PEDSnet: How A Prototype Pediatric Learning Health System Is Being Expanded Into A National Network

Abstract

Except for a few conditions, pediatric disorders are rare diseases. Because of this, no single institution has enough patients to generate adequate sample sizes to produce generalizable knowledge. Aggregating electronic clinical data from millions of children across many pediatric institutions holds the promise of producing sufficiently large data sets to accelerate knowledge discovery. However, without deliberately embedding these data in a pediatric learning health system (defined as a health care organization that is purposefully designed to produce research in routine care settings and implement evidence at the point of care), efforts to act on this new knowledge, reducing the distress and suffering that children experience when sick, will be ineffective. In this article we discuss a prototype pediatric learning health system, ImproveCareNow, for children with inflammatory bowel disease. This prototype is being scaled up to create PEDSnet, a national network that will support the efficient conduct of clinical trials, observational research, and quality improvement across diseases, specialties, and institutions.

Full text

At the Intersection of Health, Health Care and Policy
doi: 10.1377/hlthaff.2014.0127
, 33, no.7 (2014):1171-1177Health Affairs
A National Network
PEDSnet: How A Prototype Pediatric Learning Health System Is Being Expanded Into
Christopher B. Forrest, Peter Margolis, Michael Seid and Richard B. Colletti
Cite this article as:
http://content.healthaffairs.org/content/33/7/1171.full.html
available at:
The online version of this article, along with updated information and services, is
For Reprints, Links & Permissions: http://healthaffairs.org/1340_reprints.php http://content.healthaffairs.org/subscriptions/etoc.dtlE-mail Alerts : http://content.healthaffairs.org/subscriptions/online.shtmlTo Subscribe:
written permission from the Publisher. All rights reserved.
mechanical, including photocopying or by information storage or retrieval systems, without prior
may be reproduced, displayed, or transmitted in any form or by any means, electronic orAffairs HealthFoundation. As provided by United States copyright law (Title 17, U.S. Code), no part of
by Project HOPE - The People-to-People Health2014Bethesda, MD 20814-6133. Copyright ©
is published monthly by Project HOPE at 7500 Old Georgetown Road, Suite 600,Health Affairs
Not for commercial use or unauthorized distribution
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

By Christopher B. Forrest, Peter Margolis, Michael Seid, and Richard B. Colletti
PEDSnet: How A Prototype
Pediatric Learning Health System
Is Being Expanded Into A National
Network
ABSTRACT
Except for a few conditions, pediatric disorders are rare
diseases. Because of this, no single institution has enough patients to
generate adequate sample sizes to produce generalizable knowledge.
Aggregating electronic clinical data from millions of children across
many pediatric institutions holds the promise of producing sufficiently
large data sets to accelerate knowledge discovery. However, without
deliberately embedding these data in a pediatric learning health system
(defined as a health care organization that is purposefully designed to
produce research in routine care settings and implement evidence at the
point of care), efforts to act on this new knowledge, reducing the distress
and suffering that children experience when sick, will be ineffective. In
this article we discuss a prototype pediatric learning health system,
ImproveCareNow, for children with inflammatory bowel disease. This
prototype is being scaled up to create PEDSnet, a national network that
will support the efficient conduct of clinical trials, observational research,
and quality improvement across diseases, specialties, and institutions.
Except for a handful of disorders
(such as asthma, attention deficit
hyperactivity disorder, autism spec-
trum disorder, obesity, and acute
infections), virtually all childhood
conditions are rare diseases. As a result of the
infrequent occurrence of pediatric disorders, no
single institution has enough patients to gener-
ate adequate sample sizes to produce generaliz-
able knowledge. This epidemiological reality has
resulted in the reliance on “hand-me-down”re-
sults from research done among adults1and a
sparse pediatric evidence base.
One approach to overcoming these sample-
size limitations is to invest in pediatric big data.
By big data, we refer to large amounts of clinical
data obtained from electronic health records
(EHRs); patient-generated data sources; and
biospecimens aggregated from multiple institu-
tions and thousands, even millions, of children.
