This Open Access article is distributed under the terms of the Creative Commons Attribution Noncommercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.
Medicare Advantage: Issues, Insights, and Implications for the Future
Paul Cotton, Joseph P. Newhouse, PhD, Kevin G. Volpp, MD, PhD, A. Mark Fendrick, MD, Susan Lynne Oesterle, Pat Oungpasuk, Ruchi Aggarwal, Gail Wilensky, PhD, and Kathleen Sebelius
D.B. Nash, and A.Y. Schwartz
A.M. Fendrick, S.L. Oesterle, P. Oungpasuk, and R. Aggarwal
G. Wilensky and K. Sebelius
Population Health Management publishes supplements that (a) discuss new technologies, theories, and/or practice, and (b) serve as enduring materials to disseminate information from conferences and special meetings. Supplements that discuss new technologies, theories, and/or practices are subject to peer review.
The Better Medicare Alliance provided funding support to the Jefferson College of Population Health to produce this special supplement. The content and opinions expressed herein are those of the authors.
The authors declared the following potential conflicts of interest: Dr. Nash is employed by the Jefferson College of Population Health, which was compensated by the Better Medicare Alliance to produce this supplement. Ms. Schwartz is employed by the Better Medicare Alliance. Dr. Newhouse is a director and holds equity in Aetna, which sells Medicare Advantage plans. Mr. Cotton, Drs. Volpp, Fendrick, and Wilensky, and Ms. Oesterle, Ms. Oungpasuk, Ms. Aggarwal, and Ms. Sebelius declared no conflicts of interest.
* Allyson Y. Schwartz is President and CEO of the Better Medicare Alliance, Washington, DC.
Find articles by Allyson Y. SchwartzJuly 30, 2015 was the 50 th anniversary of Medicare. During those 50 years, American seniors and individuals with disabilities eligible for Medicare have come to count on Medicare for health care coverage and financial security. Understood to be one of our nation's most successful initiatives, it is viewed as stable and secure, unchanged, and guaranteed to all seniors. Yet, Medicare has been significantly expanded and modified over the decades to meet the needs of its population. These changes have expanded coverage, impacted beneficiary costs, and directed reimbursements for providers. With the passage of the Patient Protection & Affordable Care Act of 2010 (ACA), and the more recent Medicare Access and CHIP Authorization Act of 2015 (MACRA), policy makers are moving to reform the Traditional Fee-For-Service Medicare payment structure to reward value rather than the volume of care by means of bundled payments, accountable care organizations, and incentives for alternative payment models. Medicare Advantage (MA) is leading the way in this transformation away from volume-based care toward a framework of care that aligns incentives that focus on improving health status.
For roughly 40 years, Medicare beneficiaries have had the choice to receive their Medicare benefits from a private health plan instead of through Traditional Fee-For-Service Medicare. In the years since its inception, the program, also known at Part C and formerly known as Medicare+Choice, has undergone multiple updates, including major policy changes passed in 2010 as part of the ACA. Since the passage of the ACA, MA payment is updated each year based on fee-for-service spending at the county level. The per member per month payments to MA plans are also risk adjusted using a complex model that accounts for demographic, health status, income, and other variables. Risk adjustment is critical to MA; it facilitates early diagnosis and prevention, ensures that appropriate resources are available to care for the most complex patients, and prevents preferential selection. MA includes quality bonuses based on plan performance in the MA Star Rating system, as well as an out-of-pocket maximum, neither of which are included in Traditional Medicare. Today roughly 70% of MA enrollees are in high-quality MA plans rated 4+ stars (on a 5-star scale), up from 2012 when enrollment in 4+ star-rated plans was close to 30%.
There are other important distinctions between Traditional Medicare and MA. The Traditional Medicare structure is built around an acute care model that achieves suboptimal outcomes whereas MA is structured to identify health issues early, enable beneficiaries to live with chronic illness, and support providers who offer a full range of coordinated services. MA also has an emphasis on transparency, accountability, and innovation.
Today, our nation faces major challenges in Medicare. First, there is the reality of the near doubling of the number of older adults in the next decade, with 10,000 more individuals turning 65 each day. Second, many of these seniors will live longer and many more will live with multiple serious chronic conditions. There are fiscal implications as the federal government seeks to meet the expectations of these almost 75 million beneficiaries by 2026, as they, their families, and their communities struggle with cost sharing and gaps in coverage.
