Farah Acher Kaiksow, MD, MPP

Earlier this year, the Journal of Hospital Medicine updated its author guidelines to include recommendations on addressing race and racism.¹ These recommendations include explicitly naming racism (rather than race) as a determinant of health. Operationalizing these recommendations into manuscripts represents a fundamental shift in how we ask research questions, structure analyses, and interpret results.

In this issue, Maxwell et al² illustrate how to disseminate research through this lens in their retrospective cohort study of children with type 1 diabetes hospitalized with diabetic ketoacidosis (DKA). Using 6 years of data from a major academic pediatric medical center, the authors examine the association between risk for DKA admission and three factors: neighborhood poverty level, race, and type of insurance (public or private). Secondary outcomes include DKA severity and length of stay. In their unadjusted model, poverty, race, and insurance were all associated with increased hospitalizations. However, following adjustment, the association between race and hospitalizations disappeared.In line with the journal’s new guidelines, the authors point out that the statistically significant associations of poverty and insurance type with clinical outcomes suggest that racism, rather than race, is a social factor at work in their population. The authors provide further context regarding structural racism in the United States and the history of redlining, which has helped shape a society in which Black individuals are more likely to live in areas of concentrated poverty and be publicly insured.

Two other findings related to the impact of racism are notable. First, in both their univariate and multivariate models, the authors found significant A1c differences between Black and White children—higher than those of previous reports.³ These findings suggest the existence of structural factors at work in the health of their patients. Second, Black patients had longer lengths of stay when compared to White patients with the same severity of DKA. Neither poverty level nor insurance status were significantly associated with length of stay. While the analysis was limited to detecting this difference, rather than identifying its causes, the authors suggest factors at both individual and structural levels that may be impacting outcomes. Specifically, care team bias may impact discharge decisions, and factors such as less flexible times to complete diabetes education, transportation barriers, and childcare challenges could also impact discharge timing.

This work provides a template for how to address the impact of racism on health with intentionality. Moreover, individuals’ lived environments should be considered through alternative economic measurements and neighborhood definitions. The proportion of people within a census tract living below the federal poverty line is just one measure of the complex dynamics that contribute to an individual’s socioeconomic status. An alternative measure is the area deprivation index, which incorporates 17 indicators at the more granular census block group level to describe an individual’s environment⁴ and could be useful in this area of research.

Perhaps most relevant is the use of public insurance as a marker of socioeconomic status. Medicaid, although not without its flaws, provides fairly comprehensive coverage. However, many Americans have incomes too high to qualify for public insurance but too low to afford adequate insurance coverage. Theoretically, these individuals qualify for subsidies through the Affordable Care Act, yet underinsurance remains a significant issue.⁵ Future analyses to further understand and describe clinical outcomes could include this population of underinsured children as a distinct at-risk group. Maxwell et al² provide an excellent example of how we should address race and racism in disseminated literature. Although initially challenging, writing with intentionality regarding this fundamental determinant of health can provide rich and actionable information for practitioners and policy-makers.

Earlier this year, the Journal of Hospital Medicine updated its author guidelines to include recommendations on addressing race and racism.¹ These recommendations include explicitly naming racism (rather than race) as a determinant of health. Operationalizing these recommendations into manuscripts represents a fundamental shift in how we ask research questions, structure analyses, and interpret results.

In this issue, Maxwell et al² illustrate how to disseminate research through this lens in their retrospective cohort study of children with type 1 diabetes hospitalized with diabetic ketoacidosis (DKA). Using 6 years of data from a major academic pediatric medical center, the authors examine the association between risk for DKA admission and three factors: neighborhood poverty level, race, and type of insurance (public or private). Secondary outcomes include DKA severity and length of stay. In their unadjusted model, poverty, race, and insurance were all associated with increased hospitalizations. However, following adjustment, the association between race and hospitalizations disappeared.In line with the journal’s new guidelines, the authors point out that the statistically significant associations of poverty and insurance type with clinical outcomes suggest that racism, rather than race, is a social factor at work in their population. The authors provide further context regarding structural racism in the United States and the history of redlining, which has helped shape a society in which Black individuals are more likely to live in areas of concentrated poverty and be publicly insured.

Two other findings related to the impact of racism are notable. First, in both their univariate and multivariate models, the authors found significant A1c differences between Black and White children—higher than those of previous reports.³ These findings suggest the existence of structural factors at work in the health of their patients. Second, Black patients had longer lengths of stay when compared to White patients with the same severity of DKA. Neither poverty level nor insurance status were significantly associated with length of stay. While the analysis was limited to detecting this difference, rather than identifying its causes, the authors suggest factors at both individual and structural levels that may be impacting outcomes. Specifically, care team bias may impact discharge decisions, and factors such as less flexible times to complete diabetes education, transportation barriers, and childcare challenges could also impact discharge timing.

This work provides a template for how to address the impact of racism on health with intentionality. Moreover, individuals’ lived environments should be considered through alternative economic measurements and neighborhood definitions. The proportion of people within a census tract living below the federal poverty line is just one measure of the complex dynamics that contribute to an individual’s socioeconomic status. An alternative measure is the area deprivation index, which incorporates 17 indicators at the more granular census block group level to describe an individual’s environment⁴ and could be useful in this area of research.

