Are Primary Care Services a Substitute or Complement for Specialty and Inpatient Services

1
Are Primary Care Services a Substitute or
Complement for Specialty and Inpatient Services?

• John C. Fortney, Ph.D.
• Diane E. Steffick, Ph.D.
• James F. Burgess Jr., Ph.D.
• Matt L. Maciejewski, Ph.D.
• Laura A. Petersen, M.D., M.P.H.
•  Supported by Department of Veterans Affairs,
  Health Services Research and Development,
  Investigator Initiated Research award to Dr.
  Fortney (ACC 97068-2).

2
Focus on Primary Care

• Primary care stresses
• Population-based medicine
• Continuity of care over time
• Integration or coordination of care
• Frequent use of primary care services may reduce
  the need for specialty/inpatient services and
  contain health care expenditures.
• MCOs have shifted the locus of care from
  specialty and inpatient settings to the primary
  care setting.

3
Possible Substitution Mechanisms

• The prevention of illnesses, or the early
  detection of illnesses that can be treated in the
  primary care setting
• The management of chronic health conditions
• Primary care gatekeeping

4
Possible Complementation Mechanisms

• Supplemental or ancillary services
• The detection of new illnesses requiring
  specialty or inpatient treatment
• Monitoring of chronic illnesses may identify
  acute episodes requiring specialty or inpatient
  treatment

5
Contradictory Findings from Prior Studies

• Aggregate Observational Studies
• Disaggregate Observation Studies
• Pre-Post Studies
• Experimental Studies

6
Study Overview

• Objective
• Examine whether increased use of primary care due
  to improved geographic access results in
  decreases (substitution) or increases
  (complementation) in the use of other types of
  health services.
• Hypotheses
• Primary care is a substitute for specialty
  physical health.
• Primary care is a substitute for specialty mental
  health.
• Primary care is a substitute for inpatient care.
• Primary care is a complement to ancillary
  services.

7
Methods

• Study Design
• Quasi-experimental study design with reference
  group and pre-post data
• Sampling and Matching
• Sample of VA patients residing in CBOC catchment
  areas
• Reference group of matched VA patients not
  residing in CBOC catchment area
• Matched on pre-CBOC travel distance and VISN
• Pre-Post
• Pre-period defined as 18 months prior to CBOC
  establishment
• Post-period defined as months 6-24 after CBOC
  establishment

8
(No Transcript)
9
Variables

• Dependent variables were the change (post minus
  pre)
• specialty physical health visits
• specialty mental health visits
• ancillary visits
• physical health admissions
• mental health admissions
• Inpatient costs
• Outpatient costs
• Explanatory variable of interest was the change
  (post minus pre) in number of primary care clinic
  visits.
• Covariates included
• VISN (represented by fixed effects)
• age
• gender
• race
• marital status
• means test category
• percent service connected
• diagnostic risk category (DCG)

10
Analysis Plan

• Naïve Analysis
• OLS difference-in-differences analysis
• Subject to selection bias effects
• Instrumental Variables Analysis
• ? travel distance used to instrument ? primary
  care
• IV regression estimated in two stages
• Hausman test of biased OLS parameter estimates

11
Results

• The sample included 52,801 veterans.
• Veterans in the catchment areas of CBOCs
  experienced a decrease in travel distance of 23.8
  miles.
• Matched veterans experienced no decrease in
  travel distance.
• ? travel distance was a significant and
  substantial predictor of ? primary care visits
  (?0.016, t13.4, plt0.0001).

12
OLS and IV Regression Results
 
13
OLS and IV Regression Results (Cont.)
14
Policy Relevance

• Organizational strategies designed to promote the
  use of primary care services
• increases enrollees use of ancillary services
• increases enrollees use of mental health visits
• decreases enrollees use of physical health
  visits
• is cost neutral

15
Investigator Affiliations

• HSRD Centers of Excellence
• Center for Mental Health and Outcomes Research,
  North Little Rock, AR.
• Northwest Center for Outcomes Research in Older
  Adults, Seattle, WA.
• Houston Center for Quality of Care and
  Utilization Studies, Houston, TX.
• Management Science Group, Bedford, MA.
•  Affiliated Universities
• University of Arkansas for Medical Sciences -
  Division of Health Services Research, Department
  of Psychiatry, College of Medicine.
• Boston University - School of Public Health.
• University of Washington - Department of Health
  Services.
• Baylor University - Section of Health Services
  Research, College of Medicine

16
Extra Slides
17
Characteristics of VA Healthcare System and
Generalizability

• Gatekeeping without financial incentives
• Gatekeeping policies linked with financial
  incentives for primary care providers could
  strengthen the substitution effect.
• Performance measures
• Lack of performance measures targeting routine
  screening in primary care could weaken
  substitution effect.
• Few barriers to specialty mental health
• Barriers to specialty mental health referrals
  (e.g., higher patient cost-sharing levels) weaken
  complementation effect.

18
Descriptive Statistics
19
Descriptive Statistics (Continued)
20
Testing Management of Chronic Conditions
Substitution Hypothesis for Specialty Medical

• Sub-sample and stratified analyses were conducted
  to determine if the substitution effect was
  stronger for veterans with chronic illnesses and
  worse health status (DCG).
• Sub-sample analysis with 7,657 patients with
  diabetes mellitus
• Full-sample (? -0.40, p0.03)
• Sub-sample (? -0.90, p0.15)
• Stratified analysis with n15,782 for the low
  risk category, n16,375 for the average risk
  category, and n20,644 for the high risk
  category.
• Low risk stratification ? 0.008 (p0.969)
• Average risk stratification ? -0.267 (p0.145)
• High risk stratification ? -0.681 (p0.191)

21
Testing Detection Complementation Hypothesis for
Mental Health Services

• Stratified IV analyses were conducted to
  determine if the complementation effect was
  weaker for patients with primary care services in
  the pre-period.
• Stratified analysis with n13,119 for the no use
  category, n17,124 for the average use category,
  and n22,558 for the high use category.
• Outpatient Mental Health
• No use category ? 1.698 (p0.058)
• Average use category ? 0.886 (p0.546)
• High use category ? 0.617 (p0.327)
• Inpatient Mental Health
• No use category ? -0.018 (p0.51)
• Average use category ? -0.006 (p0.79)
• High use category ? 0.026 (p0.005)

22
Testing Detection Complementation Hypothesis for
Ancillary Services

• Stratified IV analyses were conducted to
  determine if the complementation effect was
  weaker for patients with primary care services in
  the pre-period.
• Stratified analysis with n13,119 for the no use
  category, n17,124 for the average use category,
  and n22,558 for the high use category.
• No use category ? 5.63 (p0.046)
• Average use category ? 6.509 (p0.162)
• High use category ? 5.71 (p0.203)