The New England Journal of Medicine - Nov 10, 2011 - Vol. 365 No. 19

99 Pages·2010·7.17 MB·English

by  M.D., Editor-in-Chief, Jeffrey M. Drazen

The NEW ENGLAND JOURNAL of MEDICINE
Perspective
november 10, 2011
Making Good on ACOs’ Promise — The Final Rule
for the Medicare Shared Savings Program
Donald M. Berwick, M.D.
uring my career as a practicing pediatrician, inadequate dissemination of us-
able clinical information, mis-
Dmy patients and I benefited from being part
aligned financial incentives, and
of a well-managed system of care, coordinated and in many cases, inertia — is rife
financed to support seamlessness and patient- with barriers to the coordinated
care that patients want, providers
centeredness. We had an elec- the U.S. health care system work want to give, and our unsustain-
tronic health record — always to deliver the highest-quality able system so desperately needs.
available and up to the minute health care they can. But as any To be sure, exactly this type of
— which reminded me when a health care provider will tell you, medicine is practiced every day
test or follow-up was due. For our system is full of roadblocks, in hundreds of places throughout
children with severe asthma, I red tape, and frustrations that the country. Innovative entrepre-
worked as part of a team, with a keep them from practicing the neurs and dedicated clinicians
home health nurse to teach skills type of medicine that most clini- have found ways to break down
and anticipate needs, an allergist cians envisioned when they chose barriers and redesign care to bet-
instantly available as a coach, their noble field. ter help their patients and com-
pharmacists to help plan care Physicians, nurses, and other munities. But bringing the best
and detect errors, and advanced health care professionals want the of our system to every commu-
practice nurses to ensure 24/7 ac- support required to work with en- nity in the country is the health
cess. As a result, my patients gaged patients to make the clini- care challenge of our time.
stayed out of emergency rooms cal decisions most appropriate to Eighteen months after Presi-
and hospital beds, remaining at their circumstances; to collabo- dent Barack Obama signed the
home and in school, where they rate with colleagues to provide a Affordable Care Act, the Depart-
belonged. Function improved and safe, seamless experience; and to ment of Health and Human Ser-
costs fell. be paid for keeping people well. vices (DHHS) has created a broad
The dedicated professionals in Instead, the status quo — with array of pathways for health care
n engl j med 365;19 nejm.org november 10, 2011 1753
The New England Journal of Medicine
PERSPECTIVE Final Rule for Medicare Shared Savings Program
Proposed Rule vs. Final Rule for Accountable Care Organizations (ACOs) in the Medicare Shared Savings Program.
Topic Proposed Rule Modifications in Final Rule
Transition to risk in ACOs could choose from two tracks, each entail- Remove two-sided risk from Track 1. Two tracks would still be
Track 1 ing a 3-year agreement. Track 1 would com- offered for ACOs at different levels of readiness, with one
prise 2 years of one-sided shared savings with a providing higher sharing rates for ACOs willing to also share
mandatory transition in year 3 to performance- in losses.
based risk under a two-sided model of shared
savings and losses. Track 2 would comprise
3 years all under the two-sided model.
Prospective vs. Retrospective assignment based on utilization of A preliminary prospective-assignment method with beneficiaries
r etrospective primary care services, with prospective identifi- identified quarterly; final reconciliation after each performance
cation of a benchmark population. year, made on the basis of patients served by the ACO.
Proposed measures 65 measures in 5 domains, including patient 33 measures in 4 domains. (Note: Claims-based measures not
to assess quality e xperience of care, utilization claims–based finalized to be used for ACO-monitoring purposes.)
measures, and measures assessing process Longer phase-in of measures over course of agreement: first year,
and outcomes. pay for reporting; second and third years, pay for reporting
Pay for full and accurate reporting first year, pay and performance.
for performance in subsequent years. Finalize as proposed.
Alignment of proposed measures with existing
quality programs and private-sector initiatives.
Sharing savings One-sided risk model: sharing beginning at sav- Share on first dollar for all ACOs in both models once minimum
ings of 2%, with some exceptions for small, savings rate has been achieved.
physician-only, and rural ACOs. Two-sided risk
model: sharing from first dollar.
Sharing beneficiary Claims data shared only for patients seen by ACO The ACO may contact beneficiaries from provided quarterly lists
identif ication primary care physician during performance to notify them of data sharing and opportunity to decline.
claims data year; beneficiaries given opportunity to decline
at the point of care.
Eligible entities The four groups specified by the Affordable Care In addition to groups included in the proposed rule, Federally
Act, as well as critical access hospitals paid Qualified Health Centers and Rural Health Clinics are also
through Method II, are eligible to form an ACO. eligible to both form and participate in an ACO. In order for
ACOs can be established with broad collabora- beneficiaries to be assigned on the basis of utilization of pri-
tion beyond these providers. mary care services, these organizations must provide a list
of practitioners who directly render primary care services in
their facilities so that beneficiaries can be assigned on the
basis of utilization of their services.