Perhaps the most promising approach for de-
veloping big data and then using it for research
and quality improvement is the formation of a
pediatric learning health system. We define a
learning health system as comprising four essen-
tial attributes: an organizational architecture
that facilitates formation of communities of pa-
tients, families, front-line clinicians, research-
ers, and health system leaders who collaborate
to produce and use pediatric big data; large elec-
tronic health and health care data sets (big data);
quality improvement for each patient at the
point of care brought about by the integration
of relevant new knowledge generated through
research;2–5and observational research and clin-
ical trials done in routine clinical care settings.
We present a model for a pediatric learning
health system that purposefully integrates re-
search and quality improvement as part of one
system of care. This model takes advantage of
pediatric big data; engages a community of
stakeholders pursuing a common purpose; and
doi: 10.1377/hlthaff.2014.0127
HEALTH AFFAIRS 33,
NO. 7 (2014): 1171–1177
©2014 Project HOPE—
The People-to-People Health
Foundation, Inc.
Christopher B. Forrest
(forrestc@email.chop.edu) is a
professor of pediatrics at the
Children’sHospitalof
Philadelphia and the
University of Pennsylvania as
well as principal investigator
for the PEDSnet learning
health system, all in
Philadelphia.
Peter Margolis is a professor
of pediatrics and director of
research at the James M.
Anderson Center for Health
Systems Excellence at the
Cincinnati Children’sHospital
Medical Center, in Ohio, and
scientific director of the
ImproveCareNow network.
Michael Seid is director of
health outcomes and quality
of care research in the
Division of Pulmonary
Medicine and a professor of
pediatrics at the Cincinnati
Children’sHospitalMedical
Center.
Richard B. Colletti is a
professor of pediatrics at the
University of Vermont College
of Medicine, in Burlington, and
network director of the
ImproveCareNow network.
July 2014 33:7 Health Affairs 1171
Building Rapid-Learning Systems
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

distributes the work of creating knowledge and
know-how across a broad population of patients
and families, clinicians, and researchers orga-
nized in networks. We illustrate the model by
describing a prototype pediatric learning health
system called the ImproveCareNow network for
children with inflammatory bowel disease and
sketch how the prototype is being scaled up to
create a national pediatric learning health sys-
tem called PEDSnet.
The Learning Health System
In today’s health system, research is done by
scientists, improvement is implemented by qual-
ity specialists, patient care is administered by
clinicians, and management is handled by health
care executives. Patients are relatively passive
consumers of these services. Communication
across these communities is scant, knowledge
is siloed, and diffusion of evidence of best prac-
tices to achieve good outcomes into clinical prac-
tice is unacceptably slow.6
We envision a transformation from the current
state to one in which research, improvement,
management, and patient care are intentionally
integrated. In such a health system, “learning
while doing”7is the default, thus ensuring that
the right care is provided to the right child at the
right time, every time.
The learning health system is more than big
data and big clinical trials. The system is predi-
cated on the active collaboration of all members
of the system, from patients to clinicians to
health system leaders, and success is defined
by the impact of the system on the health
and lives of patients. Each of these four compo-
nents—engaged communities, big data, quality
improvement, and research—should be consid-
ered within an overall system design. Clinical
research focuses on “what works.”Implementa-
tion research focuses on “how to make it work.”8
Both are needed as part of the learning health
system.
The engine of the learning health system is the
learning cycle. The cycle begins with patient-
clinician interactions at any location where care
is provided. Data from these interactions are
routinely captured electronically and combined
across patients, time, and settings, allowing for
comparative studies. Findings from these com-
parative studies coupled with existing biomedi-
cal research add to the knowledge network—the
database of current knowledge that is relevant to
improving the health and care of patients and
populations. Quality improvement methods,
such as previsit planning individualized to each
patient, are used to ensure that this evidence is
applied to meet the needs of patients. When the
learning cycle is fully operational, research in-
fluences practice and practice influenc-
es research in a virtuous cycle.5
A Pediatric Learning Health System
Prototype: ImproveCareNow
The concept of the learning health system is typ-
ically applied to a single health care organiza-
tion.5In pediatrics and rare diseases, learning
must occur among organizations and patients
who are dispersed across geography and institu-
tions to create a distributed learning health sys-
tem. There are few examples of such distributed
learning health systems involving more than one
organizationally distinct institution. One exam-
ple of an operational prototype is ImproveCare
Now. Established in 2007, ImproveCareNow was
launched to advance the quality of care for chil-
dren with Crohn’s disease and ulcerative colitis:
severe immunologic diseases referred to as in-
flammatory bowel disease that result in abdomi-
nal pain, diarrhea, bloody stools, weight loss,
stunted growth, and fatigue. Children with in-
flammatory bowel disease endure emergency de-
partment visits, colonoscopies, and x-rays, and
they risk hospitalization and surgery (such as
bowel resection and colectomy). Treatment
may require numerous daily pills and regular
intravenous infusions of medication. When pa-
tients are in remission (no symptoms, feeling
well, and fully active), they can lead normal lives.
