Healthcare Expenditures for University and

advertisement
Healthcare Expenditures for University and
Academic Medical Center Employees Enrolled in a
Pilot Workplace Health Partner Intervention
A d
AcademyHealth
H lth A
Annuall R
Research
hM
Meeting
ti
June 10, 2014
Kenton Johnston, MPH, MS
Jason Hockenberry, PhD
Kim Rask, MD, PhD
Lynn Cunningham, MS, MBA
Ken Brigham, MD
Greg Martin, MD, MSc
Project Research Funding
• Project Funding:
− Emory University
− Emory-Georgia
E
G
i T
Tech
hC
Center
t ffor H
Health
lth Di
Discovery
and Well-Being
2
Introduction
• Background
– Emory-Georgia Tech Predictive Health Institute
launched Center for Health Discovery and Well
Being (The Center)
– In 2007 The Center enrolled first cohort of Emory
University and Emory Healthcare employees in a
h lth partner
health
t
iintervention
t
ti
3
Introduction
• Rationale for Study
− Limited data on health care expenditure impact of
extensive interventions predicated on principles of
predictive health
− Collecting detailed health information and
providing individualized assessments can improve
h lth status
health
t t b
butt also
l iincrease medical
di l utilization
tili ti
4
Description of the Intervention
• Personal health information data collection
and assessment
– Multitude of data collected through biological lab
tests and health assessments
– Participants given opportunity to meet with trained
health partners to interpret individualized data
• Health Partner Intervention (HPI):
− Unlimited interactions with health partners who
were trained professionals (not MDs or nurses)
− Health partners helped to set goals based on
individualized data assessment
− Health education and behavior change coaching
provided during intervention
5
Research Objective
• Research Objective:
– Assess whether there was a difference in health
care expenditures in Emory employees attributable
to The Center
Center’s
s HPI
• Research Question:
− Was there a difference in the PMPM health care
expenditures of HPI participants
o as compared with a control group of non
nonparticipating employees
o attributable to the HPI during the intervention
ti
time
period
i d
o after controlling for other factors?
6
Study Design
• Study Design
– Nested three-year cohort study within the ongoing
longitudinal HPI program
– 12-month
12 month baseline period in 2008 and 24-month
24 month
intervention period in 2009-2010
o Emory made a major change in health insurance between
2007 and 2008,
2008 limiting access to earlier data
7
Study Population
• Study Population:
− Emory University and academic medical center
employees who were also health plan policy
holders during the three
three-year
year study period
− Intervention population: cohort of 556 employees
ever enrolled in the HPI
o 99% of participating employees enrolled in 2008 (year 1)
− Non-intervention population: cohort of 27,650
employees
p y
who did not participate
p
p
in the HPI
8
Data Sources & Variables
• Data Sources
– Administrative health care (medical & pharmacy)
claims and enrollment data on employees
– Linked file indicating HPI participation by year
yearmonth provided by third party vendor
• Variables
V i bl
− Dependent variable: annual total and categorical
(professional, facility, pharmacy) expenditures
− Independent variable: HPI participation
− Covariates: age, gender, health benefit plan type,
health plan member months, active employee
months, year dummies (2008 reference year)
9
Analytic Strategy
• Two-part regression modeling (TPM)
– TPM used to account for zero inflation and rightskewed health care expenditure distribution
– Models controlled for all covariates listed
– Part 1: Logistic regression predicting probability of
any health care utilization
– Part 2: GLM with gamma distribution and log link
to estimate expenditures conditional on utilization
– Robust clustered standard errors at the member
level to account for serial within person variation
• Sensitivity
S
iti it A
Analyses
l
− Tested removal of employees with PMPM
expenditures of $8,333 and $4,167
10
Unadjusted Descriptive Statistics
Emory
Population
Em ory Employee
Em ployee HPI
PHI Population
Emory
Population
Em ory Employee
Em ployee Non-HPI
Non-PHI Population
Year 2008
Year 2009
Year 2010
Year 2008
Year 2009
Year 2010
548
551
538
22 161
22,161
22 223
22,223
22 679
22,679
Medical Member Months
Full-time Employee Months
6,431
6,165
6,529
6,329
6,342
6,185
232,835
196,156
240,259
205,394
241,266
208,472
Plan Type as %
CDHP
POS
PPO
4.2%
68.4%
27.4%
5.1%
69.0%
26.0%
6.5%
93.3%
0.2%
2.0%
66.6%
31.5%
2.9%
67.3%
29.8%
3.8%
94.5%
1.7%
Higher
g
uptake
p
of
CDHP among HPI
participants
Dem ographics
Percent Male
Mean Age in 2008
Age 20-29 % in 2008
Age 30-39 % in 2008
Age 40-49 % in 2008
Age 50-59 % in 2008
Age 60+ % in 2008
31.9%
46 7
46.7
4.9%
21.2%
30.7%
35.6%
7.7%
31.9%
46 6
46.6
5.1%
21.4%
30.5%
35.4%
7.6%