Perhaps most relevant is the use of public insurance as a marker of socioeconomic status. Medicaid, although not without its flaws, provides fairly comprehensive coverage. However, many Americans have incomes too high to qualify for public insurance but too low to afford adequate insurance coverage. Theoretically, these individuals qualify for subsidies through the Affordable Care Act, yet underinsurance remains a significant issue.⁵ Future analyses to further understand and describe clinical outcomes could include this population of underinsured children as a distinct at-risk group. Maxwell et al² provide an excellent example of how we should address race and racism in disseminated literature. Although initially challenging, writing with intentionality regarding this fundamental determinant of health can provide rich and actionable information for practitioners and policy-makers.

Earlier this year, the Journal of Hospital Medicine updated its author guidelines to include recommendations on addressing race and racism.¹ These recommendations include explicitly naming racism (rather than race) as a determinant of health. Operationalizing these recommendations into manuscripts represents a fundamental shift in how we ask research questions, structure analyses, and interpret results.

In this issue, Maxwell et al² illustrate how to disseminate research through this lens in their retrospective cohort study of children with type 1 diabetes hospitalized with diabetic ketoacidosis (DKA). Using 6 years of data from a major academic pediatric medical center, the authors examine the association between risk for DKA admission and three factors: neighborhood poverty level, race, and type of insurance (public or private). Secondary outcomes include DKA severity and length of stay. In their unadjusted model, poverty, race, and insurance were all associated with increased hospitalizations. However, following adjustment, the association between race and hospitalizations disappeared.In line with the journal’s new guidelines, the authors point out that the statistically significant associations of poverty and insurance type with clinical outcomes suggest that racism, rather than race, is a social factor at work in their population. The authors provide further context regarding structural racism in the United States and the history of redlining, which has helped shape a society in which Black individuals are more likely to live in areas of concentrated poverty and be publicly insured.

Two other findings related to the impact of racism are notable. First, in both their univariate and multivariate models, the authors found significant A1c differences between Black and White children—higher than those of previous reports.³ These findings suggest the existence of structural factors at work in the health of their patients. Second, Black patients had longer lengths of stay when compared to White patients with the same severity of DKA. Neither poverty level nor insurance status were significantly associated with length of stay. While the analysis was limited to detecting this difference, rather than identifying its causes, the authors suggest factors at both individual and structural levels that may be impacting outcomes. Specifically, care team bias may impact discharge decisions, and factors such as less flexible times to complete diabetes education, transportation barriers, and childcare challenges could also impact discharge timing.

This work provides a template for how to address the impact of racism on health with intentionality. Moreover, individuals’ lived environments should be considered through alternative economic measurements and neighborhood definitions. The proportion of people within a census tract living below the federal poverty line is just one measure of the complex dynamics that contribute to an individual’s socioeconomic status. An alternative measure is the area deprivation index, which incorporates 17 indicators at the more granular census block group level to describe an individual’s environment⁴ and could be useful in this area of research.

Perhaps most relevant is the use of public insurance as a marker of socioeconomic status. Medicaid, although not without its flaws, provides fairly comprehensive coverage. However, many Americans have incomes too high to qualify for public insurance but too low to afford adequate insurance coverage. Theoretically, these individuals qualify for subsidies through the Affordable Care Act, yet underinsurance remains a significant issue.⁵ Future analyses to further understand and describe clinical outcomes could include this population of underinsured children as a distinct at-risk group. Maxwell et al² provide an excellent example of how we should address race and racism in disseminated literature. Although initially challenging, writing with intentionality regarding this fundamental determinant of health can provide rich and actionable information for practitioners and policy-makers.

The early days of the coronavirus disease 2019 (COVID-19) pandemic were fraught with uncertainty as hospitalists struggled to develop standards of care for noncritically ill patients. Although data were available from intensive care units (ICUs) in Asia and Europe, it was unclear whether these findings applied to the acute but noncritically ill patients who would ultimately make up most coronavirus admissions. Which therapeutics could benefit these patients? Who needs continuous cardiopulmonary monitoring? And perhaps most importantly, which patients are at risk for clinical deterioration?

In this issue, Nemer et al begin to answer these questions using a retrospective analysis of 350 noncritically ill COVID-19 patients admitted to non-ICU care at Cleveland Clinic hospitals in Ohio and Florida between March 13 and May 1, 2020.¹ The primary outcome was a composite of three endpoints: increased respiratory support (high-flow nasal cannula, noninvasive positive pressure ventilation, or intubation), ICU transfer, or death. The primary outcome occurred in 18% of all patients and the risk was greatest among patients with high admission levels of C-reactive protein (CRP). This analysis found that while clinically significant arrhythmias occurred in 14% of patients, 90% of those were in patients with either known cardiac disease or an elevated admission troponin T level and in only one case (<1%) necessitated transition to a higher level of care. Overall mortality for COVID-19 patients initially admitted to non-ICU settings was 3%.

While several tests have been proposed as clinically relevant to coronavirus disease, those recommendations are based on studies performed on critically ill patients outside of the US and have focused on survival, not clinical deterioration.^2,3 In their cohort of noncritically ill patients in the US, Nemer et al found that not only is CRP associated with clinical worsening, but that increasing levels of CRP are associated with increasing risk of deterioration. Perhaps even more interesting was the finding that no patient with a normal CRP suffered the composite outcome, including death. The authors did not report levels of other laboratory tests that have been associated with severe coronavirus disease, such as platelets, fibrin degradation products, or prolonged prothrombin time/activated partial thromboplastin time. As many clinicians will note, CRP’s lack of specificity may be its Achilles heel, potentially lowering its prognostic value. Still, given its wide availability, low cost, and rapid turnaround, CRP could serve as a screening tool to risk stratify admitted coronavirus patients, while also providing reassurance when it is normal.

