ArticlePDF Available

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



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.
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:
available at:
The online version of this article, along with updated information and services, is
For Reprints, Links & Permissions: Alerts : 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
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-downre-
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;25and 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
NO. 7 (2014): 11711177
©2014 Project HOPE
The People-to-People Health
Foundation, Inc.
Christopher B. Forrest
( is a
professor of pediatrics at the
Philadelphia and the
University of Pennsylvania as
well as principal investigator
for the PEDSnet learning
health system, all in
Peter Margolis is a professor
of pediatrics and director of
research at the James M.
Anderson Center for Health
Systems Excellence at the
Cincinnati ChildrensHospital
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
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 todays 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 doing7is 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-
nentsengaged communities, big data, quality
improvement, and researchshould 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
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 networkthe
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 Crohns 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 ( 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 Qualitys 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-
enceis 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
Nows newsletter, blog, and social media plat-
forms. This is coupled with a credo of steal
shamelessly and share seamlesslyto 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-
tersquality 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, 200714
SOURCE Data are from the ImproveCareNow pediatric inflammatory bowel disease registry for 200714. 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
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.1113
This data in once/used many timesmantra 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
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 Crohns 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
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 Crohns 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 Crohns disease patients
at the end of 2012 helped overcome these ob-
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 studys 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-
PEDSnet is a network currently composed of
eight of the nations largest pediatric academic
health centers: Childrens Hospital of Philadel-
phia, Cincinnati Childrens Hospital Medical
Center, Childrens Hospital Colorado, Nemours
Childrens Health System, Nationwide Chil-
drens Hospital, St. Louis Childrens Hospital,
Seattle Childrens Hospital, and Boston Chil-
drens Hospital. Each year these institutions care
for over two million children.24
In addition to developing data, regulatory, sci-
entific, and governance infrastructures across
childrens 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
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
childrens 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 217.25 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
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 childrens 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
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
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
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
States Surgical Care and Outcomes
Assessment Program (SCOAP) at 5
years. Surgery. 2012;151(2):14652.
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):549.
5Greene SM, Reid RJ, Larson EB.
Implementing the learning health
system: from concept to action. Ann
Intern Med. 2012;157(3):20710.
6Green LW, Ottoson JM, Garcia C,
Hiatt RA. Diffusion theory and
knowledge dissemination, utiliza-
tion, and integration in public
health. Annu Rev Public Health.
7Slutsky JR. Moving closer to a rapid-
learning health care system. Health
Aff (Millwood). 2007;26(2):w1224.
DOI: 10.1377/hlthaff.26.2.w122.
8Dougherty D, Conway PH. The 3Ts
road map to transform US health
care: the howof high-quality care.
JAMA. 2008;299(19):231921.
9Margolis PA, Peterson LE, Seid M.
Collaborative Chronic Care Net-
works (C3Ns) to transform chronic
illness care. Pediatrics. 2013;131
Suppl 4:S21923.
10 Fjeldstad ØD, Snow CC, Miles RE,
Lettl C. The architecture of collabo-
ration. Strateg Manage J. 2012;
11 Field D, Sansone S-A, Collis A, Booth
T, Dukes P, Gregurick SK, et al.
Omics data sharing. Science.
12 Kaye J, Heeney C, Hawkins N, de
Wries J, Boddington P. Data sharing
in genomicsre-shaping scientific
practice. Nat Rev Genet. 2009;
13 Margolis P, Halfon N. Innovation
networks: a strategy to transform
primary health care. JAMA. 2009;
14 Cimino JJ. Collect once, use many.
Enabling the reuse of clinical data
through controlled terminologies. J
AHIMA. 2007;78(2):249.
15 Friedman CP, Wong AK, Blumenthal
D. Achieving a nationwide learning
health system. Sci Transl Med.
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):3414.
17 Kilo CM. A framework for collabo-
rative improvement: lessons from
the Institute for Healthcare Im-
provements Breakthrough Series.
Qual Manag Health Care. 1998;
18 Odes S. How expensive is inflam-
matory bowel disease? A critical
analysis. World J Gastroenterol.
19 Parashette KR, Makam RC, Cuffari C.
Infliximab therapy in pediatric
Crohns disease: a review. Clin Exp
Gastroenterol. 2010;3:5763.
