IMS Health Real World Evidence Access Point

News, views and insights from leading international experts in RWE and HEOR
IMS HEALTH REAL-WORLD EVIDENCE SOLUTIONS & HEOR
VOLUME 5, ISSUE 9 • NOVEMBER 2014
Putting RWE at
the heart of
decision making
Diabetes special focus
Propelling stakeholder
engagement and collaboration
Optimizing resource allocation
in primary care
Harnessing transformational
methodologies
RWE x 6
1 2
3
6 = $1bn
4
5
Six ways to release
untapped RWE potential

“WeIn hsaIvgeh otbsser?v?e?d?? s?e??v?e?r?al important trends that could shape the
way companies create or use RWE, which will be of importance to
our industry moving forward.”
Headline
Headline
IMS HEALTH REAL-WORLD EVIDENCE SOLUTIONS & HEOR
"RWE is transforming a broad
understanding in diabetes with
real insights into differential
patient cohort responses, based
on powerful clinical and even
genomic data."
Welcome
Welcome to our latest AccessPoint as we continue to explore the
dynamics shaping the HEOR and real-world evidence (RWE) landscape.
In our last edition, we highlighted our evolving understanding of
oncology innovations and outcomes and the role of RWE in these areas.
This time, we expand that lens to another important disease,
diabetes, where stakeholders are seeking much deeper knowledge of
treatment outcomes in patient subgroups. RWE is transforming a
broad understanding with real insights into differential cohort
responses, based on powerful clinical and even genomic data to
evaluate benefits and risks. We also take a broader look at trends in
RWE and spotlight ongoing advances in real-world data (RWD),
methodologies and RWE applications.
We have focused this edition around these three topics
• RWE research reveals new insights into more effective ways of
researching diabetes, assessing outcomes and understanding
the implications for broader care provision. Although the quantity
of diabetes-related patient data is significant, gaps in the
completeness of datasets have impeded researchers. Now, new
mixed methods approaches such as we describe in Germany, and
analytic innovations including the IMS CORE Diabetes Model,
make research for this critical condition easier to conduct with
increased confidence and scientific rigor. A UK analysis of utility
values provides a basis for improving diabetes modeling and a recent
study in Canada shows how RWE analysis can pinpoint the resource
drivers requiring policy and clinical practice changes. This is a hopeful
time in diabetes.
• We have observed several important trends that could shape
the way companies create or use RWE, which will be of
importance to our industry moving forward. New ways of
thinking about RWD strategies are emerging, leading us to
propose a disease-centric framework to help guide those efforts.
We also comment on how involving commercial colleagues in
RWE is driving substantial value for companies that enable this
approach. And we look forward to seeing continued
collaborations with external stakeholders, namely payers, and in
new geographies, specifically Asia Pacific.
• Advancements continue to derive more value from RWE,
including improved data sourcing, methodologies and
stakeholder engagement. Predictive modeling is increasing RWE
accuracy with demonstrated benefits in risk stratification. We are
seeing leaders leverage the richness of Scandinavian data to
enable new disease-level insights. RWE also continues to support
value demonstration, such as showing the impact of adherence on
mortality, readmission risk and costs in ACS. And it is helping
companies move ‘beyond the pill’ by creating even more value
through enabling care management services.
At IMS Health, we are committed to providing insights to help advance
health and improve patient outcomes across all care settings globally.
We hope you find this edition particularly useful in your RWE journey.
AccessPoint is published twice yearly by the
IMS Health Real-World Evidence (RWE) Solutions and
Health Economics & Outcomes Research (HEOR) team.
VOLUME 5, ISSUE 9. PUbLISHEd NOVEMbER 2014.
IMS HEALTH210 Pentonville Road, London N1 9JY, UK
Tel: +44 (0) 20 3075 4800 • www.imshealth.com/rwe
RWEinfo@imshealth.com
©2014 IMS Health Incorporated and its affiliates.
All rights reserved.
Trademarks are registered in the United States
and in various other countries.
Jon Resnick
Vice President and General Manager
Real-World Evidence Solutions, IMS Health
Jresnick@imshealth.com

VOLUME 5, ISSUE 9 • NOVEMBER 2014
RWE driving deeper insights in diabetes
Major validation upholds relevance of IMS CORE diabetes Model 5
diabetes complexities drive resource consumption in Canada 15
Identifying reference utility values for economic models in diabetes 40
A collaborative foundation for new diabetes insights in Germany 45
demonstrating external validity of the IMS CORE diabetes Model 50
Perspectives and trends in RWE
Enabling disease-specific RWE through fit-for-purpose RWd 6
A roadmap for increasing RWE use in payer decisions 10
Finding the true potential of RWE through scientific-commercial collaboration 20
Preparing for RWE in Asia Pacific 36
Advances in RWD, methodology and RWE applications
Improving outcomes through predictive modeling 26
Holistic real-world data brings a new view of patients and diseases 32
Evaluating disease burden, unmet need and QoL in a chronic inflammatory disorder 56
demonstrating the impact of non-adherence to antiplatelet therapy in ACS 60
Modeling disease management above the brand with RWE 63
nEWs
2 PARTNERSHIP ENRICHES SCANDINAVIAN DATASETS
3 RESEARCH INFORMS POLICY PRIORITIES
4 FORUMS ACCELERATE RWE USE
5 IMS CDM CONFIRMS CONTEMPORARY RELEVANCE
PROJECt FOCUs
56 CHRONIC INFLAMMATORY DISORDER
Evaluating patient-reported outcomes
60 ACUTE CORONARY SYNDROME
Demonstrating the impact of non-adherence
63 RWE-BASED DISEASE MANAGEMENT
Informing the value of treatments
IMs RWEs & hEOR OVERVIEW
66 ENABLING YOUR REAL-WORLD SUCCESS
Solutions, locations and expertise
ACCESSPOINT • VOLUME 5 ISSUE 9 PAGE 1

nEWs SCANDINAVIAN RWE COLLABORATION
Partnership linkage of unique, Norwegian biobank data opens up groundbreaking
research potential with global impact
IMS Health/Lifandis AS elevate real-world insights
with enriched Scandinavian datasets
The strategic collaboration with IMS Health allows researchers to look
at a broader set of data in Norway as well as Sweden and other
Scandinavian markets through IMS Health’s existing real-world solutions
assets. Clients will now be able to benefit from the Lifandis integrated
partnership in addition to IMS Health’s other information assets,
scientific capabilities and involvement in research projects.
ESTABLISHED EXCELLENCE WITH GLOBAL IMPACT
This development enriches an already distinctive offering that allows
healthcare researchers to develop globally and locally relevant insights
into populations, diseases and treatment experience.
The ability of the IMS Health and Lifandis team to create holistic views
across settings of care over time enables Scandinavian-based affiliates
and global headquarters to answer meaningful and challenging
research questions, based on • Long-term study reviews for anonymous patients across
settings of care • Difficult-to-get patient attributes for more meaningful
treatment journeys • Information to determine the economic value of different
outcomes measures • Analytics to support research from epidemiology to
comparative effectiveness
TOWARDS A REALNTIME UNDERSTANDING
The extension of IMS Health’s RWE capabilities in Northern Europe marks
another important step in helping healthcare decision makers identify,
link and interpret real-world outcomes in near real time.
For further information on the IMS Health/Lifandis AS approach to
RWE and the exciting opportunities for integration of complex datasets
in the Scandinavian region, please email Patrik Sobocki at
Psobocki@se.imshealth.com or Christian Jonasson at cj@lifandis.com
FIGURE 1: LEGISLATION, CONSENT ANd A PERSONAL Id CREATE
POTENTIAL FOR HIGH QUALITY, COMPREHENSIVE dATASETS
Further expanding IMS Health’s distinctive and growing
real-world evidence capabilities in Northern Europe, the
company has announced a collaboration with Lifandis AS, an
independent company that works closely with the HUNT
Research Centre in Norway. The agreement combines IMS
Health’s Pygargus extraction methodology with access to
the HUNT biobank and databank, as well as other
Norwegian biobanks and health registries, enabling the
creation of significantly enhanced real-world datasets.
Underscoring the rising importance of Scandinavia as a rich
hub for RWE, this linkage affords one of the most holistic
patient-level views imaginable with potential for
unprecedented insights of both local and global relevance.
RICH SETTING FOR REALNWORLD DATA
Scandinavia is unrivalled in opportunities to generate RWE given its
well-structured public healthcare, long established high-quality
electronic medical records (EMR) and mature regulatory research
framework. In a first-of-its kind RWE approach, IMS Health brings the
most complete, integrated view of patient-level care through
anonymous EMR data along with national and disease-specific registers.
The new collaboration with Lifandis in Norway extends application of the
IMS Health Pygargus patented extraction methodology, first launched in
Sweden, to the HUNT biobank and databank, recognized by international
researchers for its value in personalized medicine (biomarker Id and
validation, disease etiology, patient subgroup stratification), epidemiology
(RWE, post-marketing studies, burden of disease, comparison of treatment
outcomes), drug discovery (target identification, target validation) and
clinical trial optimization. Containing unique patient data from 125,000
anonymous individuals, with more than 25 years of follow-up, and
covering 6,000 distinct variables, the Nord-Trøndelag Health (HUNT) Study
is one of the largest population-based health studies ever performed.1
UNIQUE FOUNDATION FOR TAILORED RESEARCH
Lifandis was founded to drive partnership between Norwegian biobanks,
academia and industry, and the company has also established a strong
foothold within register-based epidemiology. Its heritage includes
recruitment of at least 1.4 million Norwegians, around 30% of the
population, into consent-based research biobanks based on population-based
studies, with an additional 25-30 million samples in clinical
biobanks. Legislation, broad consent and the existence of a personal
identification number opens up the opportunity to build high-quality
and comprehensive datasets with access to more than 40 healthcare and
disease-specific registries, hospital and primary care EMRs and separate
endpoint registries with validated outcomes (Figure 1).
Importantly, while affording direct insights from Scandinavia, the data
can also inform scientific research to support global decisions across a
range of disease areas.
HUNT Biobank
HUNT Databank
Healthcare
Registr
Registries
ies
Electronic Medical
Records
Endpoint
ies
Registr
Registries
P
ersonal ID
Personal Da
tabank
Archival issue
samples
1 Krokstad S, et al. Cohort Profile:The HUNT Study, Norway. Int. J Epidemiol. 2013
Aug; 42(4): 968-77
PAGE 2 IMS HEALTH REAL-WORLD EVIDENCE SOLUTIONS & HEOR

