Measuring Cost and Efficiency Effects of the PCMH

1
Measuring Cost and Efficiency Effects of the PCMH

• PCMH Evaluators Collaborative
• Efficiency Workgroup
• June 30, 2009

2
What Are We Trying to Approximate Conceptually?

• CostsSpending by health plans (measured relative
  to comparison practices) is of interest even
  without quality
• CostsPractice costs to implement the PCMH for
  the purpose of calculating ROI (incremental net
  revenues over incremental net costs of operating
  as a PCMH)
• Efficiencycost (total or payer perspective) per
  unit of outcome will answer the value question
• Efficiency impact of the PCMH is the net benefit
  of the intervention incremental benefits less
  incremental costs
• In practice we are focusing on cost and
  utilization metrics that help flesh out these
  concepts substitution of equal quality, lower
  cost care (e.g., generic substitution, office
  visits for ED, group visits, email visits)

3
Principles of Measurement

• A logical connection must exist between the
  changes in practice infrastructure, process, and
  financial incentives embodied in the specific
  pilot and each cost/efficiency measure or proxy
  to be assessed.
• To the extent possible, cost/efficiency measures
  and proxies should be evidence-based and grounded
  in clinical appropriateness.
• Pilot sites will not fully implement all aspects
  of the PCMH at once. For each practice and
  demonstration, a set of most appropriate
  measures can be selected to reflect the sequence
  of PCMH elements implemented.
• Changes in patient care and health status that
  might theoretically generate cost savings and
  efficiency improvements will take time to accrue
  even after infrastructure and process
  improvements have been implemented. Reasonable
  time horizons for such changes to occur should be
  built into evaluation designs.
• Evaluations should identify intermediate outcomes
  that demonstrate the projects are successfully
  moving on the path towards improved outcomes and
  reduced spending

4
Logic Model Inputs Measurable Elements of the
Interventions

• Pay for performance targeting quality and cost
  measures (varies by pilot)
• Enhanced access expanded availability, modes of
  communication, language etc.
• Informed Care Management Disease registries,
  patient education, care management for high-risk
  populations, e-prescribing
• Coordination of Care referral and test tracking,
  management of care transitions

5
Pay for Performance

• Recent studies show modest evidence of impact,
  one (?) study with cost savings
• Pay for performance, if part of the pilot, could
  target cost or utilization measures directly (CO
  is proposing this)
• Pay for performance targeting improved care for
  chronic illness may save money (e.g., RIPA)
• Essentially, (1) if cost/utilization measures are
  in P4P and quality does not decline then
  efficiency may increase, (2) if over use measures
  are in P4P then efficiency may increase, or (3)
  if chronic care quality measures are in P4P then
  efficiency may increase with or without cost
  savings

6
Access

• Evidence of language concordance between
  attending physician and patient in inpatient
  settings leads to slightly lower costs and lowers
  return visits to the emergency department
  (Jacobs, Sadowski et al. 2007), (Hampers and
  McNulty 2002)
• Management of chronic disease in comprehensive,
  coordinated way in outpatient setting can lead to
  fewer hospitalizations in Medicaid patients
  (Billings and Mijanovich 2007) findings in
  children that more continuous care with a primary
  care provider leads to less emergency department
  visits and hospitalizations (Christakis, Mell et
  al. 2001)
• (Billings and Mijanovich 2007), (Christakis, Mell
  et al. 2001)

7
Care Management

• Studies of the chronic care model for individual
  chronic conditions have found reductions in
  admissions and readmissions. See Bodenheimer,
  Wagner Grumbach 2002 for comprehensive
  review.not so sure of the quality of this
  evidence
• Studies of the chronic care model for asthma have
  found reductions in ED visits. See Bodenheimer,
  Wagner Grumbach 2002 for comprehensive review.

8
Care Coordination

• Geisinger medical home model with practice-based
  care managers and focus on care transitions for
  Medicare patients showed 20 reduction in
  all-cause readmissions (Paulus, Davis, and Steele
  2008)
• Coleman et al. 2006 find reductions in
  readmissions with independent coaches that serve
  patients immediately post-discharge

9
Logic Model Outputs Measurable Effects that
Relate to Costs and Efficiency

• Reasonable degree of evidence/support
• Admissions (ambulatory-care sensitive)
• Readmissions
• ED visits (ambulatory-care sensitive)
• No evidence or cross-sectional only
• Lab tests and imaging (speculative)
• Referrals (based on cross-sectional evidence
  only)
• Generic prescribing

10
Summing Up Total Cost and Efficiency Measures

• Industry standard for efficiency is cost per
  episode with risk adjustment
• Disease-specific cost per episode makes sense in
  light of focus on prevalent conditions
• Total pmpm allows examination of all the cost
  increases, decreases together (takes account of
  offsetting effects)
• Eventually need to combine costs with quality to
  really measure overall efficiency impact
  although reductions in over use or misuse are
  clear efficiency gains

11
Leading Indicators?

• Cost savings/efficiency improvements may take
  time to accrue e.g., several years to prevent
  hospitalization in patients with diabetes
• Discussed what leading indicators would be
  improved intermediate outcomes (quality group)?
• Others, like reduced time to first outpatient
  visit after discharge, seemed appealing but only
  if the evidence was strong (it wasnt)

12
Practice and Plan Implementation Costs

• We recognize that implementation and
  transformation costs should be counted in the
  full assessment of the policy
• Incremental outlays of resources for the pilot?
  Fixed and variable costs of PCMH
• Unclear whether evaluations are collecting these
  what are likely to be key costs that are feasible
  to estimate that we could ask for in site visits,
  phone calls?