Start date Agreement for 3 years with uniform annual start Program established by January 1, 2012; first round of applica-
date; performance years based on calendar years. tions are due in early 2012. First ACO agreements start April
1, 2012, and July 1, 2012. ACOs will have agreements with a
first performance “year” of 18 or 21 months. ACOs starting
April 1, 2012, or July 1, 2012, have option of an interim pay-
ment if they report calendar year (CY) 2012 quality mea-
sures. ACO must report quality measures for CY 2013 to
qualify for first-performance-year shared savings.
Aggregate reports Reports will be provided at the beginning of each Additional reports will be provided quarterly.
and prel iminary performance year and include: name, date of
p rospective list birth, sex, and health insurance claim number.
Electronic health Aligning ACO requirements with EHR require- No longer a condition of participation. Retained EHR as quality
record (EHR) use ments, 50% of primary care physicians must be measure but weighted higher than any other measure for
defined as meaningful users by start of second quality-scoring purposes.
performance year.
Assignment process One-step assignment process: beneficiaries as- Two-step assignment process:
signed on the basis of a plurality of allowed Step 1: for beneficiaries who have received at least one primary
charges for primary care services rendered by care service from a physician, use plurality of allowed charges
primary care physicians (internal medicine, for primary care services rendered by primary care physicians.
general practice, family practice, and geriatric Step 2: for beneficiaries who have not received any primary care
medicine). services from a physician, use plurality of a llowed charges for
primary care services rendered by any other ACO professional.
Marketing guidelines All marketing materials must be approved by the “File and use” 5 days after submission and after certifying com-
Cent ers for Medicare and Medicaid Services pliance with marketing guidelines; CMS to provide approved
(CMS). language.
1754 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE Final Rule for Medicare Shared Savings Program
providers to begin — or in many mented by feedback at dozens of the Medicare program and im-
cases, accelerate — their care- informal listening sessions. The proves care or provides high-
improvement journey in partner- vast majority of the comments quality care; and creating a path-
ship with the Medicare and Med- we received were supportive of way for full participation of
icaid programs and in synergy the vision of the Shared Savings federally qualified health cen-
with the private sector. Today, the Program and optimistic about the ters and rural health clinics that
DHHS is taking its next major potential for ACOs to be a force provide a primary care safety net
step by finalizing the rules for for change in our broken health for Medicare beneficiaries in
the establishment of accountable care system. However, numerous underserved areas.
care organizations (ACOs) under suggestions were also offered for Taken together, these changes
the Medicare Shared Savings Pro- improvements to the proposed and numerous others create a more
gram created by Section 3022 of rule that would lead to a larger, feasible and attractive on-ramp for
the health care reform law. more pluralistic set of ACO partici- a diverse set of providers and or-
ACOs are voluntary groups of pants without compromising pa- ganizations to participate as ACOs.
physicians, hospitals, and other tient outcomes or choice. In partic- In addition, the Center for Medi-
health care providers that are ular, commenters asked CMS to care and Medicaid Innovation is
willing to assume responsibility reduce barriers to entry by stream- announcing today an advanced
for the care of a clearly defined lining governance and reporting payment initiative that will allow
population of Medicare beneficia- burdens on potential ACOs; im- small physician practices and ru-
ries attributed to them on the ba- prove the potential financial return ral community hospitals that face
sis of patients’ use of primary care for ACOs willing to make the nec- particular challenges in forming
services. If an ACO succeeds in essary, and often substantial, in- ACOs to receive up-front access to
both delivering high-quality care vestments to improve care; and needed capital.
or improving care and reducing ensure beneficiary protections. For established organizations
the cost of that care below what In response, CMS is making with a track record of providing
would otherwise have been ex- several significant changes in its robust coordinated care, the CMS
pected, it will share in the savings final rule to strengthen the ACO innovation center is offering a
it achieves for Medicare. program for providers and bene- pioneer ACO program designed
Under the ACO model, Medi- ficiaries alike (see table). Major to encourage and support the
care beneficiaries are still free to changes include providing better, next wave of innovation from van-
seek care from any Medicare pro- and more timely, information to guard organizations that are posi-
vider they wish. Indeed, Medicare ACOs at the outset of the perfor- tioned to help realize the full
beneficiaries should find their mance year through preliminary potential of the ACO model. And
care experience enhanced by a prospective alignment of benefi- for organizations and clinicians
program that supports providers ciaries (while retaining a retro- not yet prepared to make the tran-
in engaging with their patients spective reconciliation to ensure sition to ACOs, the DHHS is offer-
to deliver on the three-part aim: that ACOs are measured on the ing a menu of alternative options
better care for individuals, better basis of the patients they actually — including a comprehensive pri-
health for populations, and lower care for during the year); retaining mary care program, bundled pay-
cost growth through improve- a strong monitoring and quality- ments for care improvement, and
ments in care. measurement mechanism while a community-based transitional
The DHHS proposed its initial streamlining the metrics to focus care program — that all seek to
set of guidelines for ACOs on on what matters most, including provide the incentives and sup-
March 31, 2011, and sought wide- reducing the total number of ports necessary to move the main-
spread comment on both the di- quality measures by about half; stream of U.S. health care toward
rection and the details of this im- allowing start-up ACOs to choose accountable care.
