The Impact of Retail Clinics on Cost

1
The Impact of Retail Clinics on Cost
Utilization Are They Substitutes or Complements
to Physician Services?

• Stephen T. Parente
• University of Minnesota
• Robert Town
• University of Minnesota and NBER
• Presentation at International Health Economics
  Association Meeting
• July 15, 2009

2
Rationale for Investigation

• Retail Clinics emerged as a health market
  innovation in 2000.
• In general, innovation is welfare enhancing
  whether it is through new products or through
  reorganization of products and services.
• A concern is that these organizations can exploit
  administrative pricing irregularities and
  knowledge gaps of consumers leading to reductions
  in consumer welfare.
• Advocates of retail clinics argue they improve
  the efficiency of a highly inefficient health
  care delivery system.
• Critics of retail clinics complain that the new
  service only adds to the inefficiency in the
  system and could greatly reduce consumer well
  being from a clinical perspectives.

3
Summary of Investigation

• Objective Examine the impact of retail clinics
  on cost and use.
• Data Administrative data from a large health
  insurer across multiple US markets to examine the
  evidence of the effects of retail clinics.
• Study design Pre and post retail clinic launch
  differences in utilization between a treatment
  population of retail clinic users and retail
  clinic non-users.
• Econometrics OLS and IV estimation on cost and
  utilization measures.
• Results We find evidence that retail clinics are
  lower cost substitutes for physician office
  visits and their subsequent use does not appear
  to adversely affect measurable proxy metrics for
  quality of care.
• Implications The introduction of retail clinics
  appears to be welfare enhancing.

4
Prior Commentary Literature

• Rosenblatt, et al, 2006 Concerns of retail
  clinics taking away business from Community
  Health Centers.
• Kamerow, 2007 Title says it all Retail Health
  Clinics Threat or Promise? British Medical
  Journal
• Mehrotra et al, 2008 Used ambulatory care
  survey information to show the most common
  conditions seen be those going to retail clinics.
  Provided a descriptive analysis of differences
  in physician service use and cost between retail
  and non-retail clinic users.
• No previous direct empirical analysis of total
  cost, use or welfare impact of retail clinics.

5
Data

• Administrative insurance claims data
• Date of service, zip code of provider of care and
  insured, allowed charge
• Diagnosis and procedure code (CPT)
• Medical, institutional and pharmacy claims
• Used claims available for calendar year 2004
  through 2006
• Provider contract data
• Retail clinics operating by site of actual clinic
  (as opposed to corporate headquarters).
• Address of actual clinic served
• Start date of clinic contract
• Unit of Analysis
• Cost Use measures within 180 day intervals
  indexed on start date of retail clinic opening
  and they get a retail clinic CPT. One
  pre-interval and four-post intervals.

6
Study Population

• Retail Clinic Users (Treatment Population)
• Need to use at least one retail clinic service
• Starting N 31,392
• Enrolled for one year and had pre clinic launch
  health care use 6 month prior to launch N
  23,227
• Non-Retail Clinic Users (Control Population)
• Population who lived in area when clinic was
  operating
• Population who lives in area when clinic was
  about to open
• Need to use at least one CPT that was used in by
  a retail clinic
• Random sampled from a large N to get at starting
  N 39,111
• Enrolled for one year and had pre clinic launch
  health care use 6 month prior to launch N
  27,008

7
Patient Case-mix measured using ADGs

• Adjusted Diagnostic Groups (ACGs) were developed
  by Johns Hopkins University
• Based on combination of diagnosis, age, gender
  information during a period of time
• Can explain variation in cost as well as
  risk-adjustment for premium calculation

8
Top 10 Conditions Seen at Retail Clinics
9
Demographics of Retail Clinic and Non-Retail
Clinic User Populations 1 of 3
10
Demographics of Retail Clinic and Non-Retail
Clinic User Populations 2 of 3
11
Demographics of Retail Clinic and Non-Retail
Clinic User Populations 3 of 3
12
Empirical Framework

• To assess the impact of retail clinic utilization
  on the expenditures and the patterns of care for
  enrollee i in market m in period t we estimate
  parameters from the following model
• yimt am xit b rrit eimt
• where yimt is one of several different measures
  of expenditures or utilization.
• am is a market fixed effect, xit is a vector of
  individual demographic, condition and severity
  controls.
• rit is an indicator for whether the enrollee
  visited a retail clinic and
• eimt is a mean zero residual.
• The parameter of primary interest is r which
  captures the impact of retail clinic utilization
  on the outcome of interest.

13
Endogeneity issue

• Endogeneity of the decision to use a retail
  clinic.
• We address this concern using an instrumental
  variable approach.
• Instrument is the distance from the patients
  home zip code to the nearest retail clinic in
  operation. If there is no clinic open within
  50-miles the distance variable is set to 50
  miles. Use instrument2 as well.
• Gowrisankaran and Town (1999), Geweke,
  Gowrisankaran and Town (2003), and McNeil,
  McClellan Newhouse (1994) use a similar
  identification strategy to measure hospital
  quality.

14
First Stage Estimates of Retail Clinic Use
15
Overall Population Results
16
Chronic Sub-Sample Results
17
Pediatric Sub-Sample Results
18
Core findings

• Costs are less in general for retail clinics
• Most of the cost savings is due to reduced
  physician allowed claim cost
• Substantial reductions in admissions and
  outpatient hosts for chronically ill.
• Substantial reduction in ER use and cost for
  pediatric population.
• Results hold in IV estimation.

19
Limitations Extensions

• One insurers data. But data is internally
  consistent by person.
• Additional instruments. But distance is not a
  bad way to go for now.
• Retail clinic population needs to be weighted to
  correct for over-sampling general population.
  Question is, what is the right weight?

20
Next Potential Steps Beyond Scope

• Use clinical metrics for evidence-based medicine
  as dependent variables.
• Drug compliance requires a refill and days
  supplied variable and also requires a lot of
  caveats.
• But, we dont find pharma is really in play, but
  it could change with more precise measures.
• At the very least, if no pharma change the cost
  reductions for the chronically ill do not pose a
  quality concern.

21
Conclusions

• Patients who visit retail clinics have fewer
  costs without any reduction in our admittedly
  crude measures of quality
• Results suggest retail clinics are serving as
  substitutes for medical care with no obvious
  quality concerns
• More investigation is needed as this market
  evolves