20 Peyrin-Biroulet L. Anti-TNF therapy
in inflammatory bowel diseases: a
huge review. Minerva Gastroenterol
Dietol. 2010;56(2):23343.
21 Colombel JF, Sandborn WJ, Reinisch
W, Mantzaris GJ, Kornbluth A,
Rachmilewitz D, et al. Infliximab,
azathioprine, or combination thera-
py for Crohns disease. N Engl J Med.
22 Forrest CB, Crandall WV, Bailey LC,
Zhang P, Joffe MM, Colletti RB, et al.
Effectiveness of Anti-TNFαfor
Crohns 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 Crohns disease
in children. Gastroenterology.
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.
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;
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):
July 2014 33:7 Health Affairs 1177
at University of Pennsylvania Library on July 10, 2014Health Affairs by content.healthaffairs.orgDownloaded from
... [8][9][10][11] Many healthcare systems participate in learning networks as a part of an LHS or the infrastructure for an LHS. [12][13][14][15][16][17][18][19][20][21][22][23][24] Although real-world examples of mature pediatric LHSs are sparse in the literature, the efforts by publicly and privately funded learning networks continue to support an extensive number of outcomes and comparative effectiveness research studies that improve the literature base for building LHS capacity. 9,23,25,26 Along with PEDSnet, other examples of pediatric-specific research networks include ImproveCareNow and Learn From Every Patient. ...
... 9,23,25,26 Along with PEDSnet, other examples of pediatric-specific research networks include ImproveCareNow and Learn From Every Patient. 9,19,23,27 However, the extent of LHS infrastructure, initiatives, and guidance specific to rehabilitation research is limited. Therefore, in this paper, we aim to advance the science of LHS for rehabilitation by describing the process of developing the technical infrastructure of an LHS for a pediatric specialty care rehabilitation network, the Shriners Hospitals for Children (SHC) Health Outcomes Network (SHOnet). ...
... PEDSnet, which consists of eight large pediatric healthcare systems in the United States, modified the Observational Medical Outcomes Partnership (OMOP) CDM V5 19,20,36,37 with a defined set of pediatric common data elements. OMOP is designed to enable research on associations between healthcare interventions and outcomes for specific patient cohorts (eg, patients with a certain condition or those who have undergone a specific procedure). ...
To describe the development and implementation of learning health system (LHS) infrastructure for a pediatric specialty care health system to support LHS research in pediatric rehabilitation settings. An existing pediatric common data model (eg, PEDSnet) of standardized medical terminologies for research was expanded and leveraged for this stud, and applied to SHOnet, a clinical research data resource consisting of deidentified data extracted from the electronic health record (EHR) from the Shriners Hospitals for Children speacialty pediatric health care system. We mapped EHR data for laboratory, procedures, drugs, and conditions to standardized vocabularies including ICD‐10, CPT, RxNorm, and LOINC to the common data model using an established extraction‐transformation‐loading process. Rigorous quality checks were conducted to ensure a high degree of data conformance, completeness, and plausibility. SHOnet data elements from all sources are de‐identified and the server is managed by the SHC Information Systems Department. SHOnet data are refreshed monthly and data elements are continually expanded based on new research endeavors. Not applicable. The Shriners Health Outcomes Network (SHOnet) includes data for over 10 000 distinct observational data elements based on over two million patient encounters between 2011 and present. The systematic process to develop SHOnet is replicable and flexible for other pediatric rehabilitation research settings interested in building out their LHS capabilities. Challenges and facilitators may arise for building such LHS infrastructure for rehabilitation in areas of (a) data capture, curation, query, and governance, (b) generating knowledge from data, and (c) dissemination and implementation of new institutional knowledge. Further research studies are needed to evaluate these data resources for scalable system‐learning endeavors. SHOnet is an exemplar of an LHS for rehabilitation and specialty care settings. The success of an LHS is dependent on engagement of multiple stakeholders, shared governance, effective knowledge translation, and deep commitment to long‐term strategies for engaging clinicians, administration, and families in leveraging knowledge to improve clinical outcomes.