portant new program for Medicare. a “savings only” track without fi- Whether provided through
We at the Centers for Medicare nancial risk during their initial ACOs or an alternative innova-
and Medicaid Services (CMS) re- contract period; sharing savings tion opportunity, coordinated care
ceived more than 1200 formal with successful ACOs on a “first is meant to allow providers to
comments from throughout the dollar” basis when the ACO break away from the tyranny of
health care community, supple- achieves meaningful savings for the 15-minute visit, instill a re-
n engl j med 365;19 nejm.org november 10, 2011 1755
The New England Journal of Medicine
PERSPECTIVE Final Rule for Medicare Shared Savings Program
newed sense of collegiality, and most promising path toward fi- toward patient-centered, coordi-
return to the type of medicine nancial sustainability and away nated care.
that patients and families want. from alternatives that shift costs Disclosure forms provided by the author
For patients, coordinated care onto patients, providers, and pri- are available with the full text of this arti-
cle at NEJM.org.
means more “quality time” with vate purchasers.
their physician and care team (a We believe that today’s ACO Dr. Berwick is the administrator of the Cen-
patient’s advocate in an increas- rule is the next step in our shared ters for Medicare and Medicaid Services,
Baltimore.
ingly complex medical system) and commitment to a better, more
more collaboration in leading a lasting health care system. We This article (10.1056/NEJMp1111671) was
healthy life. And for Medicare, look forward to being a trusted published on October 20, 2011, at NEJM.org.
coordinated care represents the partner in our nation’s journey Copyright © 2011 Massachusetts Medical Society.
Getting Moving on Patient Safety — Harnessing Electronic
Data for Safer Care
Ashish K. Jha, M.D., M.P.H., and David C. Classen, M.D.
ore than a decade ago, the sentially unchanged between 2000 their adherence to evidence-based
MInstitute of Medicine re- and 2008 at 10 North Carolina treatments such as the use of as-
leased its famous report To Err Is hospitals.1 A report from the In- pirin for patients with acute myo-
Human, which set an ambitious spector General of the Department cardial infarction. Efforts by the
agenda for the United States to of Health and Human Services Joint Commission for the Accred-
reduce the number of Americans (DHHS) revealed that Medicare itation of Healthcare Organiza-
who were hurt or killed by medi- patients experienced substantial tions to systematically measure
cal errors and adverse events. In harm in U.S. hospitals as recently performance and give feedback
response, a series of new initia- as 2008.2 Finally, Classen and col- to hospitals, coupled with sub-
tives was launched, including the leagues found that almost one in sequent efforts to publicly report
funding of new research on ways three patients are harmed during performance on these measures,
of making care safer and encour- their hospital stay and that tradi- led to dramatic improvements
agement of programs shielding tional approaches to measuring in compliance.4 In the few areas
health care providers from liabil- adverse events, whether using vol- of patient safety that have seen
ity if they reported adverse events. untary reporting or patient-safety demonstrable improvement (e.g.,
Federal agencies set up patient- indicators, substantially under- catheter-related bloodstream in-
safety organizations and estab- estimate the events’ frequency.3 If fections), the changes are due, at
lished ambitious patient-safety the United States has made prog- least in part, to robust measure-
goals; accrediting organizations ress in patient safety, it has been ment programs, such as those run
set aggressive patient-safety stan- inadequate. by the Centers for Disease Control
dards; and providers hired pa- The primary reason for insuffi- and Prevention. In other areas, in-
tient-safety officers and imple- cient progress is the lack of a ro- adequate measures have hindered
mented numerous patient-safety bust measurement program: there progress, and patients continue to
initiatives. are still no nationally agreed-on suffer from the consequences of
So what are the fruits of these methods for systematically iden- unsafe care.
efforts? Recently, we have received tifying, tracking, and reporting Although there is a shortage of
some deeply disappointing news: adverse events. Here, the patient- good patient-safety metrics, poor-
three studies have called into safety movement can learn from quality measures are plentiful.
question whether we’ve made any the quality-improvement efforts The best known among these are
progress at all. Landrigan et al. that predate it. In the 1990s, patient-safety indicators, which
found that rates of injury due to emerging evidence suggested that use billing data to identify poten-
medical error had remained es- providers were inconsistent in tial complications during a hos-
1756 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE Final Rule for Medicare Shared Savings Program
newed sense of collegiality, and most promising path toward fi- toward patient-centered, coordi-
return to the type of medicine nancial sustainability and away nated care.
that patients and families want. from alternatives that shift costs Disclosure forms provided by the author
For patients, coordinated care onto patients, providers, and pri- are available with the full text of this arti-
cle at NEJM.org.
means more “quality time” with vate purchasers.
their physician and care team (a We believe that today’s ACO Dr. Berwick is the administrator of the Cen-
patient’s advocate in an increas- rule is the next step in our shared ters for Medicare and Medicaid Services,
Baltimore.
ingly complex medical system) and commitment to a better, more
more collaboration in leading a lasting health care system. We This article (10.1056/NEJMp1111671) was
healthy life. And for Medicare, look forward to being a trusted published on October 20, 2011, at NEJM.org.
coordinated care represents the partner in our nation’s journey Copyright © 2011 Massachusetts Medical Society.