nEWs EMERGING HEALTHCARE TRENDS
Research from IMS Health informs opportunities for harnessing trends to achieve
the triple aim of US health reform
Study reveals ten dynamics for policy prioritization
in US managed care
At a time of tremendous flux in the US healthcare system, a
new report, underpinned by IMS Health research, has
identified potential for strategies to achieve the triple aim
of health reform (improved care, improved health and
reduced cost) leveraging the top emerging healthcare
trends. The findings provide real-world insights into key
policy priorities for healthcare stakeholders.
The report, “Ahead of the Curve: Top 10 Emerging Health Care Trends
– Implications for Patients, Providers, Payers and Pharmaceuticals”
was developed under the direction of the American Managed Care
Pharmacy (AMCP) Foundation, in collaboration with Pfizer, Inc. The
Foundation is a research, education and philanthropic organization
established in 1990 with the goal of advancing collective knowledge and
insights into major issues associated with the practice of pharmacy in
managed healthcare settings.
In seeking to help stakeholders proactively prepare for the impact of
changes in the US healthcare marketplace, the collaborative project was
designed to systematically identify and assess current and emerging
trends impacting healthcare delivery and MCP practices.
Reflecting a strong focus on partnering with stakeholders to improve
patient outcomes and advance healthcare globally, the research was
conducted by IMS Health on behalf of the Foundation, along with
development of the report itself. The company has established excellence
in generating scientifically credible real-world evidence that drives
powerful insights for more efficient decision making. The process
employed was designed to add scientific rigor by drawing on secondary
research evidence in addition to key opinion leaders’ insights. It was
systematic and replicable and drew upon the cross-functional expertise and
knowledge base of team members from multiple practice areas.
The six-month program of research followed a two-part methodology in
which distilled information from a targeted literature review was
analyzed by an advisory panel of healthcare thought leaders from
academia, industry, managed care, government and patient advocacy.
The panel was engaged to validate, identify and prioritize trends and
provide insight into implications across healthcare stakeholders. This
process included participation in a full-day, facilitated discussion and
trends assessment.
TOP TEN TRENDS DRIVING POLICY PRIORITIES
The top ten trends identified for their impact over the next five years are
1. Migration from fee-for-service to new provider payment models
that better align incentives for cost control and high-quality patient care
2. Consolidation of healthcare stakeholders, fueling standardization
of decisions and opportunities to evolve patient care practices
3. Widespread use of data and analytics in patient care, providing
novel opportunities for improving care effectiveness and efficiency
4. Increased utilization and spending for specialty medicines,
burdening payers and manufacturers to develop novel approaches
to formulary design and pricing practices that ensure patient access
5. Medicaid expansion, shifting a larger portion of economic risk to
payers and providers and driving creation of new models for care
delivery and tactics to improve efficiency
6. Migration to a value-oriented healthcare marketplace, reflecting
new approaches to balancing care quality and cost
7. Growth and performance of accountable care organizations, with
long-term success requiring investments in data structure and
analytics and willingness to evolve new models of care
8. Greater patient engagement through technology, which will
empower patients and providers to enhance practices for managing
and coordinating healthcare
9. Increasing patient cost-sharing, to curtail costs and incentivize
patient involvement
10. Healthcare everywhere through new tools and mobile
applications, with new avenues for patient engagement and
new healthcare delivery roles as wellbeing becomes a
community-wide effort
A NEED FOR NOVEL SOLUTIONS
Overall, the report suggests an advance towards a system of patient-centric
holistic care over the next five years, with shared accountability
across stakeholders and value being the core currency of the healthcare
marketplace – changes that are expected to translate into improved
patient outcomes. In preparation, stakeholders will need to move
beyond conventional practices and generate novel solutions that
improve patient metrics and tracking, enhance patient engagement and
find the balance between driving accountability, curtailing costs and
incentivizing. Specifically, this will involve
• Providers becoming increasingly accountable for driving care
efficiency. This may require a fundamental shift from conventional
care approaches. To support the transition, providers can leverage
healthcare technologies and the expansion of patient data to drive
quality in patient care and improve care processes. • Payers designing and implementing new payment models that
share risk and drive accountability across stakeholders and
populations with varying needs and requirements. They should
increasingly leverage technology tools, patient data and health care
analytics to better engage patients and track provider performance. • Pharmaceutical companies experiencing increased demand for
proof of value and real-world effectiveness data beyond trial-based
safety and efficacy, and being asked to share the risk for supporting
improved patient outcomes. They can prepare by investing in
evidence-generation capabilities that move beyond clinical trials to
leverage real-world data from provider and payer organizations.
The report concludes that while the path forward will vary by
stakeholder, all players in the US healthcare system will need to place
the patient center stage and consider their role in supporting long-term
improvements in patient health in a more holistic manner.
For further information, the report is available to download from the
Foundation’s website at www.amcpfoundation.org

nEWs RWE DEBATE
Experts gather with IMS Health to accelerate the application of real-world evidence for
maximum utility in healthcare decision making
Stakeholders unite to improve collaboration
in realizing RWE potential
Alongside greater demand for real-world evidence and
increasing recognition of its value across the healthcare
spectrum, there are clear signs that many stakeholders still
struggle to act on its potential. Its appropriate use can
deliver benefits to all, but more open dialogue and
enhanced collaboration between relevant stakeholders is
needed. Together with other partners, IMS Health works to
help all constituent groups achieve the common goal of
advancing healthcare.
As part of the company's commitment to accelerating the application
of RWE in pricing and market access decisions, two recent initiatives in
the US and UK have broken new ground in connecting perspectives and
broadening thinking about key issues for the current use of RWE and
solutions for realizing its true value.
US: REALNWORLD EVIDENCE LEADERSHIP SYMPOSIUM
A first-of-its-kind event, the Real-World Evidence Leadership Symposium
was held on 4 November 2014.
Co-sponsored through a thought leadership partnership between
IMS Health and Johns Hopkins Center for drug Safety & Effectiveness
in baltimore, Md, “Realizing the full potential of real-world evidence
to support pricing and reimbursement decisions”, offered a forum for
invited payers, pharmaceutical executives and academicians to engage
in frank and constructive discussion on how payers and life sciences
companies were using RWE and to look for pragmatic opportunities to
maximize its utility in pricing and reimbursement decisions. A key focus
was to explore potential collaborations between pharma and payers in
RWE generation.
Under the Chairmanship of dr. Lou Garrison, Professor and Associate
director in the Pharmaceutical Outcomes Research and Policy Program,
department of Pharmacy, at the University of Washington in Seattle, the
debate was structured into three sessions
1. Review of illustrative use cases showing effective and ineffective
use of RWE, to demonstrate opportunities and limitations facing its
broader application
2. Facilitated payer panel to discuss payer views on the role of RWE
in decision making and requirements for further use
3. Discussion and proposed solutions as a starting point for action
to identify potential for united efforts to increase the value of RWE
shaping the RWE opportunity
Reactions to the symposium from both speakers and participants
underscored its value in highlighting opportunities for making RWE
more core to pricing and market access decisions, whilst also capturing
a need for life sciences companies to hear directly from payers that their
RWE can have impact in order to increase their confidence in its use.
The key discussion points and actionable outputs from the symposium
are being taken forward for further exploration in post-forum research,
the findings of which will form the basis of an authoritative white paper
to further the discussion and serve as a catalyst for more collaborative
generation and use of RWE in the future.
UK: DECISION MAKING USING REALNWORLD DATA
Pushing forward the RWE conversation in the UK, the first IMS Health
Decision Making Using Real-World Data Conference, “Understanding the
changing landscape of patient data: Informed decision making in the
UK healthcare market”, was held on 30 September, 2014. The event
was organized in response to a request from IMS Health clients to learn
more about RWE best practice in the UK and its use by other players in
the healthcare arena. bringing together life sciences industry leaders
with a variety of healthcare stakeholders, the conference afforded a
unique opportunity to explore, through open debate, the ways that real-world
data should be utilized for healthcare decision making in the UK.
The event and panel discussion were chaired by Professor Sir Alasdair
breckenridge, former Chairman of the UK Medicines and Healthcare
Products Regulatory Agency (MHRA) who brought a deep understanding
of pharmaceutical regulators, their goals and requirements.
Broadening thinking on optimizing use of RWE
The presentations offered a variety of perspectives and cross-sectional
view of decision making. Speakers included dr Sarah Gardner, Associate
director of R&d at the National Institute for Health and Care Excellence
(NICE); Kevin V. blake, Scientific Administrator, best Evidence
development Office, at the European Medicines Agency (EMA); Skip
Olson, Global Head of HEOR Excellence at Novartis; and Professor Liam
Smeeth, Professor of Clinical Epidemiology and Head of the department
of Non-communicable disease Epidemiology at the London School of
Hygiene and Tropical Medicine. IMS Health was represented by dr. Patrik
Sobocki who shared the company’s view of RWE and vision for its use.
Among the topics covered by the panel of guest speakers were • Real-world data and the changing policy landscape • EMA use of best evidence in regulatory decision making • Leadership in RWE: An industry perspective • Leveraging patient-centric data and generating evidence across the
product lifecycle • Confounding, its impact and how it can be managed to maximize
the benefit of RWE
The speakers discussed how effectively RWE is used in their sectors
currently, how they believe it should be used to help decision making
and how they see the landscape changing in the future.
Feedback from both speakers and attendees was extremely positive and
there are plans to develop and expand the "Decision Making Using Real-
World Data" conference for 2015.