Since its inception, the network has grown
from eight to sixty-six pediatric gastroenterolo-
gy care centers, now including approximately
35 percent of all US children with inflammatory
bowel disease.Without the addition of new drugs
to the therapeutic options available to patients
and clinicians, ImproveCareNow increased the
proportion of patients in remission from 55 per-
cent to 77 percent (Exhibit 1), markedly reducing
the burden of suffering. The ImproveCareNow
innovations are described in more detail below.
For the first few years of its existence, Improve
CareNow involved only clinicians and their care
centers. Over the past four years and with fund-
ing from a National Institutes of Health Trans-
formative Research grant, ImproveCareNow has
worked with the C3N Project (c3nproject.org) to
develop infrastructure and methods that are
needed to transition from a quality improvement
collaborative into a disease-specific learning
health system.9With funding from the Agency
for Healthcare Research and Quality’s Enhanced
Registry program, ImproveCareNow developed a
digital architecture based on EHRs and scientific
infrastructure for conducting comparative effec-
tiveness research.
Building Rapid-Learning Systems
1172 Health Affairs July 2014 33:7
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

Learning As A Community
ImproveCareNow has implemented a model for
engaging participants called actor-oriented col-
laboration.10 The actor-oriented architecture for
collaboration allows people and institutions
with shared values and a common purpose to
self-organize for projects that address problems
of mutual interest. ImproveCareNow depends on
the voluntary participation of care centers, clini-
cians, and patients. The network facilitates col-
laboration by providing resources, such as web-
based collaboration spaces, project manage-
ment, learning activities, and communication
resources that make it easier for geographically
disparate people to work together.
The relentless focus on improving rates of clin-
ical remission for children with inflammatory
bowel disease is the common purpose that gal-
vanizes the ImproveCareNow community. The
consistent message that “you can make a differ-
ence”is communicated through transparent
sharing of outcomes data, best practices, and
personal narratives. Sharing occurs during
monthly teleconference and semiannual, in-
person learning sessions and via ImproveCare
Now’s newsletter, blog, and social media plat-
forms. This is coupled with a credo of “steal
shamelessly and share seamlessly”to spread
good ideas to all care centers in the network.
These messages motivate contributions.
In ImproveCareNow, patients have collaborat-
ed with scientists to develop chronic care inno-
vations. Among their ideas were better use of
new technology that enables self-monitoring of
symptoms and use of treatments to improve
shared decision making between patients and
doctors. More recently, ten-year-old children
have worked with their parents to create instruc-
tional videos illustrating how to insert feeding
tubes through their own noses and into their
stomachs for nutritional therapy. Parents and
clinicians teach (and co-teach) modules at learn-
ing sessions that address how to make changes in
care delivery, how to form a parent mentoring
group, how to incorporate parents into care cen-
ters’quality improvement teams, and how to
develop an elevator pitch for ImproveCareNow.
Parents have organized to raise money and
produce materials for families with newly diag-
nosed children. They lead online discussion fo-
rums, community events such as inflammatory
bowel disease education days, and design mobile
tracking tools to supplement existing clinical
data with patient-reported data to understand
the effectiveness of nonpharmaceutical inter-
ventions such as probiotics.
To facilitate collaboration among patients and
clinicians, ImproveCareNow runs monthly team
Exhibit 1
Percentage Of Pediatric Inflammatory Bowel Disease Patients In Remission, 2007–14
SOURCE Data are from the ImproveCareNow pediatric inflammatory bowel disease registry for 2007–14. NOTES Each blue dot rep-
resents the percentage of patients in remission among care centers with more than 75 percent of their patients enrolled in Improve
CareNow in a given month. The figure shows the upper and lower confidence limits (dashed red lines in red) and the mean (green solid
lines).