Health care delivery and financing in America must respond to these changes. The good news is that Medicare, and particularly MA, the part of Medicare that offers a capitated, managed care option for beneficiaries, is actively engaged in delivering patient-centered, high-value care. Evidence shows that MA is achieving the improved outcomes for seniors. Providers—doctors, nurses, hospitals—are changing the way they deliver care to focus on primary care, early intervention, care coordination, and the socioeconomic determinants of health.
This impact is amplified as more beneficiaries choose MA – the Centers for Medicare & Medicaid Services reports that enrollment in MA is expected to reach 18.5 million in 2017, representing 32% of Medicare beneficiaries. Plans and providers report their interest in working to engage beneficiaries in their own health and to help them navigate the complex array of services and care decisions to ensure the best outcomes. We are excited by the fact that MA offers seniors a managed care option with the capacity to tailor care for patients with complex health and social needs. It is offering innovative solutions through new community partnerships, team approaches to care coordination for individuals with serious chronic conditions, attention to care transitions, home-based care, and new financing models between payors and practitioners that incentivize new, effective ways of providing care.
As we approach the upcoming transition, with a new President, new Administration and a new Congress, we call on policy makers to build on the groundbreaking work of MA. MA intersects with the tenets of population health in a number of ways. First and foremost, MA concentrates on prevention, wellness, and upstream health benefits. It focuses health care delivery on coordinating care to slow disease progression, improve outcomes, and redirect health care dollars to meet patients' needs, including vitally important social determinants of health. Importantly, MA offers seniors and individuals with disabilities a managed care option that is transparent and accountable. We are hopeful that policy makers will take the lessons learned and do all they can to promote better, smarter health care for seniors—and all Americans.
This supplement offers the perspectives of experts, researchers, and administrators on MA today and the important role it can play in Medicare for years to come. We hope you find it useful and interesting.
Policy shifts in the early 2000s resulted in Medicare Advantage (MA) payment rates that were estimated to be 14% higher per enrollee than the cost of covering Medicare beneficiaries under Traditional Fee-For-Service Medicare (TM). 1 In terms of total dollars spent, MA cost the government $14 billion more than TM in 2009 alone. 2 In addition, though MA Star Ratings were launched in 2008, at the time some MA plan sponsors told the National Committee for Quality Assurance (NCQA) that it was difficult to implement quality changes without aligned incentives. As a result, little change occurred in performance levels from 2008 to 2012.
The Patient Protection & Affordable Care Act of 2010 (ACA) sought to bring MA plan payments closer to the payments for TM over time by means of a gradual reduction in payments. The ACA also aimed to improve quality by providing bonus payments to MA plans based on achieving high scores in the Star Ratings system. Despite critics' warnings that these actions would decimate the MA program, 3 evidence shows the changes have had their intended impact and the program is thriving.
Since the ACA, MA enrollment has grown steadily and now is rising at a faster rate than overall Medicare enrollment, from 10.9 million in 2009 to a projected 18.5 million in 2017 (ie, 24% to 32% of all Medicare enrollees). 4,5 This represents a 57.8% increase in covered lives in MA and nearly 10 million more enrollees in MA than the Congressional Budget Office projected just after the ACA's enactment.
Total MA payment has dropped to 2% above TM costs in 2016, a difference that is related primarily to quality bonuses and rebates. 6 Also, MA plan premiums have fallen by nearly 13% since 2010. 5 In 2016, as migration continued to highly rated plans with lower cost sharing and additional benefits, the average premium paid by an MA enrollee decreased by $0.31 from 2015. 7 According to the Centers for Medicare & Medicaid Services (CMS), 59% of enrollees had no premium hike in 2016. 7
Perhaps more importantly, the percentage of highly-rated 4- and 5-Star plans increased from 31% in 2010 to 49% in 2017. 8 The percentage of beneficiaries enrolled in highly-rated plans rose even more, from 31% in 2012 to 68% in 2017. 8
These positive effects were realized because the ACA provided clear guidance and strong incentives for MA plans to improve. MA plan sponsors have hired new staff with titles such as “Medicare Stars Director” to focus specifically on improving MA Star Ratings scores. Conferences dedicated to this win-win challenge are common across the country. Real people – the nearly one third of the Medicare population now in MA plans – are getting the higher quality health care they deserve. More importantly, the potential for further improvement is vast, as emerging measures will take on high-priority issues such as care coordination and behavioral health.