nEWs IMS CORE DIABETES MODEL VALIDATION
IMS CORE diabetes Model demonstrates continued credibility as the leading tool for
policy and reimbursement strategy in diabetes
Major validation upholds relevance of
IMS CORE diabetes Model
The IMS CORE diabetes Model (CdM) is a well-published
and validated simulation model that predicts long-term
health outcomes and costs in type 1 and type 2 diabetes.
For those developing policy and implementing decisions
informed by CdM analyses, confirmation that the model
remains contemporary and validated is essential. Findings
from a new validation to recent diabetes outcome studies1
reaffirm the model’s suitability to support policy decisions
for improving diabetes management.
disease simulation models are increasingly being applied to inform a
wide range of issues in healthcare decision making. Their ability to
project long-term outcomes and costs on the basis of short-term study
data is particularly relevant in a chronic condition like diabetes, given
its progressive course, associated complications and high and growing
economic burden.
The market-leading CdM is designed to assess the lifetime health
outcomes and economic consequences of interventions in diabetes, and
comprises 17 interdependent sub-models that simulate the major
complications of the disease. It allows estimation of direct and indirect
costs; adjusts for quality of life; and enables users to perform both cost-effectiveness
and cost utility analyses. It is routinely used to inform
reimbursement decisions, public health issues, clinical trial design and
optimal patient management strategies.
ROBUST VALIDATION PEDIGREE
Validation to external studies has been an intrinsic part of the CdM’s
development process. In a major evaluation in 2004, its operational
predictive validity was demonstrated against 66 clinical endpoints from
11 epidemiological and clinical studies. Evolution of the model also
reflects its strong links with the Mount Hood Challenge, a recognized
biennial forum for comparing the structure and performance of diabetes
health economic models with data from clinical trials (see Insights on
page 50).
RECENT ENHANCEMENTS
An ongoing commitment to ensuring that the CdM remains the best
available tool for economic evaluations in diabetes has seen the model
undergo a series of significant updates in recent years. These include • Ability to model individual anonymous patient-level data • Incorporation of treat-to-target efficacy data for HbA1c • Inclusion of a detailed hypoglycemia sub-model • Expansion of variables for probabilistic sensitivity analysis • Addition of UKPdS 68 and 82 risk equations
ENSURING CONTEMPORARY RELEVANCE
To ensure the CdM’s continued relevance and accuracy following these
enhancements, the aim of the latest validation study, published in 2014,
was to examine the validity of the updated model to results from recent
major long-term and short-term diabetes outcome studies. Particular
emphasis was placed on cardiovascular (CV) risk.
Independent researchers with unrestricted access to the CdM and its
source code worked with IMS Health to verify (ensure the model is coded
as intended and free from errors) and externally validate (quantify how
well outcomes observed in the real world are predicted) the model. In
total 121 validation simulations were performed, stratified by study follow-up
duration, study endpoints, year of publications and diabetes type.
goodness of fit
A number of statistical measures of goodness-of-fit were used, including • Testing of null hypothesis of no difference between the
annualized event rates (observed vs. predicted) and relative risk
reduction across all validation endpoints • Assessment of whether the confidence intervals for the number of
events predicted by the model and those reported in the
validation studies overlapped • Evaluation of goodness-of-fit between simulated and observed
endpoints for trials, endpoints, treatment arm, and date of study
using the mean absolute percentage error (MAPE) and the root
mean square percentage error (RMSPE) • Scatterplots of observed vs. predicted endpoints along with the
coefficient of determination (R2)
Impact of choice of CV risk equations
The CdM currently uses, amongst others, CV risk equations derived from
the United Kingdom Prospective diabetes Study Outcomes Model
(UKPdS68) but, given the increasing choice of equations that is
emerging, assessing the continued relevance of UKPdS68 is essential.
As part of the validation exercise, the absolute level of risk and relative
risk reduction was compared for 12 CV disease risk equations developed
specifically for T2dM patients.
RESULTS
At conventional levels of statistical significance, the study found that
the CdM fitted the contemporary validation data well, supporting the
model as a credible tool for predicting the absolute number of clinical
events in dCCT- and UKPdS-like populations.
Underscoring the significance of these results, Professor Phil McEwan of
Swansea University, the lead researcher of the study, emphasized that
"Organizations developing policy and implementing decisions informed by
CDM require the reassurance that the model and its results are current and
validated. This study helps to demonstrate that the model is a validated tool
for predicting major diabetes outcomes and consequently is potentially
suitable for supporting policy decisions relating to disease management in
diabetes."
A copy of the full validation study is available to download online at:
http://www.valueinhealthjournal.com/article/S1098-3015(14)01928-7/pdf
For further information on the IMS CORE diabetes Model, please
email Mark Lamotte at Mlamotte@be.imshealth.com
1 McEwan P, Foos V, Palmer JL, Lamotte MD, Lloyd A, Grant D. Validation of the IMS
CORE Diabetes Model. Value in Health, 2014; 17: 714-724

InsIghts DISEASE-SPECIFIC RWE
The author
Enabling disease-specific RWE
through fit-for-purpose RWD
Increased stakeholder demand and the greater supply of
electronic real-world data are expanding the application of
real-world evidence across the product lifecycle. The most
successful organizations are developing RWE platforms,
capabilities and analytical methodologies focused on
therapeutic areas. Increasingly, understanding how the
characteristics of a particular disease area can influence the
availability and use of real-world data for evidence generation is
important in setting strategies that create differentiation.
Rob Kotchie, M.CHEM, MSC
is Vice President, RWE Solutions, IMS Health
Rkotchie@imshealth.com

DISEASENDRIVEN DETERMINANTS OF RWE
In seeking to inform the ease and extent of RWE
development in a particular therapeutic class, IMS Health
has identified five key characteristics of a disease area
that have influenced the evolution of RWD development
to date
1. Routine capture of clinical measures
2. Nature of the critical endpoint
3. Number of treatment settings
4. Length of follow-up
5. Available sample size
By assessing each disease area against these five
characteristics it is possible to identify the specific factors
limiting an expansion of RWD use and the levers that can
be engaged to accelerate future adoption. This point is
illustrated in Figure 2 and discussed below for the
projected top five therapy areas in 2017.
Oncology: Complex patient subgroups
For oncology, a disease area that is often more amenable
to RWE research due to the nature of the critical endpoint
and frequent short length of required patient follow-up,
analysis can be often limited by the complexity of patient
subgroups and the need to capture detailed information
on disease staging, therapy sequencing, role of surgery
and patient biomarker status.
These challenges are now being overcome to a degree
by healthcare stakeholders working together to link
important rich clinical information with genomic and
proteomic data, increasing the value and uses of RWD in
this area.
For example, RWD is increasingly being leveraged in
oncology to facilitate pricing and reimbursement of
therapies by use, enabling a mechanism for greater
alignment between manufacturers and healthcare payers
and providers on the value and costs of treatment in a
specific indication or patient population.
continued on next page
A framework for reference in key disease areas
Globally, intensified pressure to obtain better value for
healthcare spending has elevated the importance of
real-world evidence (RWE) as an enabler of improved
healthcare decision making. Increased stakeholder
demand and the greater supply of electronic real-world
data (RWD) are expanding its application across the
product lifecycle as companies become attuned to the
insights it can deliver.
Leading life sciences organizations are now using RWE to
support clinical development, improve launch
performance and drive better commercial results. The
most successful are moving beyond a product-specific,
study-based approach to develop RWE platforms,
capabilities and analytical methodologies focused on a
single or set of therapy areas to drive sustained value
across their franchises.
As these trends continue, the ability to compare and
understand how the characteristics of a particular disease
area can influence the availability and use of RWD is an
important step in setting focused and relevant RWE
strategies that create differentiation and drive
achievement of commercial goals. This article offers a
framework for assessing RWD availability by therapy area
to guide internal decision making.
NUANCED CHALLENGES FOR RWE RESEARCH
By 2017, IMS Health estimates that the largest therapeutic
classes in the developed markets will include a
combination of both traditional primary care and
specialized areas, led by oncology, diabetes, anti-TNFs,
pain and asthma/COPD (Figure 1). Each of these disease
areas presents markedly different patient populations,
unmet medical need, standards of care and disease
outcomes, leading to a nuanced set of challenges for
RWE research.
20 = 71% market
value by
2017
TOP

InsIghts DISEASE-SPECIFIC RWE
Diabetes: Extended timeframe and multiple
care settings
In diabetes the generation and application of RWE, either
by researchers to support burden of disease, comparative
effectiveness or safety research or by commercial
functions for forecasting or sales and marketing purposes,
is often hindered by the need to track patients over long
periods of time and across multiple settings of care. In
other words, in order to infer the effects of a diabetes
intervention on delaying the worsening of a secondary
condition (eg, renal disease) or a reduction in a related
complication (eg, microvascular or macrovascular events)
patients must be followed over several years. This
includes tracking their admissions and discharge to and
from hospital, and across multiple treatment centers.
Hence, to fully assess the comparative effectiveness of a
diabetes intervention in the real-world setting requires
linking one or more datasets across both ambulatory and
specialist treatment settings, and/or combining a closed
database of medical and pharmacy claims with EMR data
to provide meaningful clinical data on outcomes and
confounding factors such as Body Mass Index and HbA1c.
Despite the proliferation of data in a primary care disease
like diabetes, the challenge is in bringing it together in a
meaningful way that will increase the usability of
diabetes RWD.
Anti-tnFs/Pain: Patient-reported endpoints
In the case of anti-TNFs or therapies to treat pain, RWE
research is often limited by the lack of routine capture of
patient-reported endpoints in clinical practice. While
disease-specific instruments that are used to assess a
patient’s response to therapy are systematically applied in
clinical trials, they are typically either not routinely
recorded in clinical practice or the data is stored in
unstructured clinical notes making it challenging and
time consuming to extract, analyze and interpret.
Asthma/COPD: Routine tests and acute events
Similarly, in other chronic disease areas such as
asthma/COPD, research can be restricted by the lack of
routine capturing of test results used to assess the long-term
deterioration of the disease (eg, spirometry
measures such as FEV1) or detailed descriptions of acute
episodic events, such as admission to hospital for a major
COPD exacerbation, or the documentation of rescue
medication use for a mild to moderate exacerbation.
FIGURE 1: LEAdING THERAPEUTIC CLASSES IN 2017 WILL INCLUdE PRIMARY CARE ANd SPECIALIST AREAS
Oncology
Diabetes
Anti-TNFs
Pain
Asthma/COPD
Other CNS Drugs
Hypertension
Immunostimulants
HIV Antivirals
Dermatology
Antibiotics
Cholesterol
Anti-Epileptics
Immunosuppressants
Antipsychotics
Antiulcerants
Antidepressants
Antivirals excluding HIV
ADHD
Interferons
Developed Markets Sales in 2017 (LC$)
$74-84Bn
$34-39Bn
$32-37Bn
$31-36Bn
$31-36Bn
$26-31Bn
$23-26Bn
$22-25Bn
$22-25Bn
$22-25Bn
$18-21Bn
$16-19Bn
$15-18Bn
$15-18Bn
$13-16Bn
$12-14Bn
$10-12Bn
$8-10Bn
$7-9Bn
$6-8Bn
Source: Rickwood S, Kleinrock M, Nunez-Gaviria M. The global use of medicines: Outlook to 2017.
IMS Institute for Healthcare Informatics, 2013 Nov.
Others
29%
Top 20
Classes
71%