The results of this study could also impact use of hospital resources. The findings regarding the low risk of arrhythmias provide support for limiting the use of continuous cardiac monitoring in noncritically ill patients without previous histories of cardiac disease or elevated admission troponin levels. Patients with normal admission CRP levels could potentially be monitored safely with intermittent pulse oximetry. Continuous pulse oximetry and cardiac monitoring are already overused in many hospitals, and in the case of coronavirus the implications are even more significant given the importance of minimizing unnecessary healthcare worker exposures.

The vast majority (79% to 90%) of patients hospitalized for coronavirus will be cared for in non–ICU settings,^4,5 yet most research has thus far focused on ICU patients. Nemer et al provide much-needed information on how to care for the noncritically ill coronavirus patients whom hospitalists are most likely to treat. As a resurgence of infections is expected this winter, this work has the potential to help physicians identify not only those who have the highest probability of deteriorating, but also those who may not. In a world of limited resources, knowing which patient is unlikely to deteriorate may be just as important as recognizing which one is.

The early days of the coronavirus disease 2019 (COVID-19) pandemic were fraught with uncertainty as hospitalists struggled to develop standards of care for noncritically ill patients. Although data were available from intensive care units (ICUs) in Asia and Europe, it was unclear whether these findings applied to the acute but noncritically ill patients who would ultimately make up most coronavirus admissions. Which therapeutics could benefit these patients? Who needs continuous cardiopulmonary monitoring? And perhaps most importantly, which patients are at risk for clinical deterioration?

In this issue, Nemer et al begin to answer these questions using a retrospective analysis of 350 noncritically ill COVID-19 patients admitted to non-ICU care at Cleveland Clinic hospitals in Ohio and Florida between March 13 and May 1, 2020.¹ The primary outcome was a composite of three endpoints: increased respiratory support (high-flow nasal cannula, noninvasive positive pressure ventilation, or intubation), ICU transfer, or death. The primary outcome occurred in 18% of all patients and the risk was greatest among patients with high admission levels of C-reactive protein (CRP). This analysis found that while clinically significant arrhythmias occurred in 14% of patients, 90% of those were in patients with either known cardiac disease or an elevated admission troponin T level and in only one case (<1%) necessitated transition to a higher level of care. Overall mortality for COVID-19 patients initially admitted to non-ICU settings was 3%.

While several tests have been proposed as clinically relevant to coronavirus disease, those recommendations are based on studies performed on critically ill patients outside of the US and have focused on survival, not clinical deterioration.^2,3 In their cohort of noncritically ill patients in the US, Nemer et al found that not only is CRP associated with clinical worsening, but that increasing levels of CRP are associated with increasing risk of deterioration. Perhaps even more interesting was the finding that no patient with a normal CRP suffered the composite outcome, including death. The authors did not report levels of other laboratory tests that have been associated with severe coronavirus disease, such as platelets, fibrin degradation products, or prolonged prothrombin time/activated partial thromboplastin time. As many clinicians will note, CRP’s lack of specificity may be its Achilles heel, potentially lowering its prognostic value. Still, given its wide availability, low cost, and rapid turnaround, CRP could serve as a screening tool to risk stratify admitted coronavirus patients, while also providing reassurance when it is normal.

The results of this study could also impact use of hospital resources. The findings regarding the low risk of arrhythmias provide support for limiting the use of continuous cardiac monitoring in noncritically ill patients without previous histories of cardiac disease or elevated admission troponin levels. Patients with normal admission CRP levels could potentially be monitored safely with intermittent pulse oximetry. Continuous pulse oximetry and cardiac monitoring are already overused in many hospitals, and in the case of coronavirus the implications are even more significant given the importance of minimizing unnecessary healthcare worker exposures.

The vast majority (79% to 90%) of patients hospitalized for coronavirus will be cared for in non–ICU settings,^4,5 yet most research has thus far focused on ICU patients. Nemer et al provide much-needed information on how to care for the noncritically ill coronavirus patients whom hospitalists are most likely to treat. As a resurgence of infections is expected this winter, this work has the potential to help physicians identify not only those who have the highest probability of deteriorating, but also those who may not. In a world of limited resources, knowing which patient is unlikely to deteriorate may be just as important as recognizing which one is.

The early days of the coronavirus disease 2019 (COVID-19) pandemic were fraught with uncertainty as hospitalists struggled to develop standards of care for noncritically ill patients. Although data were available from intensive care units (ICUs) in Asia and Europe, it was unclear whether these findings applied to the acute but noncritically ill patients who would ultimately make up most coronavirus admissions. Which therapeutics could benefit these patients? Who needs continuous cardiopulmonary monitoring? And perhaps most importantly, which patients are at risk for clinical deterioration?

In this issue, Nemer et al begin to answer these questions using a retrospective analysis of 350 noncritically ill COVID-19 patients admitted to non-ICU care at Cleveland Clinic hospitals in Ohio and Florida between March 13 and May 1, 2020.¹ The primary outcome was a composite of three endpoints: increased respiratory support (high-flow nasal cannula, noninvasive positive pressure ventilation, or intubation), ICU transfer, or death. The primary outcome occurred in 18% of all patients and the risk was greatest among patients with high admission levels of C-reactive protein (CRP). This analysis found that while clinically significant arrhythmias occurred in 14% of patients, 90% of those were in patients with either known cardiac disease or an elevated admission troponin T level and in only one case (<1%) necessitated transition to a higher level of care. Overall mortality for COVID-19 patients initially admitted to non-ICU settings was 3%.

