The Effect of Quality Information on Consumer Choice of Health Plans:

advertisement
The Effect of Quality Information on
Consumer Choice of Health Plans:
Evidence from the
Buyers Health Care Action Group
Jean M. Abraham, Roger Feldman,
Caroline Carlin, and Jon Christianson
June 6, 2004
Supported by the Robert Wood Johnson Foundation’s Initiative on
Changes in Health Care Financing and Organization
Importance of Quality Information
• If employees have good information on
health plan quality, they will pick highquality plans and avoid low-quality plans
• Plans will compete to attract enrollees by
improving quality
• Result is double-barreled improvement
• Or so the theory of managed competition
says
Previous Studies
• Typical study analyzes a “natural experiment”
where an employer collects and distributes
information on health plan quality for the first
time
• Pre- and post-information behavior is compared,
e.g., Do employees switch to plans with higher
reported quality scores?
• Results suggest small effect of quality information
in getting employees to avoid low-quality plans
Our Contribution
• We also analyze a natural experiment
– Quality information program sponsored by an employer
coalition in Minneapolis
• So what’s our contribution?
– We study the value of information on a different
margin: How do employees use quality information in a
mature program?
– Information is available, but employees must be aware
of the information and then must decide to use it
– We use an economic model based on search theory to
explain information awareness and use
Conceptual Model: Basics
• Our model is based on Hirshleifer and Riley (1979)
• Individual chooses one of several health plans
subject to imperfect information about health plan
quality
• A quality shock such as a bad experience with her
current health plan leads the employee to search for
quality information
• Information changes expected quality
• Based on the new information, the employee may
decide to switch plans
Buyers Health Care Action Group
(BHCAG)
• Health insurance purchasing and reform coalition
in Minneapolis
– 24 of these employers offered “Choice Plus” in 2002
– Focus on 16 employers where Choice Plus is the sole or
dominant health plan
• Choice Plus
– Direct contracting model with 17 care systems
– Integrated teams of primary care providers, affiliated
specialists and hospitals
– Primary care physicians can affiliate with only one care
system
– Systems are placed in three cost tiers based on riskadjusted bids
Quality Program #1
• Care system quality ratings
– 16,000 enrollees are surveyed every other year by an
independent organization
– Each care system is rated on different dimensions of
quality for adults and children
– Experiences with clinics and medical care in prior year
• how people rated their clinic
• how they rated their doctor or nurse
• how well doctors communicate
• Results are summarized in booklet that some
employers distribute to employees at open
enrollment
Quality Program #2
• “Excellence in Quality Award”
– Good consumer survey scores
– Delivery of preventive care services to a large majority
of their patients
– Proof of quality improvement and outcomes of care in
at least one important way
– Demonstration of care system’s commitment to patient
safety
• Financial prizes
– Gold ($100,000); Silver ($50,000); Special Recognition
• Permission to use in marketing for 2 years
Data
• Primary data were collected in Spring 2002
–
–
–
–
Employee survey stratified by employer
Over-sampled switchers (25% vs. 8% pop.)
Focus on single employees with no dependents
N = 651
• Matched to data on care systems and
employer communication strategies
Dependent Variables
• Two quality awareness questions:
– “During the open enrollment period…do you recall
seeing the “Quality Awards and Consumer Survey
Results” that rated all the care systems on several
aspects of quality and consumer satisfaction?” (33%)
– During the open enrollment period…do you recall
seeing or hearing about the “Excellence in Quality
Awards” that care systems can earn…? (23%)
– “Yes” to either question was counted as quality
information awareness (Abraham et al, 2004)
• Switched care systems in 2002
– Self-reported and confirmed by enrollment file
Independent Variables
Personal Attributes:
•Demographics such as
sex, age, education, job
tenure, Twin Cities
tenure, and chronic
disease
•Overall satisfaction
with 2001 care system
•Have personal
doctor/nurse
•Lost personal
doctor/nurse
Employer
Communication:
Care system
Attributes:
•Performance
results booklet
distributed to all
employees
•Tax-adjusted
premium
difference vs.
best alternative
•Performance
results booklet
distributed on
request
•Quality award
comparison
•Quality rating
comparison
Model
Information Awareness = f (Personal Attributes, Employer
Communication Strategies) + e1
Switch = f (Information Awareness, Personal Attributes, Care
System Attributes) + e2
•
Model is recursive but the error terms may be correlated
(1) Two-Stage least squares on linear probability removes the
correlation
(2) Bi-variate probit explicitly models it (Greene, 1998)
• Identification: Switching equation excludes employer
communication strategies
Bi-variate Probit Results
Information Awareness - Key Results
Overall satisfaction with 2001 care system
Male
Education level
Ln(Job tenure)
Chronic disease
Booklets distributed to all
Booklets distributed on request
Coefficient
(SE)
Marginal Effect
.066*
(.039)
-.282**
(.132)
.152**
(.075)
.124*
(.071)
.014
(.135)
.255*
(.149)
.481***
(.152)
.025
*p<.10, **p<.05, ***p<.01
Note: all other variables in model are statistically insignificant
-.107
.058
.051
.005
.099
.187
Switch – Key Results
Information awareness
Have personal doctor/nurse
Lost personal doctor/nurse
Tax-adjusted premium dif. in dif.
Quality award comparison
Quality rating comparison
Overall satisfaction with 2001 care
system
Chronic disease
Constant
Coefficient
(SE)
.278
(.684)
-.478**
(.196)
.628***
(.206)
.073***
(.014)
.145
(.127)
.141
(.192)
-.029
(.047)
-.250
(.164)
-1.312***
(.375)
Marginal Effect
.029
-.063
.090
.007
.014
.014
-.003
-.023
…
Implications - 1
• Not the right information?
– Focus groups suggest employees want
information about their doctor and the best
doctors and hospitals for their illness
– Information on care systems may be too general
(although more specific than health plan
information)
• Opposing view: overall satisfaction with
2001 care system correlates highly with
things that consumers care about
Implications - 2
• Employer communication methods may be
ineffective
– Employer communication increases the
proportion of employees who are aware of
quality information, but employees don’t use
the information
– Employers may need to be more assertive in
presenting information, especially about poorquality plans
Implications - 3
• Puzzling result: more-satisfied enrollees are
more likely to be aware of quality
information
– Dissatisfied enrollees may reduce their
estimates of quality in all care systems, thus
reducing the expected benefits of search
– Need carefully-designed surveys based on
search theory
Implications - 4
• Our results don’t necessarily disagree with
those of earlier studies
– We examined a mature information program;
other studies looked at first-time behavior
– First-time dissemination of information may
help employees “sort” into the best plan
– Then they are relatively insensitive to new
information
Download