Levers
Supplementation
Supplementation NLP
Linkage
Linkage retention modeling
Pooling
FIGURE 2: FRAMEWORK FOR dETERMINING CHALLENGES OF RWE GENERATION bY dISEASE
Abundant
Hard
Single
Short
Infrequent
Soft
Multi
Long
Large Small
Oncology Diabetes Anti-TNF Pain Asthma/COPD
Routine capture
of clinical
measures
Nature of
the critical
endpoint
Number of
treatment
settings
Length of
follow up
Available
sample size
LEVERAGING PROGRESS TO REALIZE VALUE
Growing need and rapidly expanding applications of RWE
are driving the development of innovative techniques to
link, supplement and pool data sources for deeper and
more meaningful research in this area.
The deployment of data encryption engines and greater
collaboration between key players is enabling ever
increasing scope to link anonymous information across
datasets and settings of care, while preserving patient
confidentiality and appropriate use.
Innovative techniques are now available to supplement
secondary data from the electronic health record through
novel primary data collection from physician and/or
patients at the point of care (‘over the top’ data collection),
and deploy Natural Language Processing (NLP) to extract
additional rich information from clinical notes in a HIPAA-compliant
manner.
These developments are providing life science researchers
with unprecedented access to comprehensive disease area
real-world datasets spanning multiple sources and settings
of care - with sufficient sample size and patient follow-up
to power an expanded set of RWE applications.
As companies look to maximize the value of RWE in their
organization, a focus on understanding the specific needs
and challenges for evidence generation presented by
disease areas of interest will be a key step to leveraging
the progress being made and realizing its full potential
across their franchises.
Understanding how the characteristics of a disease area can influence
availability and use of real-world data for evidence generation is
increasingly important. “
”

InsIghts RWE ROADMAP
The authors
A roadmap for increasing
RWE use in payer decisions
Real-world evidence has been part of healthcare for more than
30 years. Despite this, its application to really improve the
efficiency of healthcare delivery remains uneven and siloed.
Some of the greatest opportunities lie within the realms of
collaborative and partnership initiatives between stakeholders,
especially payers.
Marla Kessler, MBA
is Vice President, IMS Consulting Group
Mkessler@imscg.com
Ragnar Linder, MSC
is Principal, RWE Solutions & HEOR, IMS Health
Rlinder@se.imshealth.com

Bridging the gap between promise and reality
Real-world evidence has been part of healthcare for over
30 years, applied at varying levels by regulators, clinicians,
payers and manufacturers to inform decisions, build
programs and improve health. IMS Health has documented
more than 100 case studies where RWE has actively
influenced product labeling, price, access and use.1
Despite this, the application of RWE to really improve the
efficiency of healthcare delivery remains uneven and
siloed. Does this suggest a lack of comprehensive, quality
data? Are healthcare professionals, policy makers and
other key stakeholders waiting for better tools? Are the
skills sets to link and analyze data not widely accessible?
In fact the evidence suggests that the ability to produce
RWE is expanding, and rather quickly. However, the gap
between the exponential increase in RWE sources and
the capacity to harness these effectively is also growing.
Our research suggests that this widening gap between
the promise and reality is due to three critical – but
manageable – barriers.
GROWING VOLUME BUT UNREALIZED POTENTIAL
The quantity and importance of RWE has expanded
tremendously in recent years (Figure 1). RWE is generated
and applied throughout the lifecycle of pharmaceuticals
and other medical interventions to demonstrate
effectiveness, safety and value. It can be used for
population health management, for example in
identifying significant health factors by geography or
demographics for the design and evaluation of
interventions to improve health. It can enable better
understanding and characterization of disease
epidemiology, treatment paradigm and associated
resource utilization. It can inform quality of care
assessment, point of care decision guides and
translational research projects. And it can also serve to
assess a drug’s performance outside the randomized
controlled trial (RCT) setting and describe any shifts in
practice once the drug is approved and used.
" " " " "
FIGURE 1: THERE HAS bEEN AN EXPLOSION OF REAL-WORLd dATA FOR ANALYSIS
of payer respondents had no confidence in
the economic evidence provided by pharma 44%

FIGURE 2: CASE STUdY bREAdTH ANd VOLUME dEMONSTRATE EXISTING RWE dEMANd
22
21
16
11
10
9
4
3 3
2
Italy USA UK Sweden Canada Spain Netherlands France Germany Denmark
Label Launch access Ongoing access Price Use
Source: Hughes B, Kessler M. RWE market impact on medicines: A lens for pharma. IMS Health AccessPoint, 2013; 3(6): 12-17
25
20
15
10
5
0
Number of case studies
While RCT data is still regarded as being top of the
evidence hierarchy, there has been an increased use of
approaches that assess patient outcomes and follow all
the care and interventions they receive. Real-world data
(RWD) is now being used to complement RCT
information, providing valuable evidence of the way
pharmaceuticals are being used in practice and in many
populations, which cannot be gained from RCTs.
The breadth and volume of demand for RWE by payers
across markets is shown in Figure 2, based on research
conducted in 2013.1 In addition, payers are involved in a
plethora of RWE activities, building RWD for commercial
purposes (eg, Humana, Lifandis), collaborating more
broadly with other payers (eg, Health Care Cost Institute),
or simply using their own data for internal assessments.
Clearly, payers have not ‘opted-out’ of RWE. And yet
examples of them accepting industry-generated RWE or
working collaboratively with pharma to generate RWE are
few. These two key players may often be on opposite sides
of a negotiating table but opportunities exist for
partnerships that could potentially improve the entire
healthcare system. While current examples do provide
hope for a more collaborative future, they also force a more
fundamental question: what are the barriers to greater use
of RWE by payers and their willingness to work with
pharma and other stakeholders to broaden its application
in pricing, reimbursement and access decision?
SOME IDENTIFIED BARRIERS
In reviewing this issue with many payers and pharma
executives and in published literature, conferences and
other forums, barriers emerge in three key areas: data
and technology; science; and collaboration. While not
exhaustive or quantified, the challenges discussed
below within these areas provide a view of the
roadblocks being encountered.
Data and technology barriers
• Data infrastructure
While fully adjudicated claims data is structured with
fewer and more consistently defined variables, the
volume of it is expanding even as it is increasingly
linked with laboratory records, medical records, patient
social media and now genomic data, stretching the
bounds of healthcare informatics. All players in the
healthcare system seek more clinical and patient
outcomes information but now appear to be drowning
in vast amounts of data without it being sufficiently
complete for effective decision making. A study from
the Health Research Institute (HRI) in the US2 notes that
payers themselves believe they lack an adequate data
infrastructure to apply RWE in areas such as outcomes-based
contracting. And although the related
technology is growing and scalable, it is too expensive
and time consuming for most stakeholders to realize
its full potential at this time.

• Data extraction and linkage
Many payers have built distinctive capabilities in
understanding claims-related data but clinical data
requires a different set of expertise. The magnitude of
the challenge is just as great for pharma although its
nature is different. Companies may have acquired
substantial data and even technology integration
solutions but the data sits in functional and geographic
silos using new and old technologies, making it
challenging to link let alone analyze.
Even in a country like Sweden, where almost all patient
data can be tied to a consistent national social security
number, linkage is possible but not immediate.
• Data programming and processing
Speed is critical. However, a well-constructed research
study involving intensive SAS programming can take
months to conduct, extended by delays in gaining
answers to questions, with knock-on implications for
the timeliness of the insights delivered.
scientific barriers
• Lack of consistent RWD methodologies
The insights to be gained from RWD are substantial,
but the growing availability of data highlights
important methodological challenges. Even at a basic
level, questions can arise. For example, what defines a
diabetic patient? Is it based on medications taken, a
recorded diagnosis code, or an actual laboratory or
series of laboratory results?
Not every patient record contains all that information
or even some of it. This quickly leads to more complex
challenges: when should data matching be
deterministic versus probabilistic? When is it
acceptable to impute missing values? How will these
decisions bias the results? How can advanced analytics,
including predictive analytics, improve the quality of
and confidence in RWE? The expertise to deal with this
exists, but not always in-house. Furthermore, payers
can be skeptical of data because there is no easy way
of ensuring that the deployed methodologies are
sufficiently robust.
• Absence of standardized measures
The current lack of consensus around many key
measures means that even issues such as how
long a patient needs to demonstrate an outcome
before a treatment is deemed cost-effective, are not
universally agreed.
The variation in approaches can significantly impact
study results. Exploring methods used to score
physician spending patterns (cost profiling), a measure
frequently assessed by payers, a Rand Health research
study showed that even slight changes in attribution
rules can dramatically change the characterization of
physician performance. For example, “Between 17 and
61 percent of physicians would be assigned to a
different cost category if an attribution rule other than
the most common rule were used.”3
Collaboration barriers
• Lack of trust
This is perhaps the elephant in the room that everyone
is willing to talk about. While payers and pharma
should be aligned around patient outcomes, economic
incentives are more complex. The previously
referenced HRI study found that 44% of payer
respondents had no confidence in the economic
evidence provided by pharma.2 Fewer than 1 in 10
were very confident in using pharma-generated
information to evaluate a drug’s comparative
effectiveness.
For data holders, the need to protect patient privacy
and the integrity of the data being used has created
many hurdles to access. Even straightforward protocols
can take months to approve if each proposal is
evaluated individually.
• Lack of imperative
While some payers see their data as entirely adequate
to support comparative effectiveness and other
analysis, others are not even sure the analysis is
required to achieve their goals. If the main objectives
are managing unit costs of treatments, payers have
other mechanisms such as rebates, formulary design
and traditional analysis of claims data, which they may
find easier to use.
In parallel, many pharma companies can be risk averse
to generating RWE with a payer without fully
understanding what will be said and how it will
be used.
Some of the greatest opportunities for achieving the goal of improved
efficiency in healthcare lie within the realms of collaborative and
partnership initiatives between stakeholders, to ensure implementation. “
”