July 2014 33:7 Health Affairs 1173
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

calls, semiannual learning sessions attended by
teams from each participating center, online
communities for parents (www.smartpatients
.com/ibd), a digital bulletin board for sharing
ideas and tools (www.improvecarenowexchange
.org), and a database that clinicians can query.
Digital Architecture
A second characteristic of a learning health sys-
tem that ImproveCareNow has implemented is
facilitated data entry from EHRs. A learning
health system relies on data from EHRs and pa-
tient registries to foster collaborative improve-
ment, research, data sharing, and innovation.11–13
This “data in once/used many times”mantra has
been the vision of leading thinkers in informat-
ics.14 Clinicians entered data into the EHR via
structured templates and received monthly re-
ports on the quality of their data. Best practices
for achieving the highest-quality data are shared
during learning sessions.
Although learning health system thought lead-
ers have advocated for a distributed data network
in which source data remain with data owners
until the data are needed for a specific pur-
pose,15,16 ImproveCareNow has pursued a cen-
tralized approach. This is because many care
centers do not have the informatics resources
to support such a distributed model or the local
capability to undertake near real-time reporting
to support care management and improvement
activities. A distributed model makes sense if
data are combined across health care organiza-
tions within the network for one research study
at a time. The distributed model becomes more
cumbersome, however, when demands for data
require near-real-time access, as is the case for
ImproveCareNow.
Quality Improvement
To make optimal use of its data, ImproveCare
Now has developed software applications that
enhance chronic care management. A variety
of reports that contrast a given health center
to its peers are provided monthly. Care centers
also have the ability to access the database and
generate reports more frequently. Daily reports
are made available for each patient, and these are
used to improve care at the individual level with
decision support and previsit planning. For ex-
ample, the patient reports include recommenda-
tions for appropriate dosing of medications and
recommended laboratory evaluations before a
visit with a patient.
A learning health system must also reduce the
interval between the discovery of new knowledge
and its impact on patients. ImproveCareNow
adapts standardized processes and tools for
chronic illness care, such as reviewing the entire
population of patients each month to identify if
there are patients who missed needed care,
learning from variations in performance, and
sharing knowledge about how to implement
changes to help care centers rapidly integrate
new information into patient care.17 More reli-
able previsit planning and regular population
review make it easier for physicians to adjust
treatments to individual patient needs. The big
increases in remission rates observed in the
ImproveCareNow patient population (Exhibit 1)
were associated with adoption of standardized
and reliable care delivery processes such as pre-
visit planning and population management.
Rapid Research
The data collected for ImproveCareNow have
been used for chronic care improvement since
2007. It was unclear whether these data could
also be used to rapidly generate new knowledge
that would be generalizable to all children with
inflammatory bowel disease. Thus, Improve
CareNow recently tested the feasibility and valid-
ity of using its registry data for comparative ef-
fectiveness research. The study contrasted the
effects of anti-tumor necrosis factor α(anti-
TNFα) therapy versus conventional care for
moderate-to-severe pediatric Crohn’s disease pa-
tients. This topic was of high priority for clini-
cians because the cost of anti-TNFαis in the
range of tens of thousands of dollars per year,
and the long-term direct and indirect costs are
substantial.18 Administration of anti-TNFαhas
been associated with serious infections, hepatic
T-cell lymphomas, systemic lupus, and blood
disorders.19 Nonetheless, in studies done on
adults, anti-TNFαhad been shown to be highly
The relentless focus
on improving rates of
clinical remission for
children with
inflammatory bowel
disease galvanizes the
ImproveCareNow
community.
Building Rapid-Learning Systems
1174 Health Affairs July 2014 33:7
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

effective for reducing or eliminating the symp-
toms of Crohn’s disease.20,21 These same compar-
ative studies had not been done in children be-
cause of practical (time and cost) and ethical
(withholding an effective treatment) challenges.
The availability of the ImproveCareNow data-
base, with over 4,000 Crohn’s disease patients
at the end of 2012 helped overcome these ob-
stacles.
