190 likes | 198 Views
This study assesses the impact of Oregon's PCPCH program on healthcare use and expenditures over a three-year period. It examines the program's effects on covered services and key service types, as well as overall program outcomes based on duration of PCPCH designation.
E N D
Oregon’s Patient Centered Primary Care Home Model after Three Years: Use and Expenditures Neal Wallace, Ph.D. Shauna Petchel, MPH Portland State University Research Support from the Oregon Health Authority (PI: Sherril Gelmon, DrPH)
Background • The Oregon Legislature established the Patient-Centered Primary Care Home (PCPCH) Program in 2009 to create access to patient-centered, high quality care and reduce costs by supporting practice transformation. • The PCPCH model is Oregon’s characterization of a “medical home,” and is defined by six core attributes each with a number of corresponding standards and measures. • The attributes include: Access to Care; Accountability; Comprehensive Whole-Person Care; Continuity; Coordination and Integration; and Person and Family Centered Care. • The Program is responsible for defining the model, administering the application process to recognize practices that have achieved the model, and providing technical assistance to clinics. • As of December 2015, over 600 practices across the state had been formally recognized by the OHA as a PCPCH.
Research Objective • To assess the impact of Oregon’s Patient Centered Primary Care Home (PCPCH) program on use and expenditures: • For all PCPCH designated clinics through the first three years of the program • Across all covered services and key service types • For the program overall and by PCPCH duration of designation.
Study Design • Generalized difference-in-difference design to accommodate staggered PCPCH implementation: • Compare patients empirically attributed to PCPCH or non-PCPCH primary care providers in each study year • Minimum one year pre-PCPCH data and maximum three years post PCPCH designation • Applied to a two part model to assess probability of use, expenditures per user and per person
Study Data • Four years of Oregon All Payer All Claims (APAC) claims, eligibility and provider data: • October 2010 through September 2014 (first PCPCH designations October 1, 2011) • APAC contains all claims for commercial, self-insured, Medicaid, and Medicare Advantage • Does not include Medicare FFS, VA claims • Excludes substance use and some sensitive condition (e.g. AIDS/HIV) related service claims
Study Population - Individuals • 1,192,435 individuals were initially identified who during a study year had: • Consistent residence in Oregon • Consistent, full year insurance coverage • At least one primary care visit to an Oregon provider • These individuals were empirically attributed to a provider billing unit during each study year based on plurality of visits or last visit (tie) • The final study population consists of 1,128,234 individuals solely (100%) attributed to either PCPCH or non-PCPCH primary care provider billing units (606,881 PCPCH and 599,990 non-PCPCH)
Study Population - Providers • 510 PCPCH practices continuously designated within the study period (on or after October 2011) • PCPCH related claims were initially identified using practice and/or organizational National Provider Identifiers (NPIs). • NPI-based claims were grouped using a “provider crosswalk key”: • Practice billing units reflecting common/linking billing information (e.g. name, address, secondary NPI, etc). • Links practitioners billing under individual NPI but otherwise identifying as part of larger practice group • Practices may have multiple “billing units”
Study Population - Observations • The unit of analysis for the study was a provider billing unit quarter. • Individual use, expenditure and demographic characteristics were aggregated to provider billing unit by quarter • 100,084 observations (7,380 PCPCH/92,704non-PCPCH) representing 510 PCPCH and 8,435 non-PCPCH billing units
Outcome Measures • Main outcome measures are percentage of subjects using service in a quarter, expenditures per service user, and expenditures per subject • The outcome measures are applied to all covered services and eight specific service types: • Primary care office visits and procedures • Specialty office visits and procedures • Outpatient mental health care • Non-therapeutic radiation • Lab • Pharmacy • Emergency Department • Inpatient
Subject Characteristics • Subject characteristics aggregated to the practice billing unit quarter as percentages include: • Gender • Age group (9) • Specific insurance type (16) • Physical and behavioral condition markers (10) including diabetes, COPD/asthma, chronic heart failure, chronic kidney disease, coronary heart disease, cerebrovascular disease, obesity, schizophrenia, affective disorders, and other behavioral health conditions
Analytic Methods • Two-way fixed effects (practice billing unit and quarter) weighted (individuals) OLS regression • Inclusion of all subject characteristics measures • Samples are not matched but >99% support found across PCPCH and non-PCPCH observations • PCPCH effects identified by a binary variable(s): • For all PCPCH related quarterly observations on or after PCPCH designation date, or • PCPCH designated quarters by year of designation (1-3) • Standard errors adjusted for clustering on practice billing unit
Results: Summary • Total expenditures were reduced by 4% overall with progressive reductions to 8% for PCPCHs in the third year of designation: • Progressive decreases in overall expenditure per user coupled with progressive increases in overall service use • Increases in primary care expenditure and use • Reductions in specialty care expenditures and use • No or inconsistent change in lab and radiology • Increases in expenditure and use of pharmacy • Decreases in expenditures for ED and inpatient: • All due to decreases in expenditures per user – use up (ED) or unchanged (IP)
Study Limitations • PCPCH program effects may be Oregon specific (e.g. culture of “transformation”) • PCPCH practices have different characteristics than non-PCPCH and may reflect unique capabilities/motivations of practices that chose to participate • APAC data missing some key services (e.g. substance abuse) and insurance groups (e.g. Medicare FFS) • Observational, natural experiment design may fail to capture “unobservables” that influence outcomes • Current design does not adjust for potential differences in reimbursement rates across insurance types over time • Anticipatory behavior by PCPCHs (changes prior to designation) detected
Conclusions • Oregon’s Patient Centered Primary Care Home Program appears to be incurring expected (or hoped for) system transformative effects – reducing expenditures in a “treatment positive” manner • Findings are consistent with other large state level implementation efforts – e.g. Colorado and Michigan • Apparent progressive effects among PCPCHs over time from designation may be important to understand from a policy, programmatic and evaluation standpoint: • Policy makers may actually have to wait for large(r) results • Progressive effects could be practice maturation (consistent with Oregon’s multi-level designation approach) and/or “dose response” as patients get more exposure to program. • Some effects may not be apparent in short-term or “average” effect evaluations
Questions? Thank You nwallace@pdx.edu Mark O. Hatfield School of Government OHSU/PSU Joint School of Public Health Portland State University