SOME POTENTIAL SOLUTIONS
None of the barriers referenced are insurmountable.
Indeed, interesting examples are already emerging of
innovative solutions on the path towards greater use of
RWE in pricing and reimbursement decisions.
• Evolution of methodologies and technology-enabled
analytics
This edition of AccessPoint alone spotlights the area of
predictive modeling where novel methodologies are
driving a new generation of applications in RWE (see
article on page 26). In these areas, researchers are
taking advantage of improved data and computing
power to run analytics that otherwise would have been
too time-consuming, if not impossible, to conduct.
• Richer data sources
Not every research question must rely on locally-sourced
data. In countries such as Scandinavia, more
than two decades of rich patient-level data exists
electronically. Technologies such as the IMS Pygargus
Customized eXtraction Program facilitate linkage
between the various sources by extracting the desired
data from an electronic medical record (EMR) to build
databases of EMR and register data. A 2014
retrospective cohort study linked national Swedish
mandatory registries to EMR data from outpatient
urology clinics to study prostate cancer (PC) patients.
The use of this approach provided a unique
understanding of the clinical course of PC that can
inform treatment and research across developed
markets – not only in Sweden.4
• Collaborations
Organizations such as the Healthcare Cost Institute
(HCCI) have been established with the goal of pooling
data (in this case, from US payers) and increasing its
quality. In reality, the value of cooperation between
stakeholders in different parts of the system – payers,
providers and pharma – will be critical, not only in
improving data sources but also in increasing buy-in to
and application of the insights from them. This check-and-
balance will enable stakeholders to put the patient
at the center of RWE and provide care that actually
improves outcomes.
In addition, it can enable a movement away from
different parties running analytics to stakeholders
working together to solve problems. For example,
RWE can support efforts to improve decision making,
adherence and efficient care delivery, where the
focus goes beyond analytics and ultimately to better
patient care.
• third-party involvement
The involvement of independent, objective third
parties can increase confidence in the underlying data
as well as the resulting analysis. It can also be an
important enabler of packaged analytics where data
can be used for a variety of applications within a
spectrum of pre-approved uses. A trusted third party
can deliver that protection. In addition, for data
providers interested in commercializing their data, a
third party can enable the full value potential of that
data to be captured across a range of research goals
involving many different types of organizations.
FULFILLING THE PROMISE
The importance of RWE is continuing to grow along with
its ability to inform critical decisions for payers, pharma
companies and other healthcare stakeholders. However,
the full impact of its potential has yet to be realized. This
article has considered some of the barriers to wider use
of RWE and proposed some solutions to address them.
Some of the greatest opportunities for achieving the
goal of improved efficiency in healthcare lie within the
realms of collaborative and partnership initiatives
between stakeholders, to ensure implementation.
Only then can we provide the best care for patients
and improve outcomes.
1 Hughes B, Kessler M. RWE market impact on medicines: A lens for pharma. IMS Health AccessPoint, 2013; 3(6): 12-17
2 Health Research Institute/PWC. Unleashing value: The changing payment landscape for the US pharmaceutical industry. May, 2012
3 Mehrotra D, Adams JL, Thomas WJ, McGlynn EA. Is physician cost profiling ready for prime time? Research Brief, Rand Health, 2010
4 Banefelt J, Liede A, Mesterton J, Stålhammar J, Hernandez RK, Sobocki P, Persson BE. Survival and clinical metastases among prostate cancer patients
treated with androgen deprivation therapy in Sweden. Cancer Epidemiology, 2014, Aug; 38(4): 442-7. doi: 10.1016/j.canep.2014.04.007. Epub 2014
May 27.

InsIghts PRIMARY CARE UTILIZATION IN CANADA
Diabetes complexities
drive resource
consumption in Canada
The authors
According to the OECD, Canada currently ranks 27 out of 34
member countries in the number of physicians per 1,000
persons.1 Around 15% of Canadians report either being unable to
access a primary care doctor or choosing not to do so.2 A new
IMS Health analysis of EMR data reveals diabetes as the main
consumer of GP resource among chronic conditions in Canada,
with key insights for improvement initiatives.
Sergey Mokin, MSC, MBA
is Consultant, CES, IMS Brogan
SMokin@ca.imsbrogan.com
Richard Borrelli, B. COMM, MBA
is Principal, CES, IMS Brogan
Rborrelli@ca.imshealth.com
Michael Sung, MSC, MBA
is Consultant, CES, IMS Brogan
Msung@ca.imsbrogan.com

A case study of EMR data in diabetes
LEVERAGING REALNWORLD EVIDENCE
Findings from the 2013 National Physician Survey in Canada
indicate that 64% of family physicians and 59% of specialists
now utilize electronic medical records (EMR) in their
practices.3 The improved availability of EMR data makes it a
powerful source of real-world evidence to better understand
demands on the healthcare system. In seeking to evaluate
primary care utilization in the country, a study was
conducted using Canadian data from the IMS Evidence 360
EMR database. This provided access to a panel of around 500
general practitioners (GPs) and specialists covering more
than 500,000 anonymous patients as a sample of the
Canadian population in major chronic indications.
Objectives
The cross-sectional EMR study had three key objectives
1. Identify medical conditions that are the highest
consumers of physicians’ time in Canada, measured in
visits per patient per year
2. Describe the contributing factors for the medical
condition associated with the most frequent visits per
patient per year
3. Propose areas of high potential impact for further
investigation and intervention
Methodology
A cohort of all patients with at least one physician visit
recorded during the study period of June 2013–May 2014
was extracted from the EMR dataset. The overall
concentration of patient visits and average visits per
patient was then determined across different diagnosed
conditions. These conditions were prioritized based on
the average visits per patient, and statistical significance
calculated to identify the top consumer of physicians’
time for both the acute and chronic conditions.
STUDY FINDINGS
Primary care system utilization overview
In the study period, a total of 122,296 unique patients
recorded visits to physicians in the EMR database. The
concentration of visits showed that 10% of patients were
responsible for nearly 40% of primary care visits (Figure 1).
FIGURE 1: 10% OF PATIENTS ACCOUNTEd FOR 40% OF PRIMARY
CARE VISITS
Frequency of visits Vs. Number of patients
concentration curve
100
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100
% Patients
% Visits
Among the patients with chronic conditions, those with
diabetes made more repeat visits to a physician, as
indicated by the significantly higher average number of
visits per patient (2.6 per year) compared to other chronic
diseases (Table 1A). Among the acute conditions (which
were not studied further), patients with diseases of the
respiratory system had the highest average number of
visits per year (1.6 per patient) over the study period
(Table 1B). The further analysis focused on diabetes given
its chronic status and the significantly larger portion of
year-to-year healthcare spending on this condition.
TAbLE 1A: CHRONIC CONdITIONS
Medical Condition
Diabetes mellitus
Mental health disorders
Hypertension & other heart diseases
Chronic musculoskeletal system & connective tissue disorders
Chronic diseases of the respiratory system
Patients
2765
5901
4764
9263
3970
Visits
7205
11425
8270
13906
5319
Visits per patient
2.61
1.94
1.74
1.50
1.34
p-value*
<0.001
<0.001
0.066
<0.001
TAbLE 1b: ACUTE CONdITIONS
Medical Condition
Acute diseases of the respiratory system
Diseases of the urinary system (cystitis)
Family planning, contraceptive advice, advice on sterilization or abortion
Immunization (all types)
Acute musculoskeletal system & connective tissue disorders
Diarrhea, gastroenteritis, viral gastroenteritis
Patients
15706
5155
3820
4702
1970
2205
Visits
25083
6609
4844
5627
2354
2522
Visits per patient
1.60
1.28
1.27
1.20
1.19
1.14
p-value*
<0.001
0.92
<0.001
0.31
<0.001
Note: ICD-9 Code 078 containing other diseases due to virus was excluded due to potential for multiple viral infections to be captured under this single code
*p-value for the
Wilcoxon rank sum
test measures the
significance of the
difference in
visits/patient
between each
medical condition
and the next highest
medical condition

More than 70% of patients were treated with metformin.
However, multiple classes of anti-diabetic medications
were used to manage the disease, with DPP-IV inhibitors
and sulphonylureas being the next two most frequently
prescribed (Table 2). Diabetic patients were also likely to
be taking medications for cholesterol and triglyceride
control as well as for hypertension or other cardiovascular
conditions (Table 3). The type and prevalence of
concomitances were consistent with an older and mostly
overweight patient population.
Of patients whose med lab test results were available and
who had been treated with an anti-diabetic, distribution
analysis of their most recent HbA1c and fasting glucose
levels (Figure 4) showed that 51% did not meet the
HbA1c control threshold and 60% were out of control
based on the fasting glucose threshold.
Patients on metformin alone were compared with those
who had metformin plus at least one other anti-diabetic
in the study period. There was a statistically significant
relationship between the medication regimen (metformin
vs. metformin plus other) and achieved control state (in
control vs. out of control) within the study period (Table 4).
Fasting glucose and HbA1c levels were significantly
higher for patients treated with metformin and another
anti-diabetic in the study period. These patients also had a
significantly higher number of GP visits (Table 5). However,
further studies are required to determine the link between
the medications prescribed and control of diabetes.
FIGURE 3: bMI dISTRIbUTION OF dIAbETIC PATIENTS (N=1697)
0.4%
17.7%
30.8%
51.0%
<18.50 18.50-24.99 25.00-29.99 >30.00
Resource use contributors in diabetes
To determine potential contributors to the high level of
resource use in diabetes, data on its associated
demographics, co-morbidities/concomitances and lab
tests was extracted and analyzed. All diabetic patients
were identified in the cohort on the basis of having at least
one ICD-9 diagnosis code 250 or at least one prescription
for an anti-diabetic described by the ATC code A10.
Body Mass Index (BMI), HbA1c and fasting glucose levels
were analyzed for the diabetic cohorts based on the latest
available result within the study period. Patients with
fasting glucose >6.9 mmol/L or HbA1c >7% were further
segmented as ‘out of control’. Those treated with a
metformin product alone for the entire study period and
those who received metformin plus another anti-diabetic
class in the study period were also segmented. Statistical
tests were conducted to determine if observed differences
between patient segments were statistically significant.
Patients
A total of 4,390 diabetic patients recorded physician visits
in the EMR dataset over the study period. More males
(55%) than females (45%) were observed among these
patients, which is representative of the Canadian diabetic
population (54% males vs. 46% females).4 The majority
(73%) were over 50 years of age (Figure 2). Of the 1,697
patients with measurable BMI, more than 50% were
classified as obese (BMI >30.00) and another 30% as
overweight (BMI 25.00–29.99) (Figure 3).
60.0
50.0
40.0
30.0
20.0
10.0
0.0
BMI
% Patients
FIGURE 2: AGE dISTRIbUTION OF dIAbETIC PATIENTS (N=4390)
30.0
25.0
20.0
15.0
10.0
5.0
0.0
0.1% 0.7%
4.1%
6.6%
15.3%
25.5%
23.4%
16.1%
8.2%
0-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90
Age Range
% Patients
The findings of the study utilizing EMR data identify diabetes as the
primary consumer of GP resource among chronic conditions in Canada. “ ”