Getting Moving on Patient Safety — Harnessing Electronic
Data for Safer Care
Ashish K. Jha, M.D., M.P.H., and David C. Classen, M.D.
ore than a decade ago, the sentially unchanged between 2000 their adherence to evidence-based
MInstitute of Medicine re- and 2008 at 10 North Carolina treatments such as the use of as-
leased its famous report To Err Is hospitals.1 A report from the In- pirin for patients with acute myo-
Human, which set an ambitious spector General of the Department cardial infarction. Efforts by the
agenda for the United States to of Health and Human Services Joint Commission for the Accred-
reduce the number of Americans (DHHS) revealed that Medicare itation of Healthcare Organiza-
who were hurt or killed by medi- patients experienced substantial tions to systematically measure
cal errors and adverse events. In harm in U.S. hospitals as recently performance and give feedback
response, a series of new initia- as 2008.2 Finally, Classen and col- to hospitals, coupled with sub-
tives was launched, including the leagues found that almost one in sequent efforts to publicly report
funding of new research on ways three patients are harmed during performance on these measures,
of making care safer and encour- their hospital stay and that tradi- led to dramatic improvements
agement of programs shielding tional approaches to measuring in compliance.4 In the few areas
health care providers from liabil- adverse events, whether using vol- of patient safety that have seen
ity if they reported adverse events. untary reporting or patient-safety demonstrable improvement (e.g.,
Federal agencies set up patient- indicators, substantially under- catheter-related bloodstream in-
safety organizations and estab- estimate the events’ frequency.3 If fections), the changes are due, at
lished ambitious patient-safety the United States has made prog- least in part, to robust measure-
goals; accrediting organizations ress in patient safety, it has been ment programs, such as those run
set aggressive patient-safety stan- inadequate. by the Centers for Disease Control
dards; and providers hired pa- The primary reason for insuffi- and Prevention. In other areas, in-
tient-safety officers and imple- cient progress is the lack of a ro- adequate measures have hindered
mented numerous patient-safety bust measurement program: there progress, and patients continue to
initiatives. are still no nationally agreed-on suffer from the consequences of
So what are the fruits of these methods for systematically iden- unsafe care.
efforts? Recently, we have received tifying, tracking, and reporting Although there is a shortage of
some deeply disappointing news: adverse events. Here, the patient- good patient-safety metrics, poor-
three studies have called into safety movement can learn from quality measures are plentiful.
question whether we’ve made any the quality-improvement efforts The best known among these are
progress at all. Landrigan et al. that predate it. In the 1990s, patient-safety indicators, which
found that rates of injury due to emerging evidence suggested that use billing data to identify poten-
medical error had remained es- providers were inconsistent in tial complications during a hos-
1756 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE Harnessing Electronic Data for Safer Care
pitalization. They generally have ingful users” of electronic health taxpayer subsidies for EHR adop-
poor sensitivity and specificity, records (EHRs). EHRs can sys- tion was sold to the U.S. public
and their utility varies with hos- tematically measure patient safety, principally as a way of making
3
pitals’ billing practices. And be- turning a laborious, manually in- health care safer. The current
cause data on them are collected tensive, and expensive process of EHR systems, if implemented
in a post hoc fashion, they fail to sifting through medical records well, may have a modest effect
engage clinicians at the time of to identify adverse events into an on safety. Requiring the presence
care delivery — and data are gen- automated one that is efficient, and use of a safety-measurement
erally unavailable to providers for consistent, and affordable. Al- module for identifying and track-
review until years after the care is though the technology is already ing adverse events would provide
delivered. In an attempt to make available, most EHRs today are a critical signal to providers that
patient-safety measures more vis- not built with this capability in monitoring adverse events is es-
ible, the Centers for Medicare and mind — and it won’t be easy (or sential. Such systems would pro-
Medicaid Services (CMS) now cheap) to retrofit EHR systems vide information to hospitals on
makes hospital performance data later. Without an explicit commit- their performance relative to their
on certain patient-safety indica- ment on the part of EHR vendors peers and their progress toward
5
tors publicly available. However, to build systems that can system- the goal of causing no harm.
these inadequate measures of atically track adverse events, most Most important, it would allow
safety, even if delivered to hospi- EHRs will fail to do it adequate- them to track the effects of their
tals more quickly, are unlikely to ly, if at all. However, the federal interventions and determine which
engage front-line clinicians in ac- government, which creates the efforts worked and which ones
tivities that will make care safer. meaningful use criteria, could in- didn’t. Data from such EHR sur-
Another approach to safety clude the ability to effectively veillance systems could have pa-
measurement has been the use of measure and report adverse event tient identifiers removed and be
voluntary (or occasionally, manda- rates as a “core” requirement of pooled across many sites, which
tory) reporting of adverse events. meaningful use. By staking out would increase the depth and
These efforts sporadically yield this ground, CMS can signal to breadth of the possible analyses
important insights — but gener- vendors the importance of in- and lead to new insight into de-
ally have very low sensitivity (most cluding such a capability in every livering safer care.