Using the ImproveCareNow data, analyses
were done over the course of a few months to
estimate the treatment effect for anti-TNFα(in-
tervention group) compared with usual care
(control group).22 The results were remarkably
consistent with those of studies that evaluated
treatment efficacy among children receiving an-
ti-TNFαbut lacked a control group, and compar-
ative controlled clinical trials done among
adults.21,23 They expand the evidence base by pro-
viding new information on the comparative ef-
fectiveness of anti-TNFαfor children managed in
routine pediatric gastroenterology settings. The
study demonstrated that prospectively collected
data from ImproveCareNow could be used to
rapidly answer important clinical questions that
cannot be addressed with controlled trials be-
cause of practical or ethical challenges. More-
over, the study’s methodology offers advantages
relative to conventional clinical trials in terms of
time, cost, recruitment, and the capacity to forgo
the use of placebo.
From Prototype To National
Pediatric Learning Health System
ImproveCareNow has been a remarkable proto-
type for learning. It has shown the way forward
in the domains of technology, governance, im-
plementation science, comparative effectiveness
research, and community engagement. To
achieve our vision of a national pediatric learn-
ing health system, however, we recognized the
need to scale up the ImproveCareNow prototype
to large pediatric health care organizations, oth-
er disease-specific communities, and national
data partners to create a network-based platform
that could support quality improvement and re-
search across all pediatric specialties, diseases,
and regions.
This vision is becoming a reality. Recently,
with funding from the Patient-Centered Out-
comes Research Institute, we established PEDS-
net. The purpose of PEDSnet is to create a
community of patients, families, clinicians, sci-
entists, and health care system leaders who work
together in a distributed learning health system
that is dedicated to discovering and implement-
ing new ways of providing the best care and en-
suring the best outcomes most efficiently. The
governance, regulatory, informatics, social, and
scientific infrastructure that PEDSnet is develop-
ing will enable research on acute, behavioral,
surgical, and chronic medical conditions in pe-
diatrics.
PEDSnet is a network currently composed of
eight of the nation’s largest pediatric academic
health centers: Children’s Hospital of Philadel-
phia, Cincinnati Children’s Hospital Medical
Center, Children’s Hospital Colorado, Nemours
Children’s Health System, Nationwide Chil-
dren’s Hospital, St. Louis Children’s Hospital,
Seattle Children’s Hospital, and Boston Chil-
dren’s Hospital. Each year these institutions care
for over two million children.24
In addition to developing data, regulatory, sci-
entific, and governance infrastructures across
children’s hospitals, we are explicitly linking
PEDSnet to three disease-specific networks—
ImproveCareNow (pediatric inflammatory bow-
el disease), the National Pediatric Cardiology
Quality Improvement Collaborative (complex
congenital heart disease), and a newly formed
Healthy Weight network (childhood obesity).
PEDSnet And Pediatric Big Data
To create a pediatric big-data resource that is
comprehensive in scope, PEDSnet has partnered
with two national data partners, ExpressScripts
and IMS Health. Over the next two years PEDS-
net will link administrative data from these data
partners to the clinical data from pediatric aca-
demic medical centers to provide retail pharma-
cy (dispensed medications) and health insur-
ance claims (health care use and costs) for
patients.
Once these linkages are complete, PEDSnet
will be the most comprehensive pediatric big-
data project in the United States and will support
the conduct of efficient clinical trials and large-
scale observational research. PEDSnet is part
of the larger National Patient-Centered Clinical
Research Network, or PCORnet (www.pcornet
.org), which includes ten other institutional net-
There exists a critical
need to develop a
national strategy for
rapidly improving
children’s health care.
July 2014 33:7 Health Affairs 1175
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

works like PEDSNet and a total of eighteen pa-
tient-powered, disease-specific networks includ-
ing ImproveCareNow. When fully functional,
PCORnet will include tens of millions of Amer-
icans, improving the capacity to rapidly learn
what works for which patients.
As a demonstration project on the validity of
using pediatric big data derived from EHRs,
PEDSnet combined data from six of the eight
institutions to accrue a data set containing infor-
mation on 1.4 million children ages 2–17.25 The
study demonstrated the feasibility of sharing
EHR-derived data for assessing obesity in large
populations of children. The time and effort re-
quired to retrieve the data were nominal, yet the
scale of the EHR-derived data was significant:
The sample from six pediatric institutions pro-
duced 6,000 body mass index assessments per
month of age for most of childhood. Not only
were these results consistent with national esti-
mates obtained by the Centers for Disease Con-
trol and Prevention, the study also demonstrated
associations between obesity and comorbidities
such as diabetes, hypertension, dyslipidemia,
liver disease, and sleep apnea, and rare diseases
such as leukemia.