TAbLE 2: dIAbETES TREATMENT LANdSCAPE
Type
Anti-diabetic
Class
Metformin
DPP-IV Inhibitor
Sulphonylurea
Human insulins and analogues
Other anti-diabetics
Total treated patients
No. of Patients
1514
624
619
212
135
2094
% Patients
72.3%
29.8%
29.6%
10.1%
6.4%
100.0%
Note: Patients treated with multiple product classes would be counted multiple times,
once within each row corresponding to each product class prescribed
TAbLE 3: TOP dIAbETES CONCOMITANCES
Indication
Anti-hyperlipidemia
Cardiovascular
Gastrointestinal
Cardiovascular
Cardiovascular
Cardiovascular
Cardiovascular
FIGURE 4: dISTRIbUTION OF dIAbETIC PATIENTS bY HbA1C ANd FASTING GLUCOSE LEVEL
Treatment type
Cholesterol & triglyceride
regulating preparations
Ace inhibitors
Antiulcerants
Calcium antagonists
Angiotensin II antagonists
Beta blocker agents
Diuretics
Control level
HbA1c: >= 7% --> Out of control (51%)
Fasting glucose: >6.9 mmol/L --> Out of control (60%)
7-<8
8-<9
9-<10
10-<11
11-<12
12-<13
13-<14
HbA1c (%) & Fasting glucose (mmol/L)
HbA1c Fasting glucose
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
Patient Distribution Between Test Levels (%)
2-<3
3-<4
4-<5
5-<6
6-<7
14-<15
15-<16
16-<17
No. of Patients
17-<18
1500
743
525
478
459
446
413
18-<19
19-<20
% Patients
34.1%
16.9%
11.9%
10.9%
10.4%
10.1%
9.4%
20+
*Refers to a treatment with metformin in
combination with any other anti-diabetic
in the study period
TAbLE 4: PEARSON CHI-SQUAREd TESTS FOR INdEPENdENCE bETWEEN TREATMENT TYPE ANd
CLINICAL OUTCOMES bY FASTING GLUCOSE ANd HbA1C TEST RESULTS
Fasting glucose level
In control
Out of control
Total
p-value
HbA1c
In control
Out of control
Total
p-value
Metformin
213
148
361
<0.001
Metformin
289
134
423
<0.001
Metformin plus other*
89
204
293
120
238
358
Total
302
352
654
Total
409
372
781
TAbLE 5: NON-PARAMETRIC TESTS FOR SIGNIFICANT dIFFERENCE IN OUTCOMES (MEASUREd bY FASTING
GLUCOSE ANd HbA1C TEST RESULTS) ANd VISITS TO A PHYSICIAN
Fasting glucose (mmol/L)
HbA1c (%)
Visits
Metformin
7.08
6.88
2.46
8.59
7.96
3.42
p-value
<0.001
<0.001
<0.001

IMPLICATIONS FOR FUTURE INTERVENTIONS
It has been estimated that by 2020 around 10.8% of the
Canadian population will be diagnosed with diabetes, a
57% increase over a 10-year period. In addition, 22.6% of
the population will be classified as pre-diabetic and at risk
of developing diabetes in the future.5 This could
significantly increase the financial burden to Canadian
healthcare; direct medical costs are projected to reach
CN$3.8 billion by 2020 (37% growth since 2010), with
about 5% attributed to GP and specialist visits.5
The findings of the study utilizing EMR data identify
diabetes as the primary consumer of GP resource among
chronic conditions in Canada. With 80% of diabetic
patients classified as being either overweight or obese
there is a clear need for weight management programs
and lifestyle counseling.
Many diabetics are also often treated for co-morbidities
with antihypertensive, gastrointestinal or hyperlipidemia
medications. This is indicative of a more complex patient,
leading to greater demands on a primary care physician
in managing these interrelated conditions.
Despite the availability of multiple treatment choices,
more than half of the diabetic patients in the study cohort
failed to achieve control of their most recent HbA1c
levels. Although the study was not designed to evaluate
the drivers of diabetes control, further investigation into
the real-world effectiveness of various therapies is
encouraged. The results could potentially inform
treatment choices, resulting in a more efficient allocation
of resources.
A further observation from the study is that treatment
complexity, as indicated by a drug regimen including
metformin plus other, is associated with poorer
HbA1c/glucose-level control and an increased demand
for physician time. Thus, patients who were unable to
achieve target control and required more complex
treatment regimens consumed a higher number of
primary care visits. This implies that maintaining better
control of patients during earlier treatment phases can
reduce the additional resource required for more
advanced diabetes care.
Finally, the study findings point to four key areas with
high potential impact for intervention to improve the
real-world management of diabetes in primary care
1. Controlling weight
2. Efficiently managing the challenges of treating a
patient for multiple conditions
3. Evaluating and identifying the most appropriate and
effective medications per patient
4. Achieving and maintaining effective early control
of diabetes.
The study findings point to four key areas with high potential impact to
improve the management of diabetes in primary care. “ ”
1 OECD Health Statistics 2014 : How does Canada compare? Available at: http://www.oecd.org/els/health-systems/Briefing-Note-CANADA-2014.pdf.
Accessed 6 October, 2014
2 Statistics Canada, Community Health Survey 2012. Available at http://www.statcan.gc.ca/pub/82-625-x/2013001/article/11832-eng.htm.
Accessed 6 October, 2014
3 2013 National Physician Survey. The College of Family Physicians of Canada, Canadian Medical Association, The Royal College of Physicians and
Surgeons of Canada. Available at: http://nationalphysiciansurvey.ca/wp-content/uploads/2013/10/2013-National-ENr.pdf. Accessed 6 October, 2014
4 Statistics Canada. Data for 2013. Available at: http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/health53a-eng.htm.
Accessed 6 October 2014
5 Canadian Diabetes Association, Diabetes Québec, 2011. Diabetes: Canada at the tipping point. Charting a new path. Available at:
http://www.diabetes.ca/CDA/media/documents/publications-and-newsletters/advocacy-reports/canada-at-the-tipping-point-english.pdf.
Accessed 6 October 2014

InsIghts SCIENTIFIC-COMMERCIAL RWE SUPPORT
The authors
Finding the true potential of
RWE through scientific-commercial
collaboration
A recent report from IMS Health demonstrates the value that
real-world evidence delivers throughout the pharmaceutical lifecycle
and proposes the more active engagement of commercial teams in
RWE – both in terms of leadership and consumption. This article
summarizes key highlights of that research and presents a framework
for increasing scientific-commercial collaboration in support of RWE.
Marla Kessler, MBA
is Vice President, IMS Consulting Group
Mkessler@imscg.com
Amanda McDonell, MSC
is Senior Consultant,
RWE Solutions & HEOR, IMS Health
Amcdonell@uk.imshealth.com
Ben Hughes, PHD, MBA, MRES, MSC
is Vice President, RWE Solutions,
IMS Health
Bhughes@uk.imshealth.com

Realizing a US$1 billion opportunity through
scientific-commercial collaboration
STEPPING UP TO UNTAPPED RWE POTENTIAL
The IMS Health report1 shows how a few leading
companies pursue RWE as a capability, implementing
RWE platforms that move beyond narrow, study-based
approaches to create sustained value across the product
lifecycle and disease franchises. By following this
approach, a top-10 pharmaco could derive US$1 billion in
value from RWE.
For commercial teams the expanding applications of RWE
come at just the right time, when their stakeholders are
demanding ever more support of a product’s value
proposition just as they and others are producing
evidence of its performance in real-life settings.
In parallel, commercial teams appreciate the
shortcomings of traditional approaches to gaining market
insights but feel they lack ready alternatives. Primary
market research is inherently limited in sample size and
depth of insight, as well as being time intensive. It can
also be inaccurate and thus an inconsistent indicator of
actual behavior. There is a growing need for more time-efficient,
fact-based research.
FOUR GOLDEN PRINCIPLES FOR TRANSFORMATION
Leading companies have recognized these challenges
and taken steps to address them. Their experiences
suggest Four Golden Principles of using RWE to transform
performance, with direct implications for commercial teams.
1. RWE capabilities converge in a platform
Leaders approach platform investments in information,
technology and analytics tools with a plan to support a
range of uses – both scientific and commercial. In these
companies, commercial teams can respond rapidly to
queries about product use and evolving treatment
paradigms rather than having to wait a year to answer the
most fundamental questions.
Leaders think carefully about the platform capabilities
they should buy versus build, and how best to balance
the benefits of centralization (economies of skill) with the
benefits of embedding capabilities within the business
unit (responsiveness to business needs) (Figure 1).
FIGURE 1: CAPAbILITIES LAYER IN AN RWE PLATFORM
RWE capabilities stack
Channels for
dissemination & engagement
CoEs for scientic
commercial analytics
Technology-enabled
tools analytics
Information, networks
data linkage
Business specic setup/build
Partially consolidated capabilities/build
Consolidated capabilities/buy
The necessary layers of capabilities are
• Information, networks and data linkage
Increasingly, technology is enabling managed access
to new information with consent. Leaders develop
relationships with healthcare stakeholders to access
specific data sources relevant to their research needs.
They are able to link datasets, comply with privacy
laws, use technologies that anonymize data at source,
or integrate routine databases with traditional
prospective data. The result is a rich end-to-end view
of patient journeys.
• technology-enabled tools and analytics
Leaders provide users with direct access to data insights
through user-friendly interfaces. Pre-defined, validated
queries under scientific leadership facilitate simple
requests. This flexibility, coupled with high-performance
architecture, reduces time to insight. It
does not replace experienced scientific and statistical
staff, but rather ensures their focus on value-added
instead of routine tasks.
5%brand growth via RWE-enabled marketing
20% launch improvement via patient pool segmentation
3-month acceleration of market access submissions
25-90%cost savings versus primary research
INCLUDING
$1bn