While several tests have been proposed as clinically relevant to coronavirus disease, those recommendations are based on studies performed on critically ill patients outside of the US and have focused on survival, not clinical deterioration.^2,3 In their cohort of noncritically ill patients in the US, Nemer et al found that not only is CRP associated with clinical worsening, but that increasing levels of CRP are associated with increasing risk of deterioration. Perhaps even more interesting was the finding that no patient with a normal CRP suffered the composite outcome, including death. The authors did not report levels of other laboratory tests that have been associated with severe coronavirus disease, such as platelets, fibrin degradation products, or prolonged prothrombin time/activated partial thromboplastin time. As many clinicians will note, CRP’s lack of specificity may be its Achilles heel, potentially lowering its prognostic value. Still, given its wide availability, low cost, and rapid turnaround, CRP could serve as a screening tool to risk stratify admitted coronavirus patients, while also providing reassurance when it is normal.

The results of this study could also impact use of hospital resources. The findings regarding the low risk of arrhythmias provide support for limiting the use of continuous cardiac monitoring in noncritically ill patients without previous histories of cardiac disease or elevated admission troponin levels. Patients with normal admission CRP levels could potentially be monitored safely with intermittent pulse oximetry. Continuous pulse oximetry and cardiac monitoring are already overused in many hospitals, and in the case of coronavirus the implications are even more significant given the importance of minimizing unnecessary healthcare worker exposures.

The vast majority (79% to 90%) of patients hospitalized for coronavirus will be cared for in non–ICU settings,^4,5 yet most research has thus far focused on ICU patients. Nemer et al provide much-needed information on how to care for the noncritically ill coronavirus patients whom hospitalists are most likely to treat. As a resurgence of infections is expected this winter, this work has the potential to help physicians identify not only those who have the highest probability of deteriorating, but also those who may not. In a world of limited resources, knowing which patient is unlikely to deteriorate may be just as important as recognizing which one is.

A 73-year-old man presents to the emergency department with sepsis secondary to community-acquired pneumonia. The patient requires supplemental oxygen and is started on intravenous antibiotics. His admitting physician expects he will need more than two nights of hospital care and suggests that inpatient status, rather than outpatient (observation) status, would be appropriate under Medicare’s “Two-Midnight Rule.” The physician also suspects the patient may need a brief stay in a skilled nursing facility (SNF) following the mentioned hospitalization and notes that the patient has a Medicare Advantage plan (Table) and wonders if the Two-Midnight Rule applies. Further, she questions whether Medicare’s “Three-Midnight Rule” for SNF benefits will factor in the patient’s discharge planning.

Since the 1970s, the Centers for Medicare & Medicaid Services (CMS) has allowed enrollees to receive their Medicare benefits from privately managed health plans through the so-called Medicare Advantage programs. CMS contracts with commercial insurers who, in exchange for a set payment per Medicare enrollee, “accept full responsibility (ie, risk) for the costs of their enrollees’ care.”¹ Over the past 20 years the percent of Medicare Advantage enrollees has nearly doubled nationwide, from 18% to 34%, and is projected to grow even further to 42% by 2028.^2,3 The reasons beneficiaries choose to enroll in Medicare Advantage over Traditional Medicare have yet to be thoroughly studied; ease of enrollment and plan administration, as well as lower deductibles, copays, and out-of-pocket maximums for in-network services, are thought to be some of the driving factors.

The federal government has asserted two goals for the development of Medicare Advantage: beneficiary choice and economic efficiency.¹ Medicare Advantage plans must be actuarially equal to Traditional Medicare but do not have to cover services in precisely the same way. Medicare Advantage plans may achieve cost savings through narrower networks, strict control of access to SNF services and acute care inpatient rehabilitation, and prior authorization requirements, the latter of which has received recent congressional attention.^4,5 On the other hand, many Medicare Advantage plans offer dental, fitness, optical, and caregiver benefits that are not included under Traditional Medicare. Beneficiaries can theoretically compare the coverage and costs of Traditional Medicare to Medicare Advantage programs and make informed choices based on their individualized needs. The second stated goal for the Medicare Advantage option assumes that privately managed plans provide care at lower costs compared with CMS; this assumption has yet to be confirmed with solid data. Indeed, a recent analysis comparing the overall costs of Medicare Advantage to those of Traditional Medicare concluded that Medicare Advantage costs CMS more than Traditional Medicare,⁶ perhaps in part due to risk adjustment practices.⁷

There are a number of areas of uncertainty regarding the specifics of how Medicare Advantage plans work, including Medicare Advantage programs’ use of outpatient (observation) stays. CMS has tried to provide guidance to healthcare organizations and clinicians regarding the appropriate use of inpatient hospitalizations for patients with Traditional Medicare, including the implementation of the Two-Midnight Rule in 2013. According to the rule, clinicians should place inpatient admission orders when they reasonably expect a patient’s care to extend across two midnights.⁸ Such admission decisions are subject to review by Medicare contractors and Quality Improvement Organizations.