Today, linking MA Star Ratings to performance-based pay appears to be a critical element in the shift from volume-driven fee-for-service incentives to value-based plans that reward clinicians on the basis of how well they meet the Triple Aim of providing high-quality, efficient, patient-centered care. Performance-based pay is being applied across all parts of Medicare, most state Medicaid, and other health programs, as well as most commercial insurers.
Medicare first implemented the Medicare Advantage Star Ratings system in 2008, based on:
• Clinical quality measures, such as those in NCQA's Healthcare Effectiveness Data & Information Set;
• Patient experience-of-care measures in the Consumer Assessment of Healthcare Providers & Systems;• Patient-reported outcomes measures in the Medicare Health Outcomes Survey 9 of randomly selected Medicare enrollees;
• Medicare administrative measures on issues such as appeals and complaints.Medicare updates measures each year to remove any that “top out” with little room for further improvement, and adjusts specifications for changes in underlying clinical evidence.
Measure weights vary by type: structure and process measures (eg, preventive screenings) have weights of 1; patient experience measures (eg, access to care) have weights of 1.5; outcome measures (eg, readmission rates) and intermediate outcomes (eg, high blood pressure control) have weights of 3. A measure of overall plan improvement has a weight of 5.
The measures address both Medicare Advantage (Part C) medical service coverage and prescription drug coverage (Part D). They are grouped into 9 domains:
1. Staying Healthy: screenings, tests, and vaccines 2. Managing Chronic (Long-Term) Conditions 3. Member Experience with Health Plan 4. Member Complaints and Changes in Health Plan Performance 5. Health Plan Customer Service 6. Drug Plan Customer Service 7. Member Complaints and Changes in the Drug Plan's Performance 8. Member Experience with the Drug Plan 9. Drug Safety and Accuracy of Drug PricingMedicare calculates results for each measure, each domain, and then an overall score, and publicly reports the results on the MA Plan Finder 10 to help inform beneficiaries' choices. Results also determine plan bonuses and rebates: MA plans with Star Ratings of 4, 4.5, or 5 receive bonuses equal to 5% of the county-level benchmark rate. MA plans may use bonuses to lower premiums and/or cost sharing, or to provide additional benefits, all of which attract enrollees to higher rated plans.
Plans receive rebates if the bid price they propose for covering their enrollees is below the benchmark. Rebates rise with Star Ratings.
• Plans with 4.5–5 Stars get 70% of the difference between the benchmark and their bid price • Plans with 3.5–4.5 stars get 65% • Plans with <3.5 Stars get 50%Medicare requires plans to use rebates to lower premiums/ cost sharing or provide additional benefits.
The highest-rated 5-Star plans are further rewarded with the ability to enroll new members outside of the Fall open enrollment period – another way to boost enrollment in the highest quality plans.
The ACA also penalizes poor performers – plans with less than 3 Stars for 3 or more years. Medicare sends letters to enrollees in these plans encouraging them to shop for alternative coverage, places an icon on Plan Finder to warn potential enrollees, and prohibits enrollment through the Plan Finder.