Another key technology barrier to forming pe-
diatric big data is the lack of standardized defi-
nitions and descriptions of clinical observations
for pediatric care and child health.26 Without a
common terminology, institutions may define
the same clinical concept differently in EHRs.
This makes combining data across research stud-
ies challenging because different definitions are
used for the same underlying concepts. To ad-
dress this need in pediatrics, PEDSnet and the
National Institute of Child Health and Human
Development have launched a pediatric research
terminology initiative that is linking pediatric
terms to existing standard terminologies.26
Conclusions
Creating big data in the absence of purposefully
designed systems that can produce new knowl-
edge (via research) and apply that knowledge at
the point of care (via quality improvement) is
unlikely to substantively improve the health
and lives of patients. There exists a critical need
to develop a national strategy for rapidly improv-
ing children’s health care. Such a strategy should
weave quality improvement and research togeth-
er into the fabric of the health system. Institu-
tions must learn how to trust one another as they
share data, patients, and the burden of the re-
search regulatory infrastructure. Designing and
developing such an infrastructure will also re-
quire forward thinking regarding its sustainabil-
ity, so that it becomes a resource not only today
but also for future generations.
Knowledge production followed by passive dif-
fusion is the status quo and is not serving anyone
well. Learning health systems are needed to build
communities of patients, clinicians, researchers,
and health system leaders dedicated to the com-
mon purpose of improving the health and lives of
children. These new systems of carewill generate
big-data resources and enable novel types of re-
search. However, by engaging all stakeholders in
the knowledge production process, we increase
the likelihood that the most important research
questions (such as those that can have substan-
tive impacts on the health and lives of patients)
are asked and answered. Lastly, the learning
health system is continuously improving clinical
operations and driving new knowledge into the
point of care when and where it is needed. The
combination of research and quality improve-
ment can greatly shorten the time from knowl-
edge acquisition to patient impact.
The success of the ImproveCareNow learning
health system for pediatric inflammatory bowel
disease has paved the way for PEDSnet to spread
the learning health system to other diseases,
specialties, and health care organizations. If suc-
cessful, PEDSnet will become a national network
of hospitals, clinics, care centers, patient com-
munities, and other data partners that collabo-
rate to create a resource that can help us reach
our aspiration of providing care for every child in
the nation within the context of a learning health
system. ▪
Knowledge production
followed by passive
diffusion is the status
quo and is not serving
anyone well.
Building Rapid-Learning Systems
1176 Health Affairs July 2014 33:7
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from

The project was supported by Agency
for Healthcare Research and Quality
Grant No. R01 HS020024, National
Institutes of Health Transformative
Research Grant No. R01 DK085719,
ImproveCareNow Care Centers, and
Patient-Centered Outcomes Research
Institute Grant No. CDRN-1306-01556.
The authors are grateful to the
clinicians and families of participating
ImproveCareNow centers who have
contributed financial and staff resources
and their time, implemented innovations,
and redesigned care delivery systems to
improve care and outcomes. The authors
also acknowledge the many families,
clinicians, and researchers working with
the C3N Project to help
ImproveCareNow transform itself into a
network-based learning health system.
More information about the C3N Project
can be found at http://c3nproject.org.
NOTES
1National Heart, Lung, and Blood
Institute. Children and clinical
studies [Internet]. Bethesda (MD):
NHLBI; [cited 2014 May 16]. Avail-
able from: http://www.nhlbi.nih
.gov/childrenandclinicalstudies/
index.php
2Olsen LA, Aisner D, McGinnis JM.
The learning healthcare system.
Washington (DC): National Acade-
mies Press; 2007.
3Kwon S, Florence M, Grigas P,
Horton M, Horvath K, Johnson M,
et al. Creating a learning healthcare
system in surgery: Washington
State’s Surgical Care and Outcomes
Assessment Program (SCOAP) at 5
years. Surgery. 2012;151(2):146–52.
4Delaney BC, Peterson KA, Speedie S,
Taweel A, Arvanitis TN, Hobbs FD.
Envisioning a learning health care
system: the electronic primary care
research network, a case study. Ann
Fam Med. 2012;10(1):54–9.
5Greene SM, Reid RJ, Larson EB.
Implementing the learning health
system: from concept to action. Ann
Intern Med. 2012;157(3):207–10.