In contrast, Medicare Advantage plans which enter into contracts with specific healthcare systems are not required to abide by CMS’ guidelines for the Two-Midnight Rule.⁹ When Medicare Advantage firms negotiate contracts with individual hospitals and healthcare organizations, CMS has been clear that such contracts are not required to include the Two-Midnight Rule when it comes to making hospitalization status decisions.¹⁰ Instead, in these instances, Medicare Advantage plans often use proprietary decision tools containing clinical criteria, such as Milliman Care Guidelines or InterQual, and/or their own plan’s internal criteria as part of the decision-making process to grant inpatient or outpatient (observation) status. More importantly, CMS has stated that for hospitals and healthcare systems that do not contract with Medicare Advantage programs, the Two-Midnight Rule should apply when it comes to making hospitalization status decisions.¹⁰

Implications for Patients

Currently, there are no data available to compare between Medicare Advantage enrollees and traditional medicine beneficiaries in terms of the frequency of observation use and out-of-pocket cost for observation stays. As alluded to in the patient’s case, the use of outpatient (observation) status has implications for a patient’s posthospitalization SNF benefit. Under Traditional Medicare, patients must be hospitalized for three consecutive inpatient midnights in order to qualify for the SNF benefit. Time spent under outpatient (observation) status does not count toward this three-day requirement. Interestingly, some Medicare Advantage programs have demonstrated innovation in this area, waiving the three inpatient midnight requirement for their beneficiaries;¹¹ there is evidence, however, that compared with their Traditional Medicare counterparts, Medicare Advantage beneficiaries are admitted to lower quality SNFs.¹² The posthospitalization consequences of an inpatient versus outpatient (observation) status determination for a Medicare Advantage beneficiary is thus unclear, further complicating the decision-making process for patients when it comes to choosing a Medicare policy, and for providers when it comes to choosing an admission status.

Implications for Clinicians and Healthcare Systems

After performing an initial history and physical exam, if a healthcare provider determines that a patient requires hospitalization, an order is placed to classify the stay as inpatient or outpatient (observation). For beneficiaries with Traditional Medicare or a Medicare Advantage plan that has not contracted with the hospital, clinicians should follow the Two-Midnight Rule for making this determination. For contracted Medicare Advantage, the rules are variable. Under Medicare’s Conditions of Participation, hospitals and healthcare organizations are required to have utilization management (UM) programs to assist physicians in making appropriate admission decisions. UM reviews can happen at any point during or after a patient’s stay, however, and physicians may have to make decisions using their best judgment at the time of admission without real-time input from UM teams.

Outpatient (observation) care and the challenges surrounding appropriate status orders have complicated the admission decision. In one study of 2014 Traditional Medicare claims, almost half of outpatient (observation) stays contained a status change.¹³ Based on a recent survey of hospitalist physicians, about two-thirds of hospitalists report at least monthly requests from patients to change their status.¹⁴ Hospital medicine physicians report that these requests “can severely damage the therapeutic bond”¹⁴ between provider and patient because the provider must assign status based on CMS rules, not patient request.

CMS could improve the current system in one of two ways. First, CMS could require that all Medicare Advantage plans follow the same polices as Traditional Medicare policies regarding the Two- and Three-Midnight Rules. This would eliminate the need for both hospitals and healthcare organizations to dedicate time and resources to negotiating with each Medicare Advantage program and to managing each Medicare Advantage patient admission based on a specific contract. Ideally, CMS could completely eliminate its outpatient (observation) policy so that all hospitalizations are treated exactly the same, classified under the same billing status and with beneficiaries having the same postacute benefit. This would be consistent with the sentiment behind the recent Office of Inspector General’s (OIG) report suggesting that CMS consider counting outpatient midnights toward the three-midnight requirement for postacute SNF care “so that beneficiaries receiving similar hospital care have similar access to these services.”¹⁵

The physician in the example above should tell their patient that they will be admitted as an inpatient given her expectation that the patient will need hospitalization for oxygen support, parenteral antibiotics, and evaluation by physical therapy to determine a medically appropriate discharge plan. The physician should document the medical necessity for the admission, specifically her expectation that the patient will require at least two midnights of medically necessary hospital care. If the patient has Traditional Medicare, this documentation, along with the inpatient status order, will fulfill the requirements for an inpatient stay. If the patient has a Medicare Advantage plan, the physician can advise the patient that the plan administrators will ultimately determine if an inpatient stay will be covered or denied.

In the proposed clinical scenario, the rules determining the patient’s hospitalization status depend on whether the hospital contracts with the patient’s Medicare Advantage plan, and if so, what the contracted criteria are in determining inpatient and outpatient (observation) status. The physician could consider real-time input from the hospital’s UM team, if available. Regardless of UM input, if the physician hospitalizes the patient as an inpatient, the Medicare Advantage plan administrators will make a determination regarding the appropriateness of the admission status, as well as whether the patient qualifies for posthospitalization Medicare SNF benefits (if requested) and, additionally, which SNFs will be covered. If denied, the hospitalist will have the option of a peer-to-peer discussion with the insurance company to overturn the denial. Given the confusion, complexity, and implications presented by this admission status decision-making process, standardization across Traditional Medicare and Medicare Advantage plans, or a budget-neutral plan to eliminate status distinction altogether, is certainly warranted.

A 73-year-old man presents to the emergency department with sepsis secondary to community-acquired pneumonia. The patient requires supplemental oxygen and is started on intravenous antibiotics. His admitting physician expects he will need more than two nights of hospital care and suggests that inpatient status, rather than outpatient (observation) status, would be appropriate under Medicare’s “Two-Midnight Rule.” The physician also suspects the patient may need a brief stay in a skilled nursing facility (SNF) following the mentioned hospitalization and notes that the patient has a Medicare Advantage plan (Table) and wonders if the Two-Midnight Rule applies. Further, she questions whether Medicare’s “Three-Midnight Rule” for SNF benefits will factor in the patient’s discharge planning.