FIGURE 2: PLATFORM dEPLOYMENT TO FUNCTIONS
RD HEOR Medical Safety Market Access Commercial
t Translational research
t Drug pathways
t Target population/
product prole
t Trial simulation/
recruitment
t Pragmatic clinical
trials (pRCTs)
t HEOR productivity
(speed quality)
t Local burden of
illness/disease/costs
Analytics CoE Analytics CoE Analytics CoE Analytics CoE
Data discovery interrogation tools
t Drug utilization/
monitoring
t Risk management
t AE/signal detection
t Rapid FDA/EMA
responses
t Speed to market
(dossier, CED1)
t New pricing mechanisms
t Formulary simulation
t Ongoing value
dierentiation
1 CED: Coverage with
Evidence Development
Technology-enabled tools analytics
Information, networks data linkage
RWE-enabled insights also have potential to accelerate drug development (eg, by improving target selection) which has not been accounted for in this assessment.
• Centers of Excellence (CoEs) for scientific and
commercial analytics
Leaders standardize analytics across markets and data
sources, pooling analysts in a flexible and scalable
service capacity. The continued tendency to manage
scientific and commercial CoEs separately allows
economies of skill where possible but also the
development of deep analytical methods specific to a
therapeutic area (TA) or function.
t RWE-enabled marketing
(eg, undertreated)
t Launch/promotion
planning via
physician-patient
segmentation
t Forecasting
t Engagement services
(eg, adherence)
Insights reporting tools
• Channels for dissemination and engagement
Leaders formalize the use of RWE across global and
local channels to engage stakeholders. This ranges
from global branding programs promoting the overall
credibility of RWE platforms to locally deployed
initiatives for improving RWE capabilities within
medical and pricing market access teams.
Internally, on-demand RWE insights are being
embedded into operational processes across functions.
Thus, the broader organization – including scientific and
commercial functions - can benefit from RWE-enabled
insights tailored to their research interests or operational
needs, as illustrated in Figure 2.
2. narrow precedes broad
Leaders focus on select TAs and markets to ensure their
investments generate differential value. Commercial
teams are often responsible for the overall franchise
performance, best positioning them to understand evidence
needs and priorities.
Companies need to funnel their investment into a
‘must-win’ TA. In our experience, they can only be
distinctive in areas of internal expertise and
products/treatments that give them credibility and
real-world experience with stakeholders. Many emerging
leaders have elected to use RWE in one or two TAs where
there is a strong pipeline and in-market portfolio, and
within mission-critical markets (to include the US and up
to three to five additional markets worldwide).
Even today, no one has full RWE-platform capabilities
across multiple TAs and geographies. However, companies
have had successes in single TAs or with single market
FIGURE 3: AdVANCEd PLATFORM STRATEGIES bY THERAPEUTIC AREA
ANd GEOGRAPHICAL SCOPE
B A A
TA
Multi C C
D
J
TA Single D
H G
Company Evolution US Multi-market
X
Therapy area (TA) scope
Market coverage
Target platform scope
(ongoing build)
Current platform
scope
PAGE 22 IMS HEALTH REAL-WORLD EVIDENCE SOLUTIONS HEOR

approaches that they have expanded over time, as shown
by the migration of individual platforms in Figure 3.
Many will debate this view, given the desire to drive
distinctive capabilities simultaneously in all key TAs,
markets and functions. In reality, it takes several years to
develop the necessary capabilities and deliver value,
which is easier to do when those involved are aligned by
common data and/or challenges, often defined by TA.
Companies outlining a transformation agenda must set
the right expectations. There is no silver bullet; success
requires a multi-year effort of continuous improvement.
3. Commercial leads the charge
HEOR and other scientific colleagues are sometimes
critical of commercial-driven RWE, as the speed to insight
is contrary to their experience of time-intensive study
design and implementation. Yet platform-based RWE
capabilities will help them deliver more and better
research publications with greater scientific and market
impact. Commercial teams must champion the overall
platform to broaden RWE’s application and value for
many reasons – including their unique ability to secure
resources – while HEOR continues to lead the
development and implementation of scientifically
rigorous studies.
The need for commercial to take the lead in this
traditionally scientific domain is not immediately obvious.
However, leaders realize that scientific can be the data
custodian and user of RWE for protocol-driven studies
while commercial can be given appropriate access to
drive strategic decisions. Strong governance, allowing
nominated individuals outside scientific access to data
insights, enables scale in RWE investments.
The largest immediate financial value of RWE is in
supporting about-to-launch and launched products,
areas where commercial drives decision making. Many
decisions related to labeling and identifying target
patients, contracting and pricing strategies, and launch
planning are transformed by RWE, requiring commercial
to be close to RWE strategy. Ultimately, only franchise
leaders can really champion the longer-term investment
in their patients and key markets.
How can commercial initiate its leadership role in a
pragmatic way? More product teams are now sharing
their priorities across functions and mapping their current
and pending evidence plans against them. One company
reoriented several expensive prospective studies to build
a platform capability linking key information sets for
required insights. Thus, longer-term evidence planning
and commercial’s ability to remove organizational barriers
is an emerging vehicle for RWE leadership.
4. speed is a goal
Leaders seek speed to insight and can perform end-to-end
scientific studies in weeks. In their vision of on-demand
insights, quality and speed are harmonious, not
trade-offs. With better, timelier information, commercial
teams can become more nimble and work more effectively
with their customers.
Platform-based RWE capabilities challenge the paradigm
that robust, scientific-led insights require significant time.
With existing data agreements in place and pre-defined
analytics established, analyses can start almost immediately.
In companies where RWE delivery teams have a customer
service mindset (at least three to our knowledge), full
scientific studies using platform-enabled analytics have
been completed in less than a month, rather than up to a year.
FIGURE 4: VALUE CAPTURE FROM RWE ACROSS LIFECYCLE FOR A TOP-10 PHARMACO
Development Launch In-market
Initial pricing
market access*
US$100m
Launch planning
tracking
US$150m
Productivity cost savings
US$100m
Clinical development*
US$100-200m
Safety value
demonstration
US$200-600m
Commercial
US$200-300m
* Selected operational opportunities only; excludes increased RD pipeline throughput and better pricing
spend eectiveness

Insights from RWE can provide commercial teams with
feedback on market changes and the impact of their
actions within weeks. Leaders realize such speed only
matters if there is willingness to act on these insights
promptly. This could mean changing sales call plans,
reprioritizing physician targets, altering or dropping
promotional plans and even engaging with payers more
frequently or differently. RWE leaders make this more real-time
information available, adopt more dynamic
marketing plans, and empower key account managers
and others to leverage the new knowledge.
SOURCES OF THE US$1 BILLION RWE
OPPORTUNITY
The experience of companies living the Four Golden
Principles demonstrates the significant value RWE can
deliver at different stages of the pharmaceutical lifecycle.
Our research identified six main areas of value capture:
clinical development; initial pricing market access;
launch planning tracking; safety value demonstration;
commercial spend effectiveness; and overall productivity
cost savings. As shown in Figure 4, most of the value is
likely to come after product launch.
Examples of impact
t 5% brand growth via RWE-enabled marketing
t 20-50% improved promotion via physician–patient segments
t Better forecasting via disease progression models
t Formulary improvement from Tier-3 to -2
t Avoidance of label changes
t 2-week responses to FDA/3rd party journal publications
t 20% launch improvement via patient pool segmentation
t Rapid adjustment of messaging/resource allocation at launch
t 3-month acceleration of market access submissions
t Payment by use/indication, more eective price negotiations
(not quantied)
t Conditional access via coverage with evidence development
t 25-90% cost saving versus primary market research
t Doubling of impact factor of publications1
t 30% improvement in trial enrolment
t Reduction in strategic trial design aws
t Better product prole design (not quantied)
FIGURE 5: CASE STUdIES OF RWE IMPACT ACROSS OPPORTUNITY AREAS
Commercial spend eectiveness
US$200-300m
Safety value demonstration
US$100 m
(upside)
US$100-500m
(downside avoidance)
Launch planning tracking
US$150m
Initial pricing market access*
US$100m
Productivity cost savings
US$100m
Clinical development*
US$100-200m
Traditional focus Leaders’ additional focus
1 Hruby GW, et al. J Am Med Inform Assoc, 2013; 20: 563-567
* Selected operational opportunities only; excludes increased RD pipeline throughput and better pricing