adverse events are never report- EHR sold in the United States. If The U.S. health care system is
ed), which makes it difficult for CMS chooses to use EHR-derived at a crossroads when it comes to
provider organizations to know safety measures for public report- ensuring patient safety and earn-
whether they’re making progress. ing or pay for performance, these ing the trust of the public. Our
Finally, some have used the “trig- metrics will need further valida- inadequate progress since To Err
ger tools” method, which, though tion, a lengthy process, but the Is Human is disheartening, but we
not extensively validated, appears agency can expedite the activi- have an opportunity to right the
to be sensitive in detecting ad- ties needed to ensure that we ship. By making systematic mea-
3
verse events. However, this ap- have validated measures quickly. surement of adverse events a requi-
proach, used primarily in research, Even without these additional site function of the EHRs that are
is resource-intensive to implement validation efforts, simply provid- eligible for financial incentives,
and has gained little traction ing better-quality EHR-derived the federal government can change
among providers as an ongoing safety data to physicians and the way safety is measured and
approach to monitoring safety. hospitals can have a profound improved throughout the health
Despite these challenges, we effect on patient-safety activities care system. Without these data,
currently have an opportunity to throughout the country. we are likely to repeat our recent
turn the tide. The Health Infor- We recognize that EHR ven- experience: good intentions, a lot
mation Technology for Economic dors face competing demands, of effort, and little demonstrable
and Clinical Health (HITECH) Act and many advocacy groups are benefit. According to IOM esti-
of 2009 provides, through CMS, clamoring to have particular mates, as many as a million
financial incentives for physicians functions included in meaning- Americans may have died owing
and hospitals to become “mean- ful use. But the $30 billion in to adverse events in U.S. hospitals
n engl j med 365;19 nejm.org november 10, 2011 1757
The New England Journal of Medicine
PERSPECTIVE Harnessing Electronic Data for Safer Care
over the past decade. We must do Health, and the VA Boston Healthcare Sys- 3. Classen DC, Resar R, Griffin F, et al.
better over the next decade. eterm S c—ie nbcoetsh Cino Brpoosrtaotnio (nA .aKn.Jd.) ;t haen dU Cnoivmerpsuitty- ‘eGvelonbtsa li nt rhigogsperit atlos oml’a ys hboe wtesn tihmate sa gdrveeartseer
EHRs can improve the safety and of Utah — both in Salt Lake City (D.C.C.). than previously measured. Health Aff (Mill-
culture of U.S. health care, but wood) 2011;30:581-9. [Erratum, Health Aff
only if the federal government, 1. Landrigan CP, Parry GJ, Bones CB, Hack- (4M. Cillhwaososdin) 2M01R1,; 30L:o1e2b1 7.J]M, Schmaltz SP,
as the nation’s largest health care barth AD, Goldmann DA, Sharek PJ. Tempo- W achter RM. Accountability measures —
p
ouayse arb, oduetm imonpsrtorvaitnegs tphaatite intt isa sfetryi-. 2rfra1ol2 mt4r- e3mn4de. sd[ Eiicnrar alr atcutaemrse ,. o NNf p EEanntiggelln Jt MhMaerdmd 2r0e1s0u;l3ti6n3g: up5.sr oiHnvgeo msmpeeintaats.l u Nrqe Emunaeglinlt Jty Mtoie npditr 2ioa0mt1iv0oe;t3es6: q3u:o6au8li3ttc-yo8 i.mme-
Disclosure forms provided by the authors 2573.] measures. Baltimore: Centers for Medicare
are available with the full text of this article 2. Levinson D. Adverse events in hospitals: & Medicaid Services, 2011 (https://www
at NEJM.org. national incidence among Medicare benefi- .cms.gov/HospitalQualityInits/
ciaries. Washington, DC: Office of the In- 20_OutcomeMeasures.asp).
From the Department of Health Policy and spector General, Department of Health and
Management, Harvard School of Public Human Services, 2010. Copyright © 2011 Massachusetts Medical Society.
Evidence-Based Medicine in the EMR Era
Jennifer Frankovich, M.D., Christopher A. Longhurst, M.D., and Scott M. Sutherland, M.D.
any physicians take great though anticoagulation is not pability.1 Through STRIDE, we
Mpride in the practice of evi- standard practice for children could rapidly review data on an
dence-based medicine. Modern with SLE even when they’re criti- SLE cohort that included pediatric
medical education emphasizes the cally ill, these additional factors patients with SLE cared for by
value of the randomized, con- put our patient at potential risk clinicians in our division between
trolled trial, and we learn early for thrombosis, and we consid- October 2004 and July 2009. This
on not to rely on anecdotal evi- ered anticoagulation. However, we “electronic cohort” was originally
dence. But the application of such were unable to find studies per- created for use in studying com-
superior evidence, however admi- taining to anticoagulation in our plications associated with pediat-
rable the ambition, can be con- patient’s situation and were there- ric SLE and exists under a proto-
strained by trials’ strict inclusion fore reluctant to pursue that col approved by our institutional
and exclusion criteria — or the course, given the risk of bleeding. review board.