1. Medicare Payment Advisory Commission. Report to the Congress: Medicare Payment Policy . March 2009. http://www.medpac.gov/docs/default-source/reports/march-2009-report-to-congress-medicare-payment-policy.pdf?sfvrsn=0 Accessed October18, 2016
2. The CMS Blog. Affordable Care Act has strengthened Medicare Advantage and Prescription Drug Program . September 22, 2016. https://blog.cms.gov/2016/09/22/affordable-care-act-has-strengthened-medicare-advantage-and-prescription-drug-program/ Accessed October18, 2016
3. Book RA, Capretta JC. Reductions in Medicare Advantage Payments: the impact on seniors by region . September 14, 2010. http://www.heritage.org/research/reports/2010/09/reductions-in-medicare-advantage-payments-the-impact-on-seniors-by-region Accessed October10, 2016
4. Congressional Budget Office (CBO). CBO's August 2010 Baseline: Medicare . August 2010. https://www.cbo.gov/sites/default/files/51302-2010-08-Medicare.pdf Accessed October10, 2016
5. Centers for Medicare & Medicaid Services Fact Sheet: Moving Medicare Advantage and Part D Forward. September 22, 2016. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2016-Fact-sheets-items/2016-09-22.html Accessed October18, 2016
6. Medicare Payment Advisory Commission. March 2016 Report to the Congress: Medicare Payment Policy. Chapter 12: The Medicare Advantage program: Status report . March 2016. http://www.medpac.gov/docs/default-source/reports/chapter-12-the-medicare-advantage-program-status-report-march-2016-report-.pdf?sfvrsn=0 Accessed October18, 2016
8. Centers for Medicare & Medicaid Services. 2017 Star Ratings . October 12, 2016. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2016-Fact-sheets-items/2016-10-12.html Accessed October18, 2016
10. Centers for Medicare & Medicaid Services. Medicare Health Outcomes Survey . http://www.hosonline.org/ Accessed September15, 2016
For the past several years I have led a group carrying out research comparing Medicare Advantage (MA) and Traditional Fee-For-Service Medicare (TM). Many of our findings are summarized in 2 publications 1,2 ; the following is a brief summary of conclusions from these papers and a discussion of some related issues.
MA plans bid against a benchmark based on average TM spending per county to provide the same covered services as TM. MA plans also must design cost sharing that does not exceed the annual MA out-of-pocket maximum, which was $6700 in 2016. TM does not have an out-of-pocket maximum. For 2016, MA plan bids for a standardized beneficiary were 94% of TM costs. 3 For health maintenance organization (HMO) MA plans, the bids were 90% of TM costs. 3 However, Medicare paid MA plans 102% of TM costs, with the 8 percentage point difference (8 = 102-94) going to beneficiaries in the form of lower premiums, lower cost sharing, additional covered services such as dental, or some combination of the 3.
We compared emergency department (ED) and hospital use in MA HMOs with a sample of TM beneficiaries matched on age, sex, race, and zip code in 2003 and again in 2009. ED use by the MA group was 25%-35% less, and hospital days were about 20%-25% less, than the TM group. Even more striking were comparisons of age-, sex-, race-, and zip code-adjusted use among individuals in the last 6 months of life. In each year between 2003 and 2009, not only were hospital admission rates 5%-14% less but ED visits per person in the MA group were fully 42%-54% less than in the comparison group. Common preventive measures such as mammography, hemoglobin A1c testing, and cholesterol testing were carried out at greater frequency in the MA group as well. 4
Prior to 1998, MA plans were paid 95% of TM spending in a given county, adjusted for age, sex, Medicaid eligibility, institutionalization, and employment. The Congressional Budget Office estimated that Medicare actually spent 8% more on those enrolled in MA at the time because they were sufficiently better risks than those enrolled in TM. As a result, the 1997 Balanced Budget Act directed the Centers for Medicare & Medicaid Services (CMS) to use diagnoses recorded on claim forms to adjust reimbursement; thus, plans would be reimbursed more for a beneficiary with a serious chronic condition (eg, cancer) than a beneficiary without such a diagnosis. CMS initiated this effort in 2000, but it was not fully implemented until 2007. Although some favorable selection into MA may remain, it now is at low levels and arguably not a large policy concern. 5–7 It was often said, only half in jest, that we would know risk adjustment was adequate when MA plans began to advertise for sick persons. Although some were skeptical about this ever happening, we now see plans in the younger than age 65 marketplaces advertising for persons with diabetes to enroll. 8
The introduction of diagnosis into the risk-adjustment scheme raised a new concern. MA plans were so aggressive about recording diagnoses that there was concern that CMS was paying more for care of the same person in MA than that person would have cost in TM. To address this issue, the Patient Protection and Affordable Care Act of 2010 (ACA) introduced a coding intensity adjustment; MA reimbursement was reduced annually beginning in 2010, with a 5.4% reduction in risk scores in 2016 and with adjustments through 2016 amounting to a more than 20% reduction in risk scores. Under current law, the reductions are to continue until risk adjustment is based on MA encounter data rather than TM data; MA is currently phasing into encounter data as a diagnosis source, and in 2017 risk adjustment will be 25% based on encounter data. There is still debate about whether the current coding intensity adjustment sufficiently accounts for coding pattern differences in MA and TM. 3,9
Although the ACA calls for risk adjustment to be based on MA data, doing so would not lead to an efficient allocation of persons between MA and TM. If MA can deliver the same outcome for persons with certain diagnoses at a lower cost than TM, it is technically efficient for those persons to be in MA and vice versa. Basing payment to MA plans on TM cost for a given mix of diagnoses does this; MA plans will seek out patients who afford them a cost advantage, and not patients for whom TM has a cost advantage.