6Green LW, Ottoson JM, Garcia C,
Hiatt RA. Diffusion theory and
knowledge dissemination, utiliza-
tion, and integration in public
health. Annu Rev Public Health.
2009;30:151–74.
7Slutsky JR. Moving closer to a rapid-
learning health care system. Health
Aff (Millwood). 2007;26(2):w122–4.
DOI: 10.1377/hlthaff.26.2.w122.
8Dougherty D, Conway PH. The “3Ts”
road map to transform US health
care: the “how”of high-quality care.
JAMA. 2008;299(19):2319–21.
9Margolis PA, Peterson LE, Seid M.
Collaborative Chronic Care Net-
works (C3Ns) to transform chronic
illness care. Pediatrics. 2013;131
Suppl 4:S219–23.
10 Fjeldstad ØD, Snow CC, Miles RE,
Lettl C. The architecture of collabo-
ration. Strateg Manage J. 2012;
33(6):734–50.
11 Field D, Sansone S-A, Collis A, Booth
T, Dukes P, Gregurick SK, et al.
’Omics data sharing. Science.
2009;326(5950):234–6.
12 Kaye J, Heeney C, Hawkins N, de
Wries J, Boddington P. Data sharing
in genomics—re-shaping scientific
practice. Nat Rev Genet. 2009;
10(5):331–5.
13 Margolis P, Halfon N. Innovation
networks: a strategy to transform
primary health care. JAMA. 2009;
302(13):1461–2.
14 Cimino JJ. Collect once, use many.
Enabling the reuse of clinical data
through controlled terminologies. J
AHIMA. 2007;78(2):24–9.
15 Friedman CP, Wong AK, Blumenthal
D. Achieving a nationwide learning
health system. Sci Transl Med.
2010;2(57):57cm29.
16 Maro JC, Platt R, Holmes JH, Strom
BL, Hennessy S, Lazarus R, et al.
Design of a national distributed
health data network. Ann Intern
Med. 2009;151(5):341–4.
17 Kilo CM. A framework for collabo-
rative improvement: lessons from
the Institute for Healthcare Im-
provement’s Breakthrough Series.
Qual Manag Health Care. 1998;
6(4):1–13.
18 Odes S. How expensive is inflam-
matory bowel disease? A critical
analysis. World J Gastroenterol.
2008;14(43):6641–7.
19 Parashette KR, Makam RC, Cuffari C.
Infliximab therapy in pediatric
Crohn’s disease: a review. Clin Exp
Gastroenterol. 2010;3:57–63.
20 Peyrin-Biroulet L. Anti-TNF therapy
in inflammatory bowel diseases: a
huge review. Minerva Gastroenterol
Dietol. 2010;56(2):233–43.
21 Colombel JF, Sandborn WJ, Reinisch
W, Mantzaris GJ, Kornbluth A,
Rachmilewitz D, et al. Infliximab,
azathioprine, or combination thera-
py for Crohn’s disease. N Engl J Med.
2010;362(15):1383–95.
22 Forrest CB, Crandall WV, Bailey LC,
Zhang P, Joffe MM, Colletti RB, et al.
Effectiveness of Anti-TNFαfor
Crohn’s disease: research in a pedi-
atric learning health system. Pedi-
atrics. Forthcoming 2014.
23 Hyams J, Crandall W, Kugathasan S,
Griffiths A, Olson A, Johanns J, et al.
Induction and maintenance inflixi-
mab therapy for the treatment of
moderate-to-severe Crohn’s disease
in children. Gastroenterology.
2007;132(3):863–73.
24 Forrest CB, Margolis PA, Bailey LC,
Marsolo K, Del Beccaro MA,
Finkelstein JA, et al. PEDSnet: a
National Pediatric Learning Health
System. J Am Med Inform Assoc.
2014;21:602–6.
25 Bailey LC, Milov DE, Kelleher K,
Kahn MG, Del Beccaro M,Yu F, et al.
Multi-institutional sharing of elec-
tronic health record data to assess
childhood obesity. PLoS One. 2013;
8(6):e66192.
26 Kahn MG, Bailey LC, Forrest CB,
Padula MA, Hirschfeld S. Building a
common pediatric research termi-
nology for accelerating child health
research. Pediatrics. 2014;133(3):
516–25.
July 2014 33:7 Health Affairs 1177
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from