Since the 1970s, the Centers for Medicare & Medicaid Services (CMS) has allowed enrollees to receive their Medicare benefits from privately managed health plans through the so-called Medicare Advantage programs. CMS contracts with commercial insurers who, in exchange for a set payment per Medicare enrollee, “accept full responsibility (ie, risk) for the costs of their enrollees’ care.”¹ Over the past 20 years the percent of Medicare Advantage enrollees has nearly doubled nationwide, from 18% to 34%, and is projected to grow even further to 42% by 2028.^2,3 The reasons beneficiaries choose to enroll in Medicare Advantage over Traditional Medicare have yet to be thoroughly studied; ease of enrollment and plan administration, as well as lower deductibles, copays, and out-of-pocket maximums for in-network services, are thought to be some of the driving factors.

The federal government has asserted two goals for the development of Medicare Advantage: beneficiary choice and economic efficiency.¹ Medicare Advantage plans must be actuarially equal to Traditional Medicare but do not have to cover services in precisely the same way. Medicare Advantage plans may achieve cost savings through narrower networks, strict control of access to SNF services and acute care inpatient rehabilitation, and prior authorization requirements, the latter of which has received recent congressional attention.^4,5 On the other hand, many Medicare Advantage plans offer dental, fitness, optical, and caregiver benefits that are not included under Traditional Medicare. Beneficiaries can theoretically compare the coverage and costs of Traditional Medicare to Medicare Advantage programs and make informed choices based on their individualized needs. The second stated goal for the Medicare Advantage option assumes that privately managed plans provide care at lower costs compared with CMS; this assumption has yet to be confirmed with solid data. Indeed, a recent analysis comparing the overall costs of Medicare Advantage to those of Traditional Medicare concluded that Medicare Advantage costs CMS more than Traditional Medicare,⁶ perhaps in part due to risk adjustment practices.⁷

There are a number of areas of uncertainty regarding the specifics of how Medicare Advantage plans work, including Medicare Advantage programs’ use of outpatient (observation) stays. CMS has tried to provide guidance to healthcare organizations and clinicians regarding the appropriate use of inpatient hospitalizations for patients with Traditional Medicare, including the implementation of the Two-Midnight Rule in 2013. According to the rule, clinicians should place inpatient admission orders when they reasonably expect a patient’s care to extend across two midnights.⁸ Such admission decisions are subject to review by Medicare contractors and Quality Improvement Organizations.

In contrast, Medicare Advantage plans which enter into contracts with specific healthcare systems are not required to abide by CMS’ guidelines for the Two-Midnight Rule.⁹ When Medicare Advantage firms negotiate contracts with individual hospitals and healthcare organizations, CMS has been clear that such contracts are not required to include the Two-Midnight Rule when it comes to making hospitalization status decisions.¹⁰ Instead, in these instances, Medicare Advantage plans often use proprietary decision tools containing clinical criteria, such as Milliman Care Guidelines or InterQual, and/or their own plan’s internal criteria as part of the decision-making process to grant inpatient or outpatient (observation) status. More importantly, CMS has stated that for hospitals and healthcare systems that do not contract with Medicare Advantage programs, the Two-Midnight Rule should apply when it comes to making hospitalization status decisions.¹⁰

Implications for Patients

Currently, there are no data available to compare between Medicare Advantage enrollees and traditional medicine beneficiaries in terms of the frequency of observation use and out-of-pocket cost for observation stays. As alluded to in the patient’s case, the use of outpatient (observation) status has implications for a patient’s posthospitalization SNF benefit. Under Traditional Medicare, patients must be hospitalized for three consecutive inpatient midnights in order to qualify for the SNF benefit. Time spent under outpatient (observation) status does not count toward this three-day requirement. Interestingly, some Medicare Advantage programs have demonstrated innovation in this area, waiving the three inpatient midnight requirement for their beneficiaries;¹¹ there is evidence, however, that compared with their Traditional Medicare counterparts, Medicare Advantage beneficiaries are admitted to lower quality SNFs.¹² The posthospitalization consequences of an inpatient versus outpatient (observation) status determination for a Medicare Advantage beneficiary is thus unclear, further complicating the decision-making process for patients when it comes to choosing a Medicare policy, and for providers when it comes to choosing an admission status.

Implications for Clinicians and Healthcare Systems

After performing an initial history and physical exam, if a healthcare provider determines that a patient requires hospitalization, an order is placed to classify the stay as inpatient or outpatient (observation). For beneficiaries with Traditional Medicare or a Medicare Advantage plan that has not contracted with the hospital, clinicians should follow the Two-Midnight Rule for making this determination. For contracted Medicare Advantage, the rules are variable. Under Medicare’s Conditions of Participation, hospitals and healthcare organizations are required to have utilization management (UM) programs to assist physicians in making appropriate admission decisions. UM reviews can happen at any point during or after a patient’s stay, however, and physicians may have to make decisions using their best judgment at the time of admission without real-time input from UM teams.