... One of the challenges faced by pediatric nephrologists is that kidney disease is uncommon in children, making it difficult to produce generalizable knowledge through research from [24]. The LHS model can also be applied across multiple care centers in an LHN. ...
Learning health systems (LHS) align science, informatics, incentives, and culture for continuous improvement and innovation. In this organizational system, best practices are seamlessly embedded in the delivery process, and new knowledge is captured as an integral byproduct of the care delivery experience aimed to transform clinical practice and improve patient outcomes. The objective of this review is to describe how building better health systems that integrate clinical care, improvement, and research as part of an LHS can improve care within pediatric nephrology. This review will provide real-world examples of how this system can be established in a single center and across multiple centers as learning health networks.
... We leveraged PEDSnet (, which was established to produce evidence by sharing data, expertise, and other resources across several of US children's hospitals 10,11 ) to produce what is to our knowledge the largest and most accurate epidemiological description of AA in children to date. This information can potentially assist clinicians in advocating for resources for children with AA and payers and health systems in planning for AA therapeutics. ...
Importance: Pediatric alopecia areata (AA) prevalence and incidence data are key to understanding the natural history of this medical disease. Objective: To determine the prevalence and incidence of AA in a pediatric population across time, age, sex, race and ethnicity, and geographic areas within the US. Design, setting, and participants: In this multicenter cohort study conducted among 5 children's hospitals, data (January 2009 to November 2020) were collected from a standardized electronic health record (PEDSnet database, version 4.0) to evaluate the incidence and prevalence of pediatric AA. The study cohort included patients younger than 18 years with at least 2 physician visits during which a diagnosis code for AA was recorded, or 1 dermatologist specialty visit for which AA was recorded. Main outcomes and measures: The prevalence denominator population comprised 5 409 919 patients. The incidence denominator population was 2 896 241. We identified 5801 children for inclusion in the AA cohort, and 2398 (41.3%) had 12 months or more of follow-up and were included in the incidence analysis. Results: Of 5801 patients in the AA cohort, the mean (SD) age was 9.0 (4.5) years, 3259 (56.2%) were female, 359 (6.2) were Asian, 1094 (18.9%) were Black, 1348 (23.2%) were Hispanic, and 2362 (40.7%) were White. The overall prevalence of pediatric AA was 0.11%, and the participants in the AA cohort were more often older, female, and members of a racial and ethnic minority group than the full PEDSnet population. The 11-year overall incidence rate of pediatric AA between 2009 and 2020 was 13.6 cases per 100 000 person-years (95% CI, 13.1-14.2). The incidence rate by age was normally distributed and peaked at age 6 years. Rates were 22.8% higher in female patients than male patients (15.1 cases per 100 000 person-years for females vs 12.3 cases per 100 000 person-years for males). Additionally, incidence rates were highest among Hispanic children (31.5 cases per 100 000 person-years). Conclusions and relevance: This cohort study examined the prevalence and incidence rates of pediatric AA in the US across time, age, sex, race and ethnicity, and region from 2009 to 2020, finding a prevalence of 0.11% (doubling during the last decade) and incidence rate of 13.6 cases per 100 000 person-years. Additionally, the results identified Asian and Hispanic children as high-risk demographic subgroups who were shown to be 2 and 3 times more likely, respectively, to receive a diagnosis of AA.
... a national network of pediatric health systems that share clinical data and conduct observational studies, clinical trials, and population surveillance. 14 ...
Background and objectives: High transmissibility of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Delta variant led to increased coronavirus disease 2019 (COVID-19) cases, but its effect on severe illness among children is less clear. This study evaluated changes in COVID-19 severity from March 1, 2020, to December 30, 2021. Methods: We examined electronic health record data from encounters that occurred in outpatient and inpatient settings in 9 health systems participating in PEDSnet. The study sample included children aged <18 years with a positive viral test for SARS-CoV-2. Severity was categorized as asymptomatic, mild (symptoms), moderate (moderately severe COVID-19-related conditions such as gastroenteritis, dehydration, and pneumonia), or severe (unstable COVID-19-related conditions, ICU admission, or mechanical ventilation). Results: The number of patients classified as asymptomatic was 54 948 (66.4%), with 22 303 (26.9%) being mild, 3781 (4.6%) being moderate, and 1766 (2.1%) being severe. In 2021, patients with moderate to severe illness peaked in June (13.5%), declining to December (8.1%). Compared with July 2020 to February 2021, the adjusted odds ratio for moderate to severe illness was highest in June 2021 (adjusted odds ratio, 2.8; 95% confidence interval, 2.2-3.6) and lower in July to December 2021, when the Delta variant predominated. The adjusted odds ratio for moderate to severe illness among children with complex chronic conditions was 4.2 (95% confidence interval, 3.9-4.5). Conclusions: Although 1 in 16 children infected with the SARS-CoV-2 virus experienced moderate or severe illness, the risk of severe disease did not change with the emergence of the Delta variant, despite its high transmissibility.
