saftinet overview for edrc

SAFTINet: Scalable Architecture for Federated Translational Inquiries Network

An OverviewMarion R. Sills, MD, MPH

SAFTINet CU Website

SAFTINet Overview• Why: To support both research and quality improvement

• What: A practice-based research network and distributed data network combining clinical and claims data collected during routine care

• Who: A network of stakeholders interested in improving health care delivery and outcomes of underserved populations




What is the value SAFTINet is designed to add?

Impetus: U.S. Healthcare Crisis• Broken healthcare system• Health expenditures account

for 17.6% of US GDP• Highest cost does not =

highest value • System not patient-centered

Stakeholder Engagement in Solutions to Crisis

In order to have SAFTINet help stakeholders, stakeholders are engaged in designing and guiding SAFTINet.

Value to Providers

• Majority of medical treatments lack valid evidence of effectiveness• Especially with regard to• Specific patient groups

• Underserved populations• Vulnerable populations

• Specific settings• Primary care, especially in safety net settings• Real world settings (i.e., outside a well-controlled

trial)• Outcomes that matter to patients

How generalizable

is the evidence to my clinic patients?

Value to Patients

• Most of the outcomes assessed are related to an intervention’s short-term effectiveness, as driven by• What researchers are trained to

measure• What funders/industry fund

• Fewer focus on patient-centered outcomes• Quality-of-life• Long-term outcomes• Often we don’t know what matters most

to patients or even how to find out

I care. Your health is

important to me.

How do you know what outcomes matter to

me?

Value to Researchers• Barriers to researching highest

priority conditions and patients • Limited funding• Challenges to study neediest

patients in their usual settings• Vulnerable populations • Data quality problems as safety net providers are

under-resourced to• Perform ongoing data collection• Integrate data collection into their workflow• Apply diligence to data quality• Electronic data not always present or of consistent

quality• True of any non-research setting

• Poor data quality = invalid research

How can I access high priority

patients and their data? How do I

obtain good quality data?

Value to Policy-Makers

• Barriers to decision-making for highest priority conditions and patients • Data on these patients• Scattered, fragmented• Varied quality• Not timely

How do we measure how

well healthcare policy decisions

are working?

Policy-Makers Take Action• 2003: MMA Section 1013 authorizes AHRQ to conduct

and support outcomes research • 2009: ARRA provides $1.1 billion to NIH/HHS/AHRQ to

stimulate the fix • funds SAFTINet, 7 other PBRN Centers for Primary Care

Research• 2010: Patient Protection and Affordable Care Act (ACA)• Funded AHRQ to perform comparative effectiveness

research (CER) • Created the Patient Centered Outcomes Research Institute

(PCORI)

Policy-Makers Take Action• 2003: MMA Section 1013 authorizes AHRQ to conduct

and support outcomes research • 2009: ARRA provides $1.1 billion to NIH/HHS/AHRQ to

stimulate the fix • funds SAFTINet, 7 other PBRN Centers for Primary Care

Research• 2010: Patient Protection and Affordable Care Act (ACA)• Funded AHRQ to perform comparative effectiveness

research (CER) • Created the Patient Centered Outcomes Research Institute

(PCORI)

All fund CER/PCOR: development of better scientific evidence on the effects of intervention on patient-centered health outcomes.

You Say CER, I Say PCOR

PCOR• Assesses the benefits and harms of

interventions to inform decision making, highlighting comparisons and outcomes that matter to people.

• Incorporates a wide variety of settings and participants to address individual differences.

• Investigates optimizing outcomes while addressing burden to individuals, availability of services and other stakeholder perspectives.

CER• The generation and synthesis of

evidence that compares the benefits and harms of alternative methods to prevent, diagnose, treat and monitor a clinical condition, or to improve the delivery of care.

• Purpose: To improve health outcomes by developing and disseminating evidence-based information to patients, clinicians, and other decision-makers, responding to their expressed needs, about which interventions are most effective for which patients under specific circumstances.

Funding for SAFTINet

• 2009: American Recovery and Reinvestment Act (ARRA) – funded SAFTINet (via AHRQ)• Project requirements• Develop a distributed data network that collects and links data from

multiple and different healthcare delivery settings• Clinical data from electronic health records (EHR) • Medicaid claims data

• Demonstrate capabilities for conducting methodologically rigorous Comparative Effectiveness Research (CER)• Capability for enhanced data collection (e.g., patient-reported outcomes)

• Other priority areas we targeted• Underserved populations• Stakeholder engagement• Measuring healthcare delivery system characteristics

Who (and Where)?

