieved. However, the small sample size (nine nurses, 45 out-of-room alarms) could raise questions about whether sufficient data were captured to make broadly generalizable conclusions, given the diverse range of patients, families, and clinical scenarios nurses encounter on an inpatient unit. Second, by instructing nurse participants to verbalize their rationale for response or nonresponse, investigators essentially asked nurses to override the “Type 1”, heuristic-based reasoning⁵ that research suggests regulates nursing responses to alarms when adapting to circumstances requiring high cognitive demand or a heavy workload.³ While innovative, it is possible that this approach prevented the investigators from fully achieving their stated objective of describing how bedside nurses think about and act upon alarms.

Nonetheless, the findings by Schondelmeyer and colleagues extend our emerging understanding of why alarm responses are disconcertingly slow. Nursing staff’s dismissal of monitor alarms that are discordant with a reassuring patient evaluation underscores the imperative to reduce nuisance alarms. Furthermore, the explicit statements justifying alarm nonresponse because of the presence of family members build upon prior findings of longer response times when family members are at the bedside³ and invite a provocative question: how would family members feel if they knew that they were being entrusted as a foundational component of safety monitoring in the hospital? In their recently published study conducted at the same hospital,⁶ Schondelmeyer’s team elicited perceptions that families are deeply concerned about staff nonresponse to alarms—as one nurse stated, parents “wonder what’s going on when no one comes in.” While there is a valuable role for integrating families into efforts to overcome threats to patient safety, as has been achieved with family error reporting⁷ and communication on family-centered rounds,⁸ this must occur in a structured, explicit, and deliberate manner, with families engaged as key stakeholders.

In summary, while Schondelmeyer and colleagues may not have exposed the depth of implicit thinking that governs nurses’ responses to alarms, they have highlighted the high-stakes decisions that nurses confront on a daily basis in an environment with exceedingly high alarm rates and low alarm actionability. The authors cite staff education among potential solutions to improve the safety of continuous monitoring, but such an intervention cannot be effective in a system that places impossible burdens on nurses. An openly family centered and multidisciplinary approach to reengineering the system for monitoring hospitalized children is needed to enable nurses to respond quickly and accurately to patients at risk of clinical deterioration.

Disclosures

The authors report no conflicts of interest.

In the past five years, it has become increasingly apparent that hospital physiologic monitoring systems are not functioning optimally for children. On pediatric wards, 26%-48% of children are continuously monitored, and these children generate between 42 and 155 alarms per day.¹ Just 1% or fewer are considered actionable or informative, slowing nurses’ response times and placing patients at risk of delayed recognition of life-threatening events.^2,3 While some factors associated with alarm response times have been elucidated,³ in order to design safe and effective monitoring systems, further work is needed to understand the complex decision-making process that nurses face when encountering alarms outside a patient’s room. It is in this area that Schondelmeyer and colleagues strive to enhance our understanding in this issue of the Journal of Hospital Medicine.⁴

Schondelmeyer et al. conducted a single-center, observational study using mixed methods in a general pediatric unit. Trained observers shadowed nine nurses one to four times each, during which nurses were asked to “think aloud” as they managed physiologic monitor alarms, rationalizing their decisions about how and why they might respond for the observer to document. Observers accumulated 61 patient-hours of observation before investigators halted data collection because new insights about alarm responses were no longer emerging from the data (thematic saturation).

Nurses thought aloud about 207 alarms during the study, which the investigators estimated comprised about one third of the alarms that occurred during observation periods. Most of the 207 occurred while the nurse was already in the patient’s room, where a response decision is uncomplicated. More interesting were the 45 alarms heard while outside the patient’s room, where nurses face the complex decision of whether to interrupt their current tasks and respond or delay their response and assume the associated risk of nonresponse to a potentially deteriorating patient. Of the 45 alarms, nurses went into the room to evaluate the patient 15 times and, after doing so, reported that five of the 15 warranted in-person responses to address technical issues with the monitor, clinical issues, or patients’ comfort. Reassuring clinical contexts—such as presence of the medical team or family in the room and recent patient assessments—were the reasons most commonly provided to explain alarm nonresponse.

This study has two key limitations. First, the authors designed the study to observe nurses’ responses until thematic saturation was achieved. However, the small sample size (nine nurses, 45 out-of-room alarms) could raise questions about whether sufficient data were captured to make broadly generalizable conclusions, given the diverse range of patients, families, and clinical scenarios nurses encounter on an inpatient unit. Second, by instructing nurse participants to verbalize their rationale for response or nonresponse, investigators essentially asked nurses to override the “Type 1”, heuristic-based reasoning⁵ that research suggests regulates nursing responses to alarms when adapting to circumstances requiring high cognitive demand or a heavy workload.³ While innovative, it is possible that this approach prevented the investigators from fully achieving their stated objective of describing how bedside nurses think about and act upon alarms.