The opportunity for RWE to add value is substantial but commercial needs
to step up and take accountability for implementing “ RWE capabilities. ”
In companies without RWE platform capabilities, the roles
of scientific and commercial are compartmentalized:
scientific teams are asked for studies to support specific
ad hoc arguments without long-term strategic input,
while commercial teams face increasing scrutiny of their
products but are often unarmed with the evidence to
defend them. Leaders have built RWE capabilities that
span both functions, enabling immediate and strategic
evidence generation.
Diving deeper into the buckets of RWE value, the research
sought to provide more information about the value drivers
and financial magnitude. Case studies enabled a richer
understanding. While RWE can help increase revenues, it
can also avoid downside risk as well as unnecessary costs.
Of particular interest were areas where leaders think
beyond traditional RWE applications (Figure 5).
IMPLICATIONS FOR SCIENTIFIC AND
COMMERCIAL COLLABORATION
The involvement of commercial does not diminish the
role of HEOR and other scientific and medical teams.
Rather, it should be complementary, serving to focus on
removing roadblocks to broader commitment for RWE
and increasing its overall application to demonstrate the
value of a franchise.
At the same time, scientific teams should champion the
treatment of RWE as a capability instead of a series of
studies to increase their overall effectiveness and
productivity. With the right RWE information and tools,
these teams can focus on the highest-value analytics
rather than lower value activities such as ad hoc data
sourcing and protocol development. Just as commercial
teams will need to generate, analyze and apply insights
more frequently, scientific colleagues will have to
integrate more seamlessly into the faster pace of decision
making enabled by systematic application of RWE.
Best practice example
A leading company provides an intriguing lens into best
practice. It began its RWE journey by creating an integrated
evidence platform in response to value and safety
demonstration challenges. When the FDA questioned the
appropriate use of its blockbuster oncology product, up to
US$500m of revenue was placed at risk due to potential
label changes. By developing the broadest RWE platform at
the time, the company enabled a variety of insights to
inform discussions with a multitude of stakeholders,
successfully responding to the FDA challenge.
Having experienced the power of RWE insights, the
company continued to invest beyond value and safety
demonstration. Commercial leaders acquainted with RWE
capabilities started to systematically lever detailed patient
pathways to understand product use, identify patterns of
under-diagnosis and under-treatment, and shape highly
targeted marketing campaigns. These campaigns nearly
doubled sales growth. Over time, RWE became the
company’s currency and competitive advantage for
engaging health systems, with granular forecasting and
disease progression models levered by a series of medical
center partners for their own service planning. For the
first time in the industry it effectively developed a closed-loop
system, using insights to engage and improve
patient pathways.
SIGNIFICANT ADDED VALUE
The opportunity for RWE to add value is thus substantial,
especially for in-market products. As the principal
organizational owners of these products, commercial
needs to step up and take accountability for implementing
RWE capabilities. Working collaboratively and cross-functionally
with scientific will ensure that investment in
RWE spans the interests of both respective functions.
1 Hughes B, Kessler M, McDonell A. Breaking New Ground with RWE: How Some Pharmacos are Poised to Realize a $1 Billion Opportunity.
A White Paper from IMS Health. August 2014.

InsIghts PREDICTIVE MODELING
The author
Improving outcomes
through predictive modeling
Predictive modeling involves assigning values to new or unseen
data. With growing promise across a wide range of fields, it is
increasingly being applied in various healthcare settings both to
reduce costs and drive quality improvements. However, while its
potential contribution is substantial, even exciting, applications
involving its use are not widespread and demonstrable evidence
on effectiveness is limited.
John Rigg, PHD
is director Predictive Analytics, RWE Solutions, IMS Health
John.rigg@uk.imshealth.com

Potential and challenges for developing successful models
Referencing real-world cases studies that have emerged,
this article discusses ways in which predictive modeling is
currently being used, considers the potential for
innovations from machine learning to extend its value
and accuracy, and highlights the challenges to
developing a successful predictive modeling application.
DIVERSE APPLICATIONS IN PRIMARY CARE
The scope of predictive modeling applications is wide
ranging, with models used to stratify risk both at a
population and patient level. At the population level, risk
stratification is routinely employed by payers/
commissioners to understand resource need and help
shape service delivery. Typically, this involves estimates of
disease prevalence, including age-demographic
adjustments. These models will likely become
increasingly advanced, helping to quantify the depth of
clinical need and define the type and scope of service.
At patient level, the applications principally focus on
identifying patients at high risk of particular events such
as unplanned hospital (re)admission, or the onset of a
chronic disease such as diabetes. High-risk patients are
then targeted with an intervention aimed at mitigating
the event.
1. Reducing hospitalizations
Identifying patients at greatest risk of unplanned hospital
readmission is currently by far the most widespread use
of predictive modeling in primary care.1 Readmissions
within thirty days of discharge are common, costly and
hazardous. Moreover, many readmissions are considered
avoidable.2 Reducing them is thus a major focus in
virtually all healthcare systems.3,4,5 It has certainly
captivated policymakers as a goal that can both improve
quality and reduce healthcare costs, seen in the US, for
example, with powerful incentives in the Patient
Protection and Affordable Care Act penalizing hospitals
that have higher-than-expected readmission rates.5
Heart failure has been a particular target, being one of the
most common reasons for hospitalization in the
developed world and accounting for the highest thirty-day
readmission rates.3
Parkland Health Hospital System: Informing CHF and
expanded disease areas
One example of a successful program is Parkland Health
Hospital System in Dallas, Texas. In 2009, Parkland began
analyzing electronic medical records (EMR) with the aim
of using predictive modeling to identify patients at high
risk of hospital readmission. The initial focus was on
congestive heart failure (CHF). Today, case managers and
other frontline providers receive details of high-risk
patients on a near real-time basis, information that is used
to prioritize workflow and allocate scarce resources to
support those most in need. Interventions are both
hospital- and community-based.6
Evaluation of the program identified a reduction in thirty-day
all-cause readmission rates from 26.2% to 21.2%.7 As
observed in an editorial by McAlister, “This effect size was
achieved even though the programme was only offered to
approximately a quarter of discharged patients, was only
deployed on weekdays (weekend discharges actually exhibit
the highest rate of readmissions) and despite the fact that
only a minority of readmissions may be truly preventable.”3
Given the observed fall in readmissions and costs for CHF
patients at Parkland, the program has been expanded to
patients with diabetes, acute myocardial infarction and
pneumonia. Preliminary data suggests similar success
with readmission rates in these conditions.6
NorthShore University HealthSystem: Supporting
hospital and primary care
Positive results have also been achieved through the use
of an effective predictive model at NorthShore University
HealthSystem in Chicago. Reports stratifying inpatients by
high, medium or low risk of readmission in 30 days are
provided to health system hospitalists on a daily basis and
scores noted as a value in every inpatient EMR.
26% 21% reduction in
re-admission rates

InsIghts PREDICTIVE MODELING
These have proved so useful that reports are also now
sent to the system’s primary care physicians listing their
patients with a high risk of readmission. The program has
seen a reduction in readmissions from 35% to 28%
among high-risk patients.8
Despite these successes, recent reviews reveal little
systematic evidence on what works in terms of community-based
alternatives to hospital admissions.4,5,9 However, there
is evidence to suggest some impact of particular initiatives
in targeted populations, such as education with self-management
in asthma, and specialist heart failure
interventions. Moreover, certain types of interventions, such
as post-discharge telephone calls, have also been identified
as effective.5 Beyond that, most other interventions appear
to have no effect in reducing emergency admissions in a
wide range of patients. There is a clear need to better
understand what works and for whom.
Interventions to reduce emergency admissions take place
within a complex environment where the nature and
structure of existing care services, individual professional
attitudes, patient and family preferences, and general
attitudes to risk management can affect their
implementation. While some interventions fail to reduce
admissions, they may have other beneficial effects, such as
reducing length of stay or improving the experience of care.4
2. Mitigating risk
NorthShore University HealthSystem: Predictive
modeling in hypertension
NorthShore is a pioneer in the use of various risk stratification
applications. One success story involves predictive modeling
to identify undiagnosed patients with hypertension (HTN).10
Although many patients with HTN are actively managed,
the condition is often overlooked. The risk stratification is
based on three screening algorithms, developed using
established HTN diagnosis guidelines, to identify patients
with consistently elevated blood pressure readings and
exclude those with only intermittent elevations. Patients
are considered at risk for undiagnosed HTN if they meet
the criteria of any of the three algorithms. The screening
tool was built using outpatient data from the NorthShore
data warehouse and the model has an accuracy rate
(Predictive Positive Value) of approximately 50%.
Veterans Health Administration (VHA): Population-wide
risk scores
The VHA has also invested heavily in risk stratification
applications, covering its entire primary care population.11
This includes models that output a patient’s percentile
scores associated with risk of hospitalization and
mortality. Updated weekly to reflect changes in individual
clinical status, the models rely on six data domains pulled
from the VHA’s extensive data platform: demographics;
diagnoses (inpatient and outpatient); vital signs;
medications; laboratory results; and prior use of health
services. Risk scores can be accessed on-line by each care
team, alongside other information such as active
diagnoses, recent visits to primary care and enrollment in
care management programs. They can also be rendered
as high-resolution geospatial maps to assist managers
with program planning and determining where new sites
for service delivery might be located.
While it is too early to determine whether the risk scores
help improve outcomes, the VHA suggests that based on
the frequency of access, healthcare providers are finding
them worthwhile. In addition, testimonials from clinicians
and care managers indicate that the scores are more
useful than clinical reminders, since each score takes into
account the patient’s unique needs and allows staff
members to focus on what is most likely to improve
future outcomes on an individual basis.
The VHA has also implemented a system for early
detection and management of chronic kidney disease,
including risk-based clinical EMR reminders which play an
important part in the effectiveness of the program.12
DEVELOPING AND APPLYING A PREDICTIVE MODEL
An outline of the main stages associated with developing,
validating and operationalizing a typical predicting
modeling application is shown in Figure 1 (page 30) and
described below.
1. Cohort creation from raw input data
In the initial stage, patient cohorts are created from the
input data. There are generally two: one cohort for model
development, the other for validation. A common
practice is to randomly split the data approximately two-thirds
and one-third between development and
validation cohorts respectively.
2. Algorithm development
In the second stage, the predictive model is estimated on
the development sample using an appropriate statistical
method such as regression analysis. The model is then used
to identify at-risk patient profiles and key predictors/
characteristics are described and clinically verified.
3. Algorithm validation
It is important that model development and validation
are carried out on separate data. This enables
independent assessment of its performance, ensuring it is
not ‘overfitting’ (where a model may accurately describe
data upon which it is estimated but poorly describe new
or unseen data). Thus, the third stage involves detailed
evaluation of model performance using a variety of
metrics. In the case of hospital readmission modeling,
for example, the metrics may include the number of

IMS Health Real World Evidence Access Point

IMS Health Real World Evidence Access Point

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Was ist angesagt?

Was ist angesagt? (20)

Andere mochten auch

Andere mochten auch (15)

Ähnlich wie IMS Health Real World Evidence Access Point

Ähnlich wie IMS Health Real World Evidence Access Point (20)

Mehr von QuintilesIMS Asia Pacific

Mehr von QuintilesIMS Asia Pacific (20)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

IMS Health Real World Evidence Access Point