The comparisons we made were with TM as it historically existed; however, TM has been changing and is likely to evolve further with the advent of accountable care organizations, bundled payments, and reimbursement reforms under the Medicare Access and CHIP Reauthorization Act of 2015 (MACRA), all of which seek to change financial incentives of providers in the TM program. How MA will compare with a remodeled TM program, and indeed how many provider organizations will seek to become MA plans, will be important to monitor going forward.
1. McGuire TG, Newhouse JP, Sinaiko AD. An economic history of Medicare Part C . Milbank Q . 2011; 89 :289–332 [PMC free article] [PubMed] [Google Scholar]
2. Newhouse JP, McGuire TG. How successful is Medicare Advantage? Milbank Q . 2014; 92 :351–394 [PMC free article] [PubMed] [Google Scholar]
3. Medicare Payment Advisory Commission. Report to Congress: Medicare Payment Policy . March, 2016. http://www.medpac.gov/docs/default-source/reports/march-2016-report-to-the-congress-medicare-payment-policy.pdf?sfvrsn=0 Accessed September12, 2016
4. Ayanian JZ, Landon BE, Zaslavsky AM, Saunders RC, Pawlson LG, Newhouse JP. Medicare beneficiaries more likely to receive appropriate ambulatory services in HMOs than in traditional Medicare . Health Aff (Millwood) . 2013; 32 :1228–1235 [PMC free article] [PubMed] [Google Scholar]
5. McWilliams JM, Hsu J, Newhouse JP. New risk-adjustment system was associated with reduced favorable selection in Medicare Advantage . Health Aff (Millwood) . 2012; 31 :2630–2640 [PMC free article] [PubMed] [Google Scholar]
6. Newhouse JP, Price M, Huang J, McWilliams JM, Hsu J. Steps to reduce favorable risk selection in Medicare Advantage largely succeeded, boding well for health insurance exchanges . Health Aff (Millwood) . 2012; 31 :2618–2628 [PMC free article] [PubMed] [Google Scholar]
7. Newhouse JP, Price M, McWilliams JM, Hsu J, McGuire TG. How much favorable selection is left in Medicare Advantage? Am J Health Econ . 2015; 1 ( 1 ):1–26 [PMC free article] [PubMed] [Google Scholar]
8. Andrews M. New health plans offer discounts for diabetes care . November 17, 2015. http://khn.org/news/new-health-plans-offer-discounts-for-diabetes-care/ Accessed September12, 2016
9. Kronick R, Welch WP. Measuring coding intensity in the Medicare Advantage program . MMRR . 2014; 4 ( 2 ):E1-E19 [PMC free article] [PubMed] [Google Scholar]
Recent years have highlighted a shift away from the reactive, visit-based model of health care financing and toward a new era of proactive, health outcomes-based models (eg, primary care medical homes, accountable care organizations) wherein provider organizations are at financial risk for quality of care and insurance premiums are increasingly linked to patients' health behaviors and/or outcomes. This has heightened the importance of patient engagement. This is especially true for those in Medicare with chronic conditions who account for a disproportionate share of health care expenditures and may find it difficult to manage their own care. Medicare Advantage's capitated system provides a framework that enables plans to test different care management strategies.
Today, providers as well as public and private insurers, including those that offer Medicare Advantage plans, are engaging in a variety of tactics aimed at promoting positive health behaviors. Although these programs are well-intended, many are built on traditional economic models that do not take into account common decision errors and predictably irrational behavior. These programs often have minimal impact or produce unintended consequences. Programs built on a behavioral economic model leverage evidence about human psychology and decision making into their designs, making them more effective.