complete absence of a relevant A survey of our pediatric rheu- Of the 98 patients in our pedi-
trial. For those of us practicing matology colleagues — a review atric lupus cohort, 10 patients de-
pediatric medicine, this reality is of our collective Level V evidence, veloped thrombosis, documented
all too familiar. In such situa- so to speak — was equally fruit- in the EMR, while they were acute-
tions, we are used to relying on less and failed to produce a con- ly ill. The prevalence was higher
evidence at Levels III through V — sensus. among patients who had persis-
expert opinion — or resorting to Without clear evidence to guide tent nephrotic-range proteinuria
anecdotal evidence. What should us and needing to make a deci- and pancreatitis (see table). As
we do, though, when there aren’t sion swiftly, we turned to a new compared with our patients with
even meager data available and approach, using the data captured lupus who did not have these
we don’t have a single anecdote in our institution’s electronic med- risk factors, the risk of thrombo-
on which to draw? ical record (EMR) and an innova- sis was 14.7 (95% confidence in-
We recently found ourselves in tive research data warehouse. The terval [CI], 3.3 to 96) among pa-
such a situation as we admitted platform, called the Stanford tients with persistent nephrosis
to our service a 13-year-old girl Translational Research Integrated and 11.8 (95% CI, 3.8 to 27)
with systemic lupus erythemato- Database Environment (STRIDE), among those with pancreatitis.
sus (SLE). Our patient’s presenta- acquires and stores all patient This automated cohort review was
tion was complicated by nephrotic- data contained in the EMR at conducted in less than 4 hours
range proteinuria, antiphospholipid our hospital and provides imme- by a single clinician. On the ba-
antibodies, and pancreatitis. Al- diate advanced text searching ca- sis of this real-time, informatics-
1758 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE Harnessing Electronic Data for Safer Care
over the past decade. We must do Health, and the VA Boston Healthcare Sys- 3. Classen DC, Resar R, Griffin F, et al.
better over the next decade. eterm S c—ie nbcoetsh Cino Brpoosrtaotnio (nA .aKn.Jd.) ;t haen dU Cnoivmerpsuitty- ‘eGvelonbtsa li nt rhigogsperit atlos oml’a ys hboe wtesn tihmate sa gdrveeartseer
EHRs can improve the safety and of Utah — both in Salt Lake City (D.C.C.). than previously measured. Health Aff (Mill-
culture of U.S. health care, but wood) 2011;30:581-9. [Erratum, Health Aff
only if the federal government, 1. Landrigan CP, Parry GJ, Bones CB, Hack- (4M. Cillhwaososdin) 2M01R1,; 30L:o1e2b1 7.J]M, Schmaltz SP,
as the nation’s largest health care barth AD, Goldmann DA, Sharek PJ. Tempo- W achter RM. Accountability measures —
p
ouayse arb, oduetm imonpsrtorvaitnegs tphaatite intt isa sfetryi-. 2rfra1ol2 mt4r- e3mn4de. sd[ Eiicnrar alr atcutaemrse ,. o NNf p EEanntiggelln Jt MhMaerdmd 2r0e1s0u;l3ti6n3g: up5.sr oiHnvgeo msmpeeintaats.l u Nrqe Emunaeglinlt Jty Mtoie npditr 2ioa0mt1iv0oe;t3es6: q3u:o6au8li3ttc-yo8 i.mme-
Disclosure forms provided by the authors 2573.] measures. Baltimore: Centers for Medicare
are available with the full text of this article 2. Levinson D. Adverse events in hospitals: & Medicaid Services, 2011 (https://www
at NEJM.org. national incidence among Medicare benefi- .cms.gov/HospitalQualityInits/
ciaries. Washington, DC: Office of the In- 20_OutcomeMeasures.asp).
From the Department of Health Policy and spector General, Department of Health and
Management, Harvard School of Public Human Services, 2010. Copyright © 2011 Massachusetts Medical Society.
Evidence-Based Medicine in the EMR Era
Jennifer Frankovich, M.D., Christopher A. Longhurst, M.D., and Scott M. Sutherland, M.D.
any physicians take great though anticoagulation is not pability.1 Through STRIDE, we
Mpride in the practice of evi- standard practice for children could rapidly review data on an
dence-based medicine. Modern with SLE even when they’re criti- SLE cohort that included pediatric
medical education emphasizes the cally ill, these additional factors patients with SLE cared for by
value of the randomized, con- put our patient at potential risk clinicians in our division between
trolled trial, and we learn early for thrombosis, and we consid- October 2004 and July 2009. This
on not to rely on anecdotal evi- ered anticoagulation. However, we “electronic cohort” was originally
dence. But the application of such were unable to find studies per- created for use in studying com-
superior evidence, however admi- taining to anticoagulation in our plications associated with pediat-
rable the ambition, can be con- patient’s situation and were there- ric SLE and exists under a proto-
strained by trials’ strict inclusion fore reluctant to pursue that col approved by our institutional
and exclusion criteria — or the course, given the risk of bleeding. review board.