... ICN participants share goals, standards and resources and their continuous use of measurements demonstrate success, including continually improved clinical outcomes. The ICN has served as a prototype for a national paediatric LHS, the PED-Snet [18,19]. No images of the Cincinnati Collaborative LHS Model are presented in the papers we reviewed. ...
Full-text available
Background Co-production of health is defined as ‘the interdependent work of users and professionals who are creating, designing, producing, delivering, assessing, and evaluating the relationships and actions that contribute to the health of individuals and populations’. It can assume many forms and include multiple stakeholders in pursuit of continuous improvement, as in Learning Health Systems (LHSs). There is increasing interest in how the LHS concept allows integration of different knowledge domains to support and achieve better health. Even if definitions of LHSs include engaging users and their family as active participants in aspects of enabling better health for individuals and populations, LHS descriptions emphasize technological solutions, such as the use of information systems. Fewer LHS texts address how interpersonal interactions contribute to the design and improvement of healthcare services. Objective We examined the literature on LHS to clarify the role and contributions of co-production in LHS conceptualizations and applications. Method First, we undertook a scoping review of LHS conceptualizations. Second, we compared those conceptualizations to the characteristics of LHSs first described by the US Institute of Medicine. Third, we examined the LHS conceptualizations to assess how they bring four types of value co-creation in public services into play: co-production, co-design, co-construction and co-innovation. These were used to describe core ideas, as principles, to guide development. Result Among 17 identified LHS conceptualizations, 3 qualified as most comprehensive regarding fidelity to LHS characteristics and their use in multiple settings: (i) the Cincinnati Collaborative LHS Model, (ii) the Dartmouth Coproduction LHS Model and (iii) the Michigan Learning Cycle Model. These conceptualizations exhibit all four types of value co-creation, provide examples of how LHSs can harness co-production and are used to identify principles that can enhance value co-creation: (i) use a shared aim, (ii) navigate towards improved outcomes, (iii) tailor feedback with and for users, (iv) distribute leadership, (v) facilitate interactions, (vi) co-design services and (vii) support self-organization. Conclusions The LHS conceptualizations have common features and harness co-production to generate value for individual patients as well as for health systems. They facilitate learning and improvement by integrating supportive technologies into the sociotechnical systems that make up healthcare. Further research on LHS applications in real-world complex settings is needed to unpack how LHSs are grown through coproduction and other types of value co-creation.
... 7,8 EHR distributed data networks (DDNs) can leverage federal investments for research, quality improvement and public health, valued domains for any LHS. Federated data sharing, recognized by funding agencies, 9 and adopted by clinical data research networks, 10,11 can preserve privacy and security as data remain behind firewalls of DDN-participating healthcare organizations, until queried for specific approved uses. The importance of IM in a DDN is likely influenced by the specific use case and geographic proximity of participating organizations. ...