Outpatient (observation) care and the challenges surrounding appropriate status orders have complicated the admission decision. In one study of 2014 Traditional Medicare claims, almost half of outpatient (observation) stays contained a status change.¹³ Based on a recent survey of hospitalist physicians, about two-thirds of hospitalists report at least monthly requests from patients to change their status.¹⁴ Hospital medicine physicians report that these requests “can severely damage the therapeutic bond”¹⁴ between provider and patient because the provider must assign status based on CMS rules, not patient request.

CMS could improve the current system in one of two ways. First, CMS could require that all Medicare Advantage plans follow the same polices as Traditional Medicare policies regarding the Two- and Three-Midnight Rules. This would eliminate the need for both hospitals and healthcare organizations to dedicate time and resources to negotiating with each Medicare Advantage program and to managing each Medicare Advantage patient admission based on a specific contract. Ideally, CMS could completely eliminate its outpatient (observation) policy so that all hospitalizations are treated exactly the same, classified under the same billing status and with beneficiaries having the same postacute benefit. This would be consistent with the sentiment behind the recent Office of Inspector General’s (OIG) report suggesting that CMS consider counting outpatient midnights toward the three-midnight requirement for postacute SNF care “so that beneficiaries receiving similar hospital care have similar access to these services.”¹⁵

The physician in the example above should tell their patient that they will be admitted as an inpatient given her expectation that the patient will need hospitalization for oxygen support, parenteral antibiotics, and evaluation by physical therapy to determine a medically appropriate discharge plan. The physician should document the medical necessity for the admission, specifically her expectation that the patient will require at least two midnights of medically necessary hospital care. If the patient has Traditional Medicare, this documentation, along with the inpatient status order, will fulfill the requirements for an inpatient stay. If the patient has a Medicare Advantage plan, the physician can advise the patient that the plan administrators will ultimately determine if an inpatient stay will be covered or denied.

In the proposed clinical scenario, the rules determining the patient’s hospitalization status depend on whether the hospital contracts with the patient’s Medicare Advantage plan, and if so, what the contracted criteria are in determining inpatient and outpatient (observation) status. The physician could consider real-time input from the hospital’s UM team, if available. Regardless of UM input, if the physician hospitalizes the patient as an inpatient, the Medicare Advantage plan administrators will make a determination regarding the appropriateness of the admission status, as well as whether the patient qualifies for posthospitalization Medicare SNF benefits (if requested) and, additionally, which SNFs will be covered. If denied, the hospitalist will have the option of a peer-to-peer discussion with the insurance company to overturn the denial. Given the confusion, complexity, and implications presented by this admission status decision-making process, standardization across Traditional Medicare and Medicare Advantage plans, or a budget-neutral plan to eliminate status distinction altogether, is certainly warranted.

A 73-year-old man presents to the emergency department with sepsis secondary to community-acquired pneumonia. The patient requires supplemental oxygen and is started on intravenous antibiotics. His admitting physician expects he will need more than two nights of hospital care and suggests that inpatient status, rather than outpatient (observation) status, would be appropriate under Medicare’s “Two-Midnight Rule.” The physician also suspects the patient may need a brief stay in a skilled nursing facility (SNF) following the mentioned hospitalization and notes that the patient has a Medicare Advantage plan (Table) and wonders if the Two-Midnight Rule applies. Further, she questions whether Medicare’s “Three-Midnight Rule” for SNF benefits will factor in the patient’s discharge planning.

Since the 1970s, the Centers for Medicare & Medicaid Services (CMS) has allowed enrollees to receive their Medicare benefits from privately managed health plans through the so-called Medicare Advantage programs. CMS contracts with commercial insurers who, in exchange for a set payment per Medicare enrollee, “accept full responsibility (ie, risk) for the costs of their enrollees’ care.”¹ Over the past 20 years the percent of Medicare Advantage enrollees has nearly doubled nationwide, from 18% to 34%, and is projected to grow even further to 42% by 2028.^2,3 The reasons beneficiaries choose to enroll in Medicare Advantage over Traditional Medicare have yet to be thoroughly studied; ease of enrollment and plan administration, as well as lower deductibles, copays, and out-of-pocket maximums for in-network services, are thought to be some of the driving factors.

The federal government has asserted two goals for the development of Medicare Advantage: beneficiary choice and economic efficiency.¹ Medicare Advantage plans must be actuarially equal to Traditional Medicare but do not have to cover services in precisely the same way. Medicare Advantage plans may achieve cost savings through narrower networks, strict control of access to SNF services and acute care inpatient rehabilitation, and prior authorization requirements, the latter of which has received recent congressional attention.^4,5 On the other hand, many Medicare Advantage plans offer dental, fitness, optical, and caregiver benefits that are not included under Traditional Medicare. Beneficiaries can theoretically compare the coverage and costs of Traditional Medicare to Medicare Advantage programs and make informed choices based on their individualized needs. The second stated goal for the Medicare Advantage option assumes that privately managed plans provide care at lower costs compared with CMS; this assumption has yet to be confirmed with solid data. Indeed, a recent analysis comparing the overall costs of Medicare Advantage to those of Traditional Medicare concluded that Medicare Advantage costs CMS more than Traditional Medicare,⁶ perhaps in part due to risk adjustment practices.⁷

There are a number of areas of uncertainty regarding the specifics of how Medicare Advantage plans work, including Medicare Advantage programs’ use of outpatient (observation) stays. CMS has tried to provide guidance to healthcare organizations and clinicians regarding the appropriate use of inpatient hospitalizations for patients with Traditional Medicare, including the implementation of the Two-Midnight Rule in 2013. According to the rule, clinicians should place inpatient admission orders when they reasonably expect a patient’s care to extend across two midnights.⁸ Such admission decisions are subject to review by Medicare contractors and Quality Improvement Organizations.