Practices by Organization

SAFTINet Goals and Progress

• Stakeholder Engagement• Research and QI• Infrastructure

Stakeholder Engagement• SAFTINet Objective• Establish and maintain a broad, safety-net focused, research

partnership and learning community to govern relationships, establish priorities, provide data quality oversight, and evaluate the purpose and value of the community’s effort

• Goal • Create a trusted, valued multi-state community of safety net

stakeholders and researchers to lead and participate in a learning community to address evidence-gaps relevant to the safety net populations – with special emphasis upon those populations served by Medicaid and State Child Health Insurance Program (SCHIP)

Stakeholder Engagement• What we are doing: ongoing engagement of stakeholders in

research (CER/PCOR) and quality improvement, including:• Evidence prioritization, generation, synthesis, and interpretation

and integration• Dissemination and application• Feedback and assessment

Stakeholder Engagement: Example 1• Stakeholders wanted more

standardized collection of patient-reported outcomes

• What mattered to providers, practice administrators and researchers• Asthma control• Medication adherence

• Stakeholders selected measures and implemented them• Asthma Control Test: useful to

providers, widely used• Medication adherence: providers

question validity, not widely used

Stakeholder Engagement: Example 1• Patient-reported outcome measure implementation

Researcher request

Organization 1 Organization 2 Organization 3 Organization 4

Frequency of Administration

At least annually; all visits preferred

Annually At all visits At all visits At hypertension visits (2x/year)

Mode of Administration (by whom, in what setting, with what materials)

No request made

Care coordinators administer by telephone as part of larger patient survey

Administered by medical assistant, responses recorded on paper upon check-in for visit

Administered as part of written pre-appointment survey at check-in

Electronic survey; Provider asks the questions out loud and records the answers in the EHR template

Stakeholder Engagement: Example 2• Stakeholders (providers, practice administrators) wanted a

measure of Medical Home – ness that had better face validity than existing measures

• Iterative process of review and adaptation of existing measures: the Delivery of Coordinated Care Survey (DoCCS)

• Administered at practice-level, self-report

What the DoCCS MeasuresDomain Example GoalsPersonal Clinician & Sustained Partnership

Clearly link patients to a clinician and/or care team so both the patient and provider/care team recognize each other as partners in care.

Personal Clinician Led/ Team-Based Care

Team-based care led by clinician

Coordinated and Integrated Care Link patients with community resources to facilitate referrals and respond to social service needs.

Patient/Family-Centered Care/Support Shared Decision-Making

Assess and respect patient and family values and expressed needs.

Quality Improvement & Safety Establish and monitor metrics to evaluate improvement efforts and outcomes and provide feedback.

Use of Organized Care & Evidence-based Medicine

Use point of care reminders based on clinical guidelines.

Access Provide scheduling options that are patient- and family -centered and accessible to all patients.

Engaged Leadership Provide visible and sustained leadership overall culture change and specific strategies to improve quality and sustain and spread change.

Registries, Performance Reporting and QI Programs

Use of patient tracking registries to monitor and inform clinical interventions for persons with specific health care needs.

Distributions: All DoCCS responses

Worse DoCCS score in practices• With more Latino and

Spanish-preferring patients

• Located in areas of greater population density

Stakeholder Engagement: Example 3• Stakeholders (providers, practice administrators) wanted more

standardized performance reporting across practices • What mattered to providers, administrators and researchers• Able to compare practices on data of comparable quality• Able to use data to meet reporting requirements

• Stakeholders are engaged in designing reports• Vermont practices have used these for both QI and reporting• Other practices: work-in-progress

Stakeholder Engagement: Lessons Learned• Balancing fidelity to protocol with flexibility• Aligning multiple stakeholder perspectives requires

significant compromise and investment of effort from all parties• Estimated 2000 person-hours• Many competing priorities

• Experience matters • Most successful organization had experience with PRO data

collection in the past, and an existing, proven data collection mechanism

Stakeholder Engagement• Focus of ongoing work: sustainability• Stakeholder engagement (focus groups, key informant interviews)

in adapting SAFTINet to best meet the ongoing needs of stakeholders

SAFTINet Research Goals

• Improve health care quality and value• Reduce health disparities• Generate evidence that will inform health policy, health

services delivery, and practice at the point of care• Target population:• Low-income• Chronic conditions• Cared for in public safety net clinics