Full-text available
Introduction Learning health systems can help estimate chronic disease prevalence through distributed data networks (DDNs). Concerns remain about bias introduced to DDN prevalence estimates when individuals seeking care across systems are counted multiple times. This paper describes a process to deduplicate individuals for DDN prevalence estimates. Methods We operationalized a two-step deduplication process, leveraging health information exchange (HIE)-assigned network identifiers, within the Colorado Health Observation Regional Data Service (CHORDS) DDN. We generated prevalence estimates for type 1 and type 2 diabetes among pediatric patients (0-17 years) with at least one 2017 encounter in one of two geographically-proximate DDN partners. We assessed the extent of cross-system duplication and its effect on prevalence estimates. Results We identified 218 437 unique pediatric patients seen across systems during 2017, including 7628 (3.5%) seen in both. We found no measurable difference in prevalence after deduplication. The number of cases we identified differed slightly by data reconciliation strategy. Concordance of linked patients' demographic attributes varied by attribute. Conclusions We implemented an HIE-dependent, extensible process that deduplicates individuals for less biased prevalence estimates in a DDN. Our null pilot findings have limited generalizability. Overlap was small and likely insufficient to influence prevalence estimates. Other factors, including the number and size of partners, the matching algorithm, and the electronic phenotype may influence the degree of deduplication bias. Additional use cases may help improve understanding of duplication bias and reveal other principles and insights. This study informed how DDNs could support learning health systems' response to public health challenges and improve regional health.
... a member of The National Patient-Centered Clinical Research Network (PCORnet,, is a multi-center network of pediatric healthcare 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 systems containing data on > 6.7 million children. PEDSnet is a national-scale learning health system that enables identification and study of patient cohorts using case defining algorithms (13)(14)(15). It has the potential to be a useful resource for rare diseases that are challenging to study outside of specific patient registries. ...
Background: Performing adequately powered clinical trials in pediatric diseases such as systemic lupus erythematosus (SLE) is challenging. Improved recruitment strategies are needed for identifying patients. Design, Setting, Participants, and Measurements: Electronic health record (EHR) algorithms were developed and tested to identify children with SLE both with and without lupus nephritis. We used single center EHR data to develop computable phenotypes comprised of diagnosis, medication, procedure, and utilization codes. These were evaluated iteratively against a manually assembled SLE patient database. The highest performing phenotypes were then evaluated across institutions in PEDSnet, a national healthcare systems network of >6.7 million children. Reviewers blinded to case status used standardized forms to review random samples of cases (n=350) and non-cases (n=350). Results: Final algorithms consisted of both utilization and diagnostic criteria. For both, utilization criteria included ≥2 in-person visits with nephrology or rheumatology and ≥60 days follow-up. SLE diagnostic criteria included absence of neonatal lupus, ≥1 hydroxychloroquine exposure, and either ≥3 qualifying diagnosis codes separated by ≥30 days, or ≥1 diagnosis code and ≥1 kidney biopsy procedure code. Sensitivity was 100% (95%CI, 99-100); specificity, 92% (95% CI, 88-94); positive predictive value, 91% (95%CI, 87-94); negative predictive value, 100% (95%CI, 99-100). Lupus nephritis diagnostic criteria included either ≥3 qualifying lupus nephritis diagnosis codes (or SLE codes on same day as glomerular/kidney codes) separated by ≥30 days, or ≥1 SLE diagnosis code and ≥1 kidney biopsy procedure code. Sensitivity was 90% (95%CI, 85-94); specificity, 93% (95% CI, 89-97); positive predictive value, 94% (95%CI, 89-97); negative predictive value, 90% (95%CI, 84-94). Algorithms identified 1,508 children with SLE at PEDSnet institutions (537 with LN), 809 of whom were seen in the past 12 months. Conclusions: EHR-based algorithms for SLE and Lupus nephritis demonstrated excellent classification accuracy across PEDSnet institutions.
Introduction: The strength of the evidence base for the comparative effectiveness of three common surgical modalities for paediatric nephrolithiasis (ureteroscopy, shockwave lithotripsy and percutaneous nephrolithotomy) and its relevance to patients and caregivers are insufficient. We describe the methods and rationale for the Pediatric KIDney Stone (PKIDS) Care Improvement Network Trial with the aim to compare effectiveness of surgical modalities in paediatric nephrolithiasis based on stone clearance and lived patient experiences. This protocol serves as a patient-centred alternative to randomised controlled trials for interventions where clinical equipoise is lacking. Methods and analysis: The PKIDS is a collaborative learning organisation composed of 26 hospitals that is conducting a prospective pragmatic clinical trial comparing the effectiveness of ureteroscopy, shockwave lithotripsy and percutaneous nephrolithotomy for youth aged 8-21 years with kidney and/or ureteral stones. Embedded within clinical care, the PKIDS trial will collect granular patient-level, surgeon-level and institution-level data, with a goal enrolment of 1290 participants over a 21-month period. The primary study outcome is stone clearance, defined as absence of a residual calculus of >4 mm on postoperative ultrasound. Secondary outcomes include patient-reported physical, emotional and social health outcomes (primarily using the Patient-Reported Outcome Measurement Information System), analgesic use and healthcare resource use. Timing and content of secondary outcomes assessments were set based on feedback from patient partners. Heterogeneity of treatment effect for stone clearance and patient-reported outcomes by participant and stone characteristics will be assessed. Ethics and dissemination: This study is approved by the central institutional review board with reliance across participating sites. Participating stakeholders will review results and contribute to development dissemination at regional, national and international meetings. Trial registration number: NCT04285658; Pre-results.