Incentive-based wellness programs (eg, a $150 reward – paid at the end of a year – for going to the gym 100 times) have become increasingly popular. However, a national survey estimated that participation rates in many such programs are low (5%-10%), particularly in programs that target smoking or obesity. One reason for low enrollment is failure to take basic principles of behavioral economics into account; the 100-visit threshold is too hard to reach for someone who does not already go to the gym, the feedback at the end of the year is far away in time and not very motivating at the beginning of the year, and the incentive is relatively small ($1.50 per visit). 1
A recent study on the effect of financial incentives (to primary care physicians, patients, or both) on lipid levels concluded that shared incentives resulted in a statistically significant difference in the reduction of low-density lipoprotein cholesterol (LDL-C) levels at 12 months, whereas incentives to physicians or patients alone did not. 2 This is because many chronic diseases require action by both providers and patients and although the provider incentive increased medication intensification and the patient incentive increased adherence, only the shared incentive arm resulted in both intensification of therapy by providers and higher patient adherence. Forty-nine percent of patients in the shared patient-physician incentive group achieved the LDL-C goal compared to 36%-40% of patients in comparison groups.
For more than 2 decades, disease management programs have attempted to engage patients in managing their chronic conditions (eg, diabetes) and, more recently, transitional care models have been touted as an effective means for coordinating care post hospitalization. These conventional “hovering” approaches are personnel based (eg, visiting nurses, telemedicine services) and expensive, and results have been mixed. For instance, a large multicenter trial of telemonitoring for patients with heart disease showed no effect on the primary outcomes of rehospitalization and death. Moreover, 14% of those assigned to the intervention group did not use the system at all and nearly half of those who did lost interest over time.
As medication adherence has taken on greater importance, home-based biometric assessments of various indicators have emerged as a longitudinal part of clinical care. The challenge lies in automating the hovering – thereby reducing its cost – and incorporating this monitoring into patients' lives in ways that are acceptable, convenient, and even welcomed. Three recent developments suggest that automated hovering may offer promise:
1. The payment mechanisms that support accountability for health outcomes (eg, non-reimbursement for preventable admissions, bundled payments around goals of care) also provide the financial engine to support automated hovering initiatives.
2. Although most people want better health and typically know what is required to achieve it, distractions and urgencies of the moment often get in the way of pursuing long-term wellness goals. Behavioral economics explains why individuals behave this way and provides tools for redirecting behavior toward health goals with carefully deployed nudges and financial incentives.
3. The expanded reach of today's simple and more sophisticated technologies (eg, cell phones, wireless devices) that can help experts connect to people during their everyday lives were not part of traditional disease management programs in the past.
This type of automated hovering must be targeted to the right clinical and social circumstances (eg, patients with conditions whose management depends substantially on their own behavior – those with heart failure or acute coronary artery syndromes that must strictly adhere to medication). In one study, patients taking warfarin via a home-based pill dispenser that was electronically tethered to a lottery system were automatically entered in a daily random drawing with a small chance of winning $100 and a larger chance of winning $10. Each day, patients were electronically notified if their number had come up – but they were eligible for the prize only if they'd taken their warfarin the previous day as signaled by the dispenser. The system provided daily engagement, the chance of a prize, and a sense of anticipated regret. Although the expected value was less than $3 a day, the system reduced the rate of incorrect doses from 22% to ∼3% and reduced the rate of out-of-range international normalized ratios from 35% to 12%.
Much is yet to be understood regarding the types of patients, approaches, conditions, and settings in which these programs would be most helpful in Medicare. Clinical and research opportunities in behavioral economics should be pursued in conjunction with careful, iterative testing because these new forms of patient engagement will be central to improving population health in the future. Understanding how different payment frameworks enable or inhibit care management strategies is another important element that must be investigated.
1. Loewenstein G, Asch DA, Volpp KG. Behavioral economics holds potential to deliver better results for patients, insurers, and employers . Health Aff (Millwood) . 2013; 32 :1244–1250 [PubMed] [Google Scholar]
2. Asch DA, Troxel AB, Stewart WF, et al.. Effect of financial incentives to physicians, patients or both on lipid levels . JAMA . 2015; 314 :1926–1935 [PMC free article] [PubMed] [Google Scholar]
3. Asch DA, Muller RW, Volpp KG. Automated hovering in health care – watching over the 5000 hours . New Engl J Med . 2012; 367 :1–3.2 [PubMed] [Google Scholar]