SAFTINet Research Objective• To develop and enhance four sentinel cohort pairs of patients

distinguished by their health care delivery characteristics, which can support comparative effectiveness research• Asthma (pediatric and adult)• Hypertension• Hypercholesterolemia

DELIVERY SYSTEM FACTORS + PROCESSES OF

CARE + PATIENT FACTORS → OUTCOMES

SAFTINet Research Goal• To improve health care quality and value, and reduce health

disparities, in low income patients with chronic conditions who receive care in public safety net clinics by generating evidence that will inform health policy, health services delivery, and practice at the point of care

DELIVERY SYSTEM FACTORS + PROCESSES OF

CARE + PATIENT FACTORS → OUTCOMES

Research Conceptual Model

Research in Progress• Asthma• Does practice-level concordance with the Patient-Centered

Medical Home model impact asthma outcomes?• Does adoption of the Colorado Asthma Toolkit Program impact

asthma outcomes?• Other projects• Diabetes control• Methods projects

• Data quality methods• Multilevel analytic methods for clustered data

Asthma Toolkit Project

Patient-Centered Medical Home Project

Methods: Hypothesized Causal Structure

Asthma Control (tn)-ACT (good, poor, very poor control)-Exacerbations 1+ of: -ED/hospital -Steroid burst -Az visit cluster -Inhaled beta-agonist

Medical Home (tn)-Personal clinician-Teams-Coordination/integration-Pt centeredness-QI-Organized/EBM-Access-Leadership-Registries/perf rep-Enhanced asthma care

Intermediate (tn)-Attention to asthma -More data on asthma-Provision of care -More access for in-house az tx-Use of spirometry-Patient adherence to tx-Provider guideline concordance-Continuity of care (1o care util)-Smoking status

Practice-level variables (to)-Practice Size/Diversity -# providers (FTE) -# specialties in house-Practice location -MSA size (urban/rural) -Practice structure (FQHC, ACO)

-Practice population -Total unique pts, pt visits -Payer mix -Race/ethnicity overall mix -Total az patients

Patient-level variables (to)-Patient demographics -Age -Race & Ethnicity -SES (family income, size) -Gender (maybe) -Urban/rural

Patient-level variables (to)-Risk for poor asthma control -Prior exacerbations -Prior utilization -Prior ACT scores-Asthma severity -Persistent/Intermittent (meds)

Patient-level variables (t1)-Risk for poor asthma control -comorbidities (allergy, rhinitis/sinusitis, GERD)

SAFTINet Research Lessons Learned• Stakeholder engagement is valuable and resource-intensive• Essential to high-value data• More resource intensive than often projected• Focus on relationships, resources

• Real world data from diverse, non-research settings is messy• Data mapping and cleaning: very resource intensive • Complex data structure: complex modeling methods• Non-RCT trials or observational studies require rigorous methods

to address confounding and bias

Network Infrastructure• SAFTINET Objective• Build, deploy and assess a safety-net focused distributed research

network which combines ambulatory and inpatient clinical data and Medicaid claims and eligibility data for clinical and research purposes

• Goal• Build the technology necessary to support a valued, virtual

organization that securely federates clinical EHR and Medicaid/CHIP data, to promote quality care and provide enhanced data for comparative effectiveness research

Questions?

Distributed Data Networks

• Build on and expand existing electronic health infrastructure • Broad, scalable and sustainable systems• Enable the collection of longitudinal and comprehensive data

across diverse health care settings• Evaluate effectiveness of clinical interventions, processes of

care, and reimbursement initiatives for diverse clinical conditions, persons and settings

Web PortalQueries and

Data TransfersInternetInternal Network

TRIADROSITA

Local Data

Warehouse

Other EHR

Clear Text PHI

LimitedDataset

Grid-enabledDatabase

DMZ

Firewall Firewall

Technology infrastructure: Single Node

Technology Infrastructure: Medicaid

MedicaidData

sFTP

/Clea

r tex

t PHI

sFTP/Clear text PHI

sFTP/PPRL encrypted PHI

Web PortalQueries and

Data Transfers

Internet

Technology infrastructure: Multi-node

Common Data Model (OMOP)

Person

Drug_exposure

Condition_occurrence

Procedure_occurrence

Visit_occurrence

Observation

Procedure_cost

Drug_cost

Observation_period

Payer_plan_period

Provider

Care_site

Organization

Location

• Health Outcomes of Interest

• Drugs of Interest• Interventions

Cohort

Death

Drug_era

Condition_era

Standardized Vocabulary

SAFTINet Data Specifications