Context Diabetes and cardiovascular diseases are common among men with Klinefelter syndrome (KS) and contribute to higher morbidity and mortality. Objective To determine if cardiometabolic-related diagnoses are more prevalent among youth with KS compared to matched controls in a large population-based cohort. Design Secondary data analysis from electronic health records Setting Six pediatric institutions in the United States (PEDSnet) Patients All youth with KS in the database (n=1,080) and 4,497 youth without KS matched for sex, age (mean 13 years at last encounter), year of birth, race, ethnicity, insurance, site, and duration of care (mean 7 years). Main outcome measures Prevalence of five cardiometabolic-related outcomes including overweight/obesity, dyslipidemia, dysglycemia, hypertension, and liver dysfunction Results The odds of overweight/obesity (OR 1.6 (95%CI 1.4-1.8)), dyslipidemia (3.0 (2.2-3.9)), and liver dysfunction (2.0 (1.6-2.5)) were all higher in KS compared to controls. Adjusting for covariates (obesity, testosterone treatment, and antipsychotic use) attenuated the effect of KS on these outcomes, however boys with KS still had 45% greater odds of overweight/obesity (CI 1.2-1.7) and 70% greater odds of liver dysfunction (1.3-2.2) compared to controls, and both dyslipidemia (1.6 (1.1-2.4)) and dysglycemia (1.8 (1.1-3.2)) were higher in KS but of borderline statistical significance when accounting for multiple comparisons. The odds of hypertension were not different between groups. Conclusions This large, population-based cohort of youth with KS had a higher odds of most cardiometabolic-related diagnoses compared to matched controls.
Full-text available
Despite significant gains by pediatric collaborative improvement networks, the overall US system of chronic illness care does not work well. A new paradigm is needed: a Collaborative Chronic Care Network (C3N). A C3N is a network-based production system that harnesses the collective intelligence of patients, clinicians, and researchers and distributes the production of knowledge, information, and know-how over large groups of people, dramatically accelerating the discovery process. A C3N is a platform of "operating systems" on which interconnected processes and interventions are designed, tested, and implemented. The social operating system is facilitated by community building, engaging all stakeholders and their expertise, and providing multiple ways to participate. Standard progress measures and a robust information technology infrastructure enable the technical operating system to reduce unwanted variation and adopt advances more rapidly. A structured approach to innovation design provides a scientific operating system or "laboratory" for what works and how to make it work. Data support testing and research on multiple levels: comparative effectiveness research for populations, evaluating care delivery processes at the care center level, and N-of-1 trials and other methods to select the best treatment of individual patient circumstances. Methods to reduce transactional costs to participate include a Federated IRB Model in which centers rely on a protocol approved at 1 central institutional review board and a "commons framework" for organizational copyright and intellectual property concerns. A fully realized C3N represents a discontinuous leap to a self-developing learning health system capable of producing a qualitatively different approach to improving health.