In contrast, Medicare Advantage plans which enter into contracts with specific healthcare systems are not required to abide by CMS’ guidelines for the Two-Midnight Rule.⁹ When Medicare Advantage firms negotiate contracts with individual hospitals and healthcare organizations, CMS has been clear that such contracts are not required to include the Two-Midnight Rule when it comes to making hospitalization status decisions.¹⁰ Instead, in these instances, Medicare Advantage plans often use proprietary decision tools containing clinical criteria, such as Milliman Care Guidelines or InterQual, and/or their own plan’s internal criteria as part of the decision-making process to grant inpatient or outpatient (observation) status. More importantly, CMS has stated that for hospitals and healthcare systems that do not contract with Medicare Advantage programs, the Two-Midnight Rule should apply when it comes to making hospitalization status decisions.¹⁰

Implications for Patients

Currently, there are no data available to compare between Medicare Advantage enrollees and traditional medicine beneficiaries in terms of the frequency of observation use and out-of-pocket cost for observation stays. As alluded to in the patient’s case, the use of outpatient (observation) status has implications for a patient’s posthospitalization SNF benefit. Under Traditional Medicare, patients must be hospitalized for three consecutive inpatient midnights in order to qualify for the SNF benefit. Time spent under outpatient (observation) status does not count toward this three-day requirement. Interestingly, some Medicare Advantage programs have demonstrated innovation in this area, waiving the three inpatient midnight requirement for their beneficiaries;¹¹ there is evidence, however, that compared with their Traditional Medicare counterparts, Medicare Advantage beneficiaries are admitted to lower quality SNFs.¹² The posthospitalization consequences of an inpatient versus outpatient (observation) status determination for a Medicare Advantage beneficiary is thus unclear, further complicating the decision-making process for patients when it comes to choosing a Medicare policy, and for providers when it comes to choosing an admission status.

Implications for Clinicians and Healthcare Systems

After performing an initial history and physical exam, if a healthcare provider determines that a patient requires hospitalization, an order is placed to classify the stay as inpatient or outpatient (observation). For beneficiaries with Traditional Medicare or a Medicare Advantage plan that has not contracted with the hospital, clinicians should follow the Two-Midnight Rule for making this determination. For contracted Medicare Advantage, the rules are variable. Under Medicare’s Conditions of Participation, hospitals and healthcare organizations are required to have utilization management (UM) programs to assist physicians in making appropriate admission decisions. UM reviews can happen at any point during or after a patient’s stay, however, and physicians may have to make decisions using their best judgment at the time of admission without real-time input from UM teams.

Outpatient (observation) care and the challenges surrounding appropriate status orders have complicated the admission decision. In one study of 2014 Traditional Medicare claims, almost half of outpatient (observation) stays contained a status change.¹³ Based on a recent survey of hospitalist physicians, about two-thirds of hospitalists report at least monthly requests from patients to change their status.¹⁴ Hospital medicine physicians report that these requests “can severely damage the therapeutic bond”¹⁴ between provider and patient because the provider must assign status based on CMS rules, not patient request.

CMS could improve the current system in one of two ways. First, CMS could require that all Medicare Advantage plans follow the same polices as Traditional Medicare policies regarding the Two- and Three-Midnight Rules. This would eliminate the need for both hospitals and healthcare organizations to dedicate time and resources to negotiating with each Medicare Advantage program and to managing each Medicare Advantage patient admission based on a specific contract. Ideally, CMS could completely eliminate its outpatient (observation) policy so that all hospitalizations are treated exactly the same, classified under the same billing status and with beneficiaries having the same postacute benefit. This would be consistent with the sentiment behind the recent Office of Inspector General’s (OIG) report suggesting that CMS consider counting outpatient midnights toward the three-midnight requirement for postacute SNF care “so that beneficiaries receiving similar hospital care have similar access to these services.”¹⁵

The physician in the example above should tell their patient that they will be admitted as an inpatient given her expectation that the patient will need hospitalization for oxygen support, parenteral antibiotics, and evaluation by physical therapy to determine a medically appropriate discharge plan. The physician should document the medical necessity for the admission, specifically her expectation that the patient will require at least two midnights of medically necessary hospital care. If the patient has Traditional Medicare, this documentation, along with the inpatient status order, will fulfill the requirements for an inpatient stay. If the patient has a Medicare Advantage plan, the physician can advise the patient that the plan administrators will ultimately determine if an inpatient stay will be covered or denied.

In the proposed clinical scenario, the rules determining the patient’s hospitalization status depend on whether the hospital contracts with the patient’s Medicare Advantage plan, and if so, what the contracted criteria are in determining inpatient and outpatient (observation) status. The physician could consider real-time input from the hospital’s UM team, if available. Regardless of UM input, if the physician hospitalizes the patient as an inpatient, the Medicare Advantage plan administrators will make a determination regarding the appropriateness of the admission status, as well as whether the patient qualifies for posthospitalization Medicare SNF benefits (if requested) and, additionally, which SNFs will be covered. If denied, the hospitalist will have the option of a peer-to-peer discussion with the insurance company to overturn the denial. Given the confusion, complexity, and implications presented by this admission status decision-making process, standardization across Traditional Medicare and Medicare Advantage plans, or a budget-neutral plan to eliminate status distinction altogether, is certainly warranted.