Nonetheless, the findings by Schondelmeyer and colleagues extend our emerging understanding of why alarm responses are disconcertingly slow. Nursing staff’s dismissal of monitor alarms that are discordant with a reassuring patient evaluation underscores the imperative to reduce nuisance alarms. Furthermore, the explicit statements justifying alarm nonresponse because of the presence of family members build upon prior findings of longer response times when family members are at the bedside³ and invite a provocative question: how would family members feel if they knew that they were being entrusted as a foundational component of safety monitoring in the hospital? In their recently published study conducted at the same hospital,⁶ Schondelmeyer’s team elicited perceptions that families are deeply concerned about staff nonresponse to alarms—as one nurse stated, parents “wonder what’s going on when no one comes in.” While there is a valuable role for integrating families into efforts to overcome threats to patient safety, as has been achieved with family error reporting⁷ and communication on family-centered rounds,⁸ this must occur in a structured, explicit, and deliberate manner, with families engaged as key stakeholders.

In summary, while Schondelmeyer and colleagues may not have exposed the depth of implicit thinking that governs nurses’ responses to alarms, they have highlighted the high-stakes decisions that nurses confront on a daily basis in an environment with exceedingly high alarm rates and low alarm actionability. The authors cite staff education among potential solutions to improve the safety of continuous monitoring, but such an intervention cannot be effective in a system that places impossible burdens on nurses. An openly family centered and multidisciplinary approach to reengineering the system for monitoring hospitalized children is needed to enable nurses to respond quickly and accurately to patients at risk of clinical deterioration.

Disclosures

The authors report no conflicts of interest.

In the past five years, it has become increasingly apparent that hospital physiologic monitoring systems are not functioning optimally for children. On pediatric wards, 26%-48% of children are continuously monitored, and these children generate between 42 and 155 alarms per day.¹ Just 1% or fewer are considered actionable or informative, slowing nurses’ response times and placing patients at risk of delayed recognition of life-threatening events.^2,3 While some factors associated with alarm response times have been elucidated,³ in order to design safe and effective monitoring systems, further work is needed to understand the complex decision-making process that nurses face when encountering alarms outside a patient’s room. It is in this area that Schondelmeyer and colleagues strive to enhance our understanding in this issue of the Journal of Hospital Medicine.⁴

Schondelmeyer et al. conducted a single-center, observational study using mixed methods in a general pediatric unit. Trained observers shadowed nine nurses one to four times each, during which nurses were asked to “think aloud” as they managed physiologic monitor alarms, rationalizing their decisions about how and why they might respond for the observer to document. Observers accumulated 61 patient-hours of observation before investigators halted data collection because new insights about alarm responses were no longer emerging from the data (thematic saturation).

Nurses thought aloud about 207 alarms during the study, which the investigators estimated comprised about one third of the alarms that occurred during observation periods. Most of the 207 occurred while the nurse was already in the patient’s room, where a response decision is uncomplicated. More interesting were the 45 alarms heard while outside the patient’s room, where nurses face the complex decision of whether to interrupt their current tasks and respond or delay their response and assume the associated risk of nonresponse to a potentially deteriorating patient. Of the 45 alarms, nurses went into the room to evaluate the patient 15 times and, after doing so, reported that five of the 15 warranted in-person responses to address technical issues with the monitor, clinical issues, or patients’ comfort. Reassuring clinical contexts—such as presence of the medical team or family in the room and recent patient assessments—were the reasons most commonly provided to explain alarm nonresponse.

This study has two key limitations. First, the authors designed the study to observe nurses’ responses until thematic saturation was achieved. However, the small sample size (nine nurses, 45 out-of-room alarms) could raise questions about whether sufficient data were captured to make broadly generalizable conclusions, given the diverse range of patients, families, and clinical scenarios nurses encounter on an inpatient unit. Second, by instructing nurse participants to verbalize their rationale for response or nonresponse, investigators essentially asked nurses to override the “Type 1”, heuristic-based reasoning⁵ that research suggests regulates nursing responses to alarms when adapting to circumstances requiring high cognitive demand or a heavy workload.³ While innovative, it is possible that this approach prevented the investigators from fully achieving their stated objective of describing how bedside nurses think about and act upon alarms.

Nonetheless, the findings by Schondelmeyer and colleagues extend our emerging understanding of why alarm responses are disconcertingly slow. Nursing staff’s dismissal of monitor alarms that are discordant with a reassuring patient evaluation underscores the imperative to reduce nuisance alarms. Furthermore, the explicit statements justifying alarm nonresponse because of the presence of family members build upon prior findings of longer response times when family members are at the bedside³ and invite a provocative question: how would family members feel if they knew that they were being entrusted as a foundational component of safety monitoring in the hospital? In their recently published study conducted at the same hospital,⁶ Schondelmeyer’s team elicited perceptions that families are deeply concerned about staff nonresponse to alarms—as one nurse stated, parents “wonder what’s going on when no one comes in.” While there is a valuable role for integrating families into efforts to overcome threats to patient safety, as has been achieved with family error reporting⁷ and communication on family-centered rounds,⁸ this must occur in a structured, explicit, and deliberate manner, with families engaged as key stakeholders.

In summary, while Schondelmeyer and colleagues may not have exposed the depth of implicit thinking that governs nurses’ responses to alarms, they have highlighted the high-stakes decisions that nurses confront on a daily basis in an environment with exceedingly high alarm rates and low alarm actionability. The authors cite staff education among potential solutions to improve the safety of continuous monitoring, but such an intervention cannot be effective in a system that places impossible burdens on nurses. An openly family centered and multidisciplinary approach to reengineering the system for monitoring hospitalized children is needed to enable nurses to respond quickly and accurately to patients at risk of clinical deterioration.

Disclosures

The authors report no conflicts of interest.