Full-text available
Objectives: ImproveCareNow (ICN) is the largest pediatric learning health system in the nation and started as a quality improvement collaborative. To test the feasibility and validity of using ICN data for clinical research, we evaluated the effectiveness of anti-tumor necrosis factor-α (anti-TNFα) agents in the management of pediatric Crohn disease (CD). Methods: Data were collected in 35 pediatric gastroenterology practices (April 2007 to March 2012) and analyzed as a sequence of nonrandomized trials. Patients who had moderate to severe CD were classified as initiators or non-initiators of anti-TNFα therapy. Among 4130 patients who had pediatric CD, 603 were new users and 1211 were receiving anti-TNFα therapy on entry into ICN. Results: During a 26-week follow-up period, rate ratios obtained from Cox proportional hazards models, adjusting for patient and disease characteristics and concurrent medications, were 1.53 (95% confidence interval [CI], 1.20-1.96) for clinical remission and 1.74 (95% CI, 1.33-2.29) for corticosteroid-free remission. The rate ratio for corticosteroid-free remission was comparable to the estimate produced by the adult SONIC study, which was a randomized controlled trial on the efficacy of anti-TNFα therapy. The number needed to treat was 5.2 (95% CI, 3.4-11.1) for clinical remission and 5.0 (95% CI, 3.4-10.0) for corticosteroid-free remission. Conclusions: In routine pediatric gastroenterology practice settings, anti-TNFα therapy was effective at achieving clinical and corticosteroid-free remission for patients who had Crohn disease. Using data from the ICN learning health system for the purpose of observational research is feasible and produces valuable new knowledge.
Full-text available
A learning health system (LHS) integrates research done in routine care settings, structured data capture during every encounter, and quality improvement processes to rapidly implement advances in new knowledge, all with active and meaningful patient participation. While disease-specific pediatric LHSs have shown tremendous impact on improved clinical outcomes, a national digital architecture to rapidly implement LHSs across multiple pediatric conditions does not exist. PEDSnet is a clinical data research network that provides the infrastructure to support a national pediatric LHS. A consortium consisting of PEDSnet, which includes eight academic medical centers, two existing disease-specific pediatric networks, and two national data partners form the initial partners in the National Pediatric Learning Health System (NPLHS). PEDSnet is implementing a flexible dual data architecture that incorporates two widely used data models and national terminology standards to support multi-institutional data integration, cohort discovery, and advanced analytics that enable rapid learning.
Full-text available
Longitudinal observational clinical data on pediatric patients in electronic format is becoming widely available. A new era of multi-institutional data networks that study pediatric diseases and outcomes across disparate health delivery models and care settings are also enabling an innovative collaborative rapid improvement paradigm called the Learning Health System. However, the potential alignment of routine clinical care, observational clinical research, pragmatic clinical trials, and health systems improvement requires a data infrastructure capable of combining information from systems and workflows that historically have been isolated from each other. Removing barriers to integrating and reusing data collected in different settings will permit new opportunities to develop a more complete picture of a patient's care and to leverage data from related research studies. One key barrier is the lack of a common terminology that provides uniform definitions and descriptions of clinical observations and data. A well-characterized terminology ensures a common meaning and supports data reuse and integration. A common terminology allows studies to build upon previous findings and to reuse data collection tools and data management processes. We present the current state of terminology harmonization and describe a governance structure and mechanism for coordinating the development of a common pediatric research terminology that links to clinical terminologies and can be used to align existing terminologies. By reducing the barriers between clinical care and clinical research, a Learning Health System can leverage and reuse not only its own data resources but also broader extant data resources.
Full-text available
To evaluate the validity of multi-institutional electronic health record (EHR) data sharing for surveillance and study of childhood obesity. We conducted a non-concurrent cohort study of 528,340 children with outpatient visits to six pediatric academic medical centers during 2007-08, with sufficient data in the EHR for body mass index (BMI) assessment. EHR data were compared with data from the 2007-08 National Health and Nutrition Examination Survey (NHANES). Among children 2-17 years, BMI was evaluable for 1,398,655 visits (56%). The EHR dataset contained over 6,000 BMI measurements per month of age up to 16 years, yielding precise estimates of BMI. In the EHR dataset, 18% of children were obese versus 18% in NHANES, while 35% were obese or overweight versus 34% in NHANES. BMI for an individual was highly reliable over time (intraclass correlation coefficient 0.90 for obese children and 0.97 for all children). Only 14% of visits with measured obesity (BMI ≥95%) had a diagnosis of obesity recorded, and only 20% of children with measured obesity had the diagnosis documented during the study period. Obese children had higher primary care (4.8 versus 4.0 visits, p<0.001) and specialty care (3.7 versus 2.7 visits, p<0.001) utilization than non-obese counterparts, and higher prevalence of diverse co-morbidities. The cohort size in the EHR dataset permitted detection of associations with rare diagnoses. Data sharing did not require investment of extensive institutional resources, yet yielded high data quality. Multi-institutional EHR data sharing is a promising, feasible, and valid approach for population health surveillance. It provides a valuable complement to more resource-intensive national surveys, particularly for iterative surveillance and quality improvement. Low rates of obesity diagnosis present a significant obstacle to surveillance and quality improvement for care of children with obesity.