complete absence of a relevant A survey of our pediatric rheu- Of the 98 patients in our pedi-
trial. For those of us practicing matology colleagues — a review atric lupus cohort, 10 patients de-
pediatric medicine, this reality is of our collective Level V evidence, veloped thrombosis, documented
all too familiar. In such situa- so to speak — was equally fruit- in the EMR, while they were acute-
tions, we are used to relying on less and failed to produce a con- ly ill. The prevalence was higher
evidence at Levels III through V — sensus. among patients who had persis-
expert opinion — or resorting to Without clear evidence to guide tent nephrotic-range proteinuria
anecdotal evidence. What should us and needing to make a deci- and pancreatitis (see table). As
we do, though, when there aren’t sion swiftly, we turned to a new compared with our patients with
even meager data available and approach, using the data captured lupus who did not have these
we don’t have a single anecdote in our institution’s electronic med- risk factors, the risk of thrombo-
on which to draw? ical record (EMR) and an innova- sis was 14.7 (95% confidence in-
We recently found ourselves in tive research data warehouse. The terval [CI], 3.3 to 96) among pa-
such a situation as we admitted platform, called the Stanford tients with persistent nephrosis
to our service a 13-year-old girl Translational Research Integrated and 11.8 (95% CI, 3.8 to 27)
with systemic lupus erythemato- Database Environment (STRIDE), among those with pancreatitis.
sus (SLE). Our patient’s presenta- acquires and stores all patient This automated cohort review was
tion was complicated by nephrotic- data contained in the EMR at conducted in less than 4 hours
range proteinuria, antiphospholipid our hospital and provides imme- by a single clinician. On the ba-
antibodies, and pancreatitis. Al- diate advanced text searching ca- sis of this real-time, informatics-
1758 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE Evidence-Based Medicine in the EMR Era
Results of Electronic Search of Patient Medical Records (for a Cohort of 98 Pediatric Patients with Lupus) Focused
on Risk Factors for Thrombosis Relevant to Our 13-Year-Old Patient with Systemic Lupus Erythematosus.*
Keywords Used to Conduct Prevalence Relative Risk
Outcome or Risk Factor E xpedited Electronic Search of Thrombosis (95% CI)
no./total no (%)
Outcome — thrombosis “Thrombus,” “Thrombosis,” 10/98 (10) Not applicable
“Blood clot”
Thrombosis risk factor
Heavy proteinuria (>2.5 g per deciliter)
Present at any time “Nephrosis,” “Nephrotic,” 8/36 (22) 7.8 (1.7–50)
“Proteinuria”
Present >60 days “Urine protein” 7/23 (30) 14.7 (3.3–96)
Pancreatitis “Pancreatitis,” “Lipase” 5/8 (63) 11.8 (3.8–27)
Antiphospholipid antibodies “Aspirin” 6/51 (12) 1.0 (0.3–3.7)
* In all cases, the sentences surrounding the keywords were manually reviewed to determine their relevance to our patient. Pancre-
atitis was defined as an elevated lipase level (twice the upper limit of normal) coexisting with abdominal pain. We used the word
“aspirin” as a proxy for antiphospholipid antibodies, since it is standard practice at our institution to give all patients with these
antibodies aspirin; if “aspirin” was found in the chart, than antiphospholipid-antibody status was confirmed by investigating
the laboratory results.
5
enabled data analysis, we made sion making has already trans- by intelligence.” In the practice
4
the decision to give our patient formed other industries, and the of medicine, one can’t do better
anticoagulants within 24 hours growing prevalence of EMRs along than that.
after admission. with the development of sophisti- Disclosure forms provided by the authors
are available with the full text of this article
Our case is but one example cated tools for real-time analysis
at NEJM.org.
of a situation in which the exist- of deidentified data sets will no
From the Division of Rheumatology (J.F.),
ing literature is insufficient to doubt advance the use of this data-
the Division of Systems Medicine (C.A.L.),
guide the clinical care of a pa- driven approach to health care de- and the Division of Nephrology (S.M.S.),
Department of Pediatrics, Stanford Univer-
tient. But it illustrates a novel livery. We look forward to a fu-
sity School of Medicine, Palo Alto, CA.
process that is likely to become ture in which health information
This article (10.1056/NEJMp1108726) was
much more standard with the systems help physicians learn from
published on November 2, 2011, at NEJM.org.
widespread adoption of EMRs every patient at every visit and
1. Lowe HJ, Ferris TA, Hernandez PM, Weber
and more sophisticated informat- close the feedback loop for clini-
SC. STRIDE — an integrated standards-based
ics tools. Although many other cal decision making in real time. translational research informatics platform.
groups have highlighted the sec- Did we make the correct deci- AMIA Annu Symp Proc 2009;Nov 14:391-5.
2. Prokosch HU, Ganslandt T. Perspectives
ondary use of EMR data for clini- sion for our patient? Thrombosis
for medical informatics: reusing the elec-
2,3
cal research, we have now seen did not develop, and the patient tronic medical record for clinical research.
how the same approach can be did not have any sequelae related Methods Inf Med 2009;48:38-44.