31.4%
46 5
46.5
5.2%
21.0%
31.0%
35.1%
7.4%
32.7%
42 8
42.8
17.4%
28.0%
24.3%
19.3%
10.8%
32.6%
42 3
42.3
19.4%
26.9%
24.2%
19.0%
10.2%
32.8%
41 6
41.6
21.5%
26.2%
23.8%
18.3%
9.5%
Mean age of HPI
participants ~4
years older
Em ployee Type as %
Active Full Time
Active Part Time Seasonal
COBRA Continuee
Early Retiree
Medicare Eligible Retiree
Other/Unknow n
95.8%
3.6%
0.0%
0 0%
0.0%
0.0%
0.5%
96.2%
2.7%
0.0%
0 4%
0.4%
0.2%
0.5%
97.8%
0.7%
0.0%
0 4%
0.4%
0.4%
0.7%
84.9%
7.7%
0.6%
0 6%
0.6%
3.7%
2.5%
86.0%
7.6%
0.0%
0 5%
0.5%
3.8%
2.2%
87.1%
6.2%
0.0%
0 5%
0.5%
3.9%
2.3%
31.7
14 9
14.9
2.5
11.5
11.6
38.6
33 1
33.1
2.9
12.3
12.0
54.9
49 2
49.2
3.6
11.1
11.7
76.5
58 5
58.5
4.1
10.3
12.1
73.5
56 6
56.6
3.7
10.8
12.7
56.3
59 9
59.9
3.5
10.0
12.2
Population (Unique Mem bers)
Healthcare Utilization
ED Visits per 1000 members
H
Hospital
it l St
Stays per 1000 members
b
Average Length of Stay in days
Professional Encounters Mean PMPY
Prescriptions Filled Mean PMPY
Total of 556 unique
HPI participants
HPI participants
~10% points more
likely to be full time
Lower hospital &
highe professional
higher
p ofessional
utilization at
baseline for HPI 11
Unadjusted PMPM Expenditures
HPI participants
PMPM expenditures
19% lower than nonparticipants in 2008
Figure
g
1. Unadjusted
j
PMPM Expenditures
p
for 2008-2010
PHI
HPIMembers
Members
Non-HPI Members
Members
Non-PHI
$437
$407
$414
$404
$389
$329
Year 2008
Year 2009
Year 2010
Emory
Employee
Population
Em
ory Em
ployee HPI
PHI Population
Population (Unique Mem bers)
Medical Member Months
Full-time Employee Months
Unadjusted trend
among participants
higher than among
non-participants
Emory
Population
Em ory Employee
Em ployee Non-HPI
Non-PHI Population
Year 2008
Year 2009
Year 2010
Year 2008
Year 2009
Year 2010
548
551
538
22,161
22,223
22,679
6,431
6,165
6,529
6,329
6,342
6,185
232,835
196,156
240,259
205,394
241,266
208,472
Mean Expenditures - Per Mem ber Per Month
Total
$
329.15
Facility
$
80.07
Professional
$
155.10
Pharmacy
$
93.98
$
$
$
$
389.10
117.95
169.37
101.78
$
$
$
$
404.16
135.88
168.72
99.55
$
$
$
$
406.73
154.62
148.17
104.37
$
$
$
$
437.03
168.15
157.75
111.13
$
$
$
$
By 2010 participants’
PMPM expenditures
were 2% lower than
non-participants
414.07
159.57
143.41
111.08
12
TPM Regression Results
Total
Expenditures
Facility
Expenditures
Professional
Expenditures
Pharmacy
Expenditures
PMPM
PMPM
PMPM
P-Value
PMPM
P-Value
$ 0.004 1.000
$ (33.66) 0.009
PHI Participants
$ 25.74 0.275
$ 24.95 0.177
PHI Participants in 2009
$ 4.24 0.881
$ 8.90 0.621
PHI Participants in 2010
(PHI non-participants and year 2008 are the referent groups)
$ 22.80
$ 5.09
$ 4.08
0.009
0.551
0.777
$ 3.46
$ (0.87)
$ (8.01)
0.681
0.860
0.230
$ 154.71 <.001
POS Plan Type (vs CDHP)
$ 130.79 <.001
PPO Plan Type (vs CDHP)
(CDHP plan type is the referent group)
$ 51.36
$ 53.79
<.001
<.001
$ 42.55
$ 25.40
<.001
0.011
$ 53.17
$ 49.85
<.001
<.001
p
Unique
q
PHI Participant
PHI Non-Participant Unique
554
27,486
---
554
27,486
---
554
27,486
---
P-Value
P-Value
(HPI non-participants and year 2008 are the referent groups)
554
27,486
---
− No difference in total expenditures at baseline
− Participants had significantly lower facility costs and higher
professional costs at baseline
g
PMPM expenditures
p
− No differences in total or categorical
attributable to the HPI in participants for 2009 and 2010
− Employees not in CDHP had much higher total PMPM
expenditures than employees in the CDHP
− Further sensitivity analyses confirmed findings
13
Summary of Findings
• We find no difference in total or categorical
expenditures attributable to the HPI
– After excluding outliers and controlling for age,
gender health benefit plan
gender,
plan, and employment
– This is in contrast to unadjusted data that show a
higher cost trend in HPI participants
• Employees in traditional health benefit plans
had significantly higher PMPM expenditures
than employees in the high deductible CDHP
14
Limitations
• Due to an absence of baseline period (2007)
data, our regression models treated the first
intervention year (2008) as the baseline
– 2007 insurance
i
claims
l i
d
data
t unavailable
il bl
• We did not have access to individual-level
linked data on extent of participants’
program engagement over time
• Because the intervention was introduced
during the recession, caution should be
taken in interpreting these findings
− Poor macroeconomic conditions are known to
impact health
h l h services use and
d cost
15
Conclusions & Implications
• HPI does not appear to have raised health
care costs in the short term
– This despite the increased provision of diagnostic
and preventive care which could potentially induce
more treatment
• It may be
b possible
ibl tto iintroduce
t d
iintensive
t
i
predictive health interventions without
increasing short-term
short term health care costs
16
Download