A distributed health data network is a system that allows secure remote analysis of separate data sets, each comprising a different medical organization’s or health plan’s records. Distributed health data networks are currently being planned that could cover millions of people, permitting studies of comparative clinical effectiveness, best practices, diffusion of medical technologies, and quality of care. These networks could also support assessment of medical product safety and other public health needs. Distributed network technologies allow data holders to control all uses of their data, which overcomes many practical obstacles related to confidentiality, regulation, and proprietary interests. Some of the challenges and potential methods of operation of a multipurpose, multi-institutional distributed health data network are described.
The article describes the Breakthrough Series, a collaborative improvement model developed by the Institute for Healthcare Improvement. The model adapts and applies existing knowledge to multiple, similar sites to accomplish common aims. It has been used to address several of the most pressing issues in health care today. The article outlines key elements of the Breakthrough Series to provide a framework for future collaborative improvement efforts. (C)1998Aspen Publishers, Inc.
Firms increasingly face competitive pressures related to rapid and continuous adaptation to a complex, dynamic, and highly interconnected global environment. Pressing challenges include keeping pace with shorter product life cycles, incorporating multiple technologies into the design of new products, cocreating products and services with customers and partners, and leveraging the growth of scientific and technical knowledge in many sectors. In response, we observe experimentation with new organization designs that are fundamentally different from existing forms of organizing. We propose that these new designs are based on an actor-oriented architectural scheme composed of three main elements: (1) actors who have the capabilities and values to s elf-organize; (2) commons where the actors accumulate and share resources; and (3) protocols, processes, and infrastructures that enable multi-actor collaboration. We demonstrate the usefulness of the actor-oriented scheme by applying it to organizations drawn from four different sectors: global professional services, open source software development, computer equipment, and national defense. We discuss the implications of the actor-oriented architectural scheme for future research on organizational forms as well as for managers who are involved in designing organizations.
Clinicians and health systems are facing widespread challenges, including changes in care delivery, escalating health care costs, and the need to keep up with rapid scientific discovery. Reorganizing U.S. health care and changing its practices to render better, more affordable care requires transformation in how health systems generate and apply knowledge. The "rapid-learning health system"-posited as a conceptual strategy to spur such transformation-leverages recent developments in health information technology and a growing health data infrastructure to access and apply evidence in real time, while simultaneously drawing knowledge from real-world care-delivery processes to promote innovation and health system change on the basis of rigorous research. This article describes an evolving learning health system at Group Health Cooperative, the 6 phases characterizing its approach, and examples of organization-wide applications. This practical model promotes bidirectional discovery and an open mind at the system level, resulting in willingness to make changes on the basis of evidence that is both scientifically sound and practice-based. Rapid learning must be valued as a health system property to realize its full potential for knowledge generation and application.
We report a patient with Castleman’s disease arising from the gallbladder neck, which caused difficulty in making the differential diagnosis against gallbladder malignancies. A 50-year-old woman presented to our institution with epigastric pain. An abdominal computed tomography scan (CT) and magnetic resonance cholangiopancreatography (MRCP) study showed a 20-mm tumor located in the gallbladder neck for which malignancy could not be completely ruled out. For the definitive diagnosis and treatment, cholecystectomy was performed. In the operation, the main tumor and resection margins of the cystic duct were submitted for frozen section. The tumor was composed of a proliferation of lymphoid tissue with no signs of dysplasia. The ductal margin was free of tumor. The final histopathological diagnosis was unicentric Castleman’s disease, a hyaline vascular variant that developed in the gallbladder. The patient is currently in good condition without any signs of recurrence 28months after the operation. This is the first detailed report of Castleman’s disease of the gallbladder. Making a correct diagnosis was very difficult before the operation, and only a surgical approach enabled confirmation of the diagnosis for this patient.