3. Gunn PW, Hansen ML, Kaelber DC. Under-
used to guide real-time clinical to her anticoagulation; truthfully,
diagnosis of pediatric hypertension — an
decisions. The rapid electronic though, we may never really example of a new era of clinical research en-
chart review and analysis were know. We will, however, know abled by electronic medical records. AMIA
Annu Symp Proc 2007;October 11:966.
not only feasible, but also more that we made the decision on the
4. Halevy A, Norvig P, Pereira F. The Unrea-
helpful and accurate than physi- basis of the best data available sonable Effectiveness of Data. IEEE Intelli-
cian recollection and pooled col- — acting, as the fictional detec- gent Systems, March/April 2009:8-12.
5. Stout R. In the best families. New York:
league opinion. Such real-time tive Nero Wolfe would say, “in
Viking Press, 1950:71.
availability of data to guide deci- the light of experience as guided Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011 1759
The New England Journal of Medicine
PERSPECTIVE Cost-Effectiveness and U.S. Vaccination Policy
The Role of Cost-Effectiveness in U.S. Vaccination Policy
Jane J. Kim, Ph.D.
accination policy is driven by organizations, such as the Amer- ing its cost-effectiveness. Since
Vseveral factors, including vac- ican Academy of Family Physi- 2007, routine HPV vaccination has
cine safety and efficacy, avertable cians and the American Academy been recommended for girls 11 to
disease burden, acceptability, and of Pediatrics. 12 years of age (and as early as
societal value. One measure of Historically, ACIP recommen- 9 years), with “catch-up” vaccina-
value is an intervention’s cost- dations have influenced coverage tion recommended up to the age
effectiveness, defined as the addi- decisions by both private and pub- of 26, despite evidence of rapidly
tional cost required per additional lic insurers. Through a separate diminishing marginal returns and
unit of health benefit produced process, the ACIP also determines decreasing cost-effectiveness af-
as compared with the next-most- what vaccines are to be covered by ter 21 years of age.2
effective alternative. It is impor- the federal Vaccines for Children After the Food and Drug Ad-
tant to differentiate cost-effective- (VFC) program, which covers chil- ministration (FDA) approved the
ness (value for money) from dren who are Medicaid-eligible, quadrivalent HPV (HPV4) vaccine
affordability (financial resources uninsured or underinsured, or for males in 2009, the ACIP voted
required); indeed, interventions American Indians or Alaska na- for “permissive” — but not rou-
with high value may not always tives up to the age of 18. With tine — use of it in boys and men
be affordable. Although informa- nearly 50% of U.S. children eli- 9 to 26 years of age for preven-
tion on the cost-effectiveness of gible for VFC coverage,1 the ACIP tion of genital warts. Despite this
health interventions is increas- faces dual pressures: it must max- less enthusiastic stance, the ACIP
ingly being used in health policy imize underserved children’s ac- voted in favor of VFC coverage
globally, the extent to which this cess to vaccines while selecting for eligible males 9 to 18 years of
information influences decisions vaccines that provide the most age. The committee was persuad-
varies by country. For example, bang for the buck. This pressure ed not to recommend routine
the governments in Britain and will increase with the rollout of male HPV vaccination in part by
Australia explicitly and routinely the Affordable Care Act, which evidence that it may not be cost-
incorporate findings from cost- mandates coverage of all ACIP- effective, especially if vaccine up-
effectiveness analyses into cover- recommended childhood immu- take in girls and young women is
age and reimbursement decisions; nizations. high, given the sexual transmis-
in contrast, in the United States, With low cost and high effica- sion of HPV infections and ex-
it has been essentially taboo for cy, many vaccines are estimated pected herd-immunity benefits
anyone in the public sector to re- to be cost-saving — the up-front through female-only vaccination.
fer explicitly to cost as a factor in expenditure for vaccination is Recent data on uptake among
health decisions. entirely offset by costs averted adolescent girls, however, show
One exception is the Advisory through disease prevention. How- less than 50% completion of the
Committee on Immunization ever, newly licensed and expensive three-dose series, suggesting that
Practices (ACIP), an independent vaccines, such as those against HPV vaccination of boys may be
expert advisory board that for- human papillomavirus (HPV, the cost-effective at this time. Further-
mally includes cost-effectiveness virus causally linked to cervical more, since the 2009 guidelines
among the types of evidence it cancer) and meningococcal dis- were issued, the indications for
considers when making vaccine- ease, are being considered for use HPV4 have expanded to include
policy recommendations to the in ways that raise questions re- prevention of anal cancers. Rou-
Centers for Disease Control and garding their overall public health tine male HPV vaccination, espe-
Prevention (CDC). The ACIP value as estimated in cost-effec- cially if targeted at an early age,
strives to be transparent and bal- tiveness analyses. when the vaccines are expected
anced, inviting perspectives from In late October, the ACIP is ex- to have highest benefit, would
stakeholders ranging from scien- pected to vote on routine HPV maximize protection for men
tists to patient groups, and tries vaccination in boys and young who have sex with men, a group
to harmonize its recommenda- men and to discuss meningococ- at high risk for HPV-related can-
tions with those of professional cal vaccination in infants, includ- cers that would receive little
1760 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine