University of Southern California Dissertations and Theses

The impact of Patient-Centered Medical Home on a managed Medicaid plan

(USC Thesis Other)

The impact of Patient-Centered Medical Home on a managed Medicaid plan

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

The Impact of Patient-Centered Medical Home on a Managed Medicaid Plan
by
Li-Hao Chu

Dissertation Submitted in Partial Fulfillment
of the Requirements for the Degree of
Doctor of Philosophy
Pharmaceutical Economics and Policy

University of Southern California
May 2016

i
Abstract
The California Medicaid program, also known as Medi-Cal is the largest
Medicaid program in the nation. With the rollout of Medicaid expansion, the total Medi-
Cal beneficiaries have reached about 12 million, constituting nearly 30% of the state's
population. Knowing that the resource and capacity at the provider side can hardly keep
up with the growth of the Medi-Cal population, the identification of a model of care to
direct patients to the proper setting for care has emerged as a top priority for the state’s
health policy.
Safety-net clinics play a pivotal role in delivering both primary and specialty care
to millions of low-income people, and yet we know little about their performance under
different health care delivery models. With the implementation of Patient-Centered
Medical Home (PCMH) in early 2012, where patient engagement, health information
technology, coordination of care, quality of care and access to care was integrated into
clinics’ daily practice, it would be interesting to examine the impact of this model on the
healthcare use of Medi-Cal beneficiaries.
This dissertation included two main objectives. The first objective evaluated the
impact of PCMH on non-disabled Medi-Cal beneficiaries. The analysis shows that among
clinics with less than 10% Seniors and Persons with Disabilities (SPD) membership,
transformation to PCMH was associated with increased use of office visits and reduced
use of emergency departments (ED). In particular, PCMH clinics (relative to non-PCMH
clinics) reduced ED visits by an average 70 visits per thousand members per year
(PTMPY) and avoidable ED visits by 20 visits PTMPY. No significant change in office

ii
visits or reduction in ED use was found in clinics with SPD membership greater than
10% suggesting that the beneficial effects of the PCMH model in safety net clinics can be
muted by a sudden influx of heavy utilizers.
The second objective evaluated the impact of PCMH on Medi-Cal beneficiaries
with disabilities. The finding shows that among patients who had at least one office visit
in a year, the odds ratio (OR) of having at least one ED use drops around 22-24% (p-
value < 0.05) when comparing PCMH clinics with non-PCMH clinics. Similarly, the OR
of having at least two ED visit drops around 40% (p<0.05).
Both findings led to the same conclusion that the adoption of the PCMH model in
safety-net clinics can effectively reduce ED use. Improved access to care was recognized
as the key attribute of the success of the PCMH model by the leaders from safety-net
clinics.

iii
Dedication
This dissertation work is dedicated to my wife, Yuchia Chung, who is my most
enthusiastic cheerleader and the most amazing wife and mother. In the past few years, she
helped me to focus on my dissertation by running all household errands and taking care
of our two years old boy. I wouldn’t be able to complete this degree without her giving
up so much. In addition, my little toddler, who brings so much joy, love and happiness to
the family has been the biggest inspiration to me. I regret that you had less of my time
and attention in the first two years of your life. I will try my best to make it up for you
and your dear mom.

iv
Acknowledgment
It is so fortunate to have Dr. Neeraj Sood as my advisor as well as dissertation
chair. Starting from the introduction of Applied Pharmacoeconometrics to the preparation
of publication, his expertise, knowledge and flexibility in guiding me through this
dissertation has been indispensable to the completion of this PhD program.
Professionally, he is an example of excellence as a researcher, whose insight and opinion
has advanced this dissertation to be more methodologically and theoretically sound.
Personally, he is a great mentor who continually raises the bar for students and allows
them to strive for their best performance. I am so grateful to his instruction and support.
I would like to thank Dr. Jennifer Sayles for inspiring this dissertation. With her
support and permission to participate in a real-life project, I was able to gain the first-
hand experiences in research planning and program evaluation. Above all, her generosity
in sharing her operational experiences and offering resource and time to this study
enriched my understanding in implementing a program in a real world setting. Working
with her was such a pleasant learning experience. Thank you, Dr. Jennifer Sayles!
I am also very grateful to have Dr. Geoffrey Joyce in my committee. His rigorous
feedback and comments ensured that our findings were on firm footings.
Both Drs. Joel Hay and John Romley also contributed their time and expertise in
shaping the dissertation at the stage of qualifying exam. I am very thankful of their help.
In addition to the support of dissertation committee members, my colleagues from
L. A. Care Health plan have provided meaningful and enjoyable discussion. It was very
grateful to work with them.

v

Table of Contents
Abstract ................................................................................................................................ i
Dedication .......................................................................................................................... iii
Acknowledgement ............................................................................................................. iv
Table of Contents .................................................................................................................v
List of Tables .................................................................................................................... vii
List of Figures .................................................................................................................. viii
Chapter 1: Introduction ........................................................................................................1
Background ....................................................................................................................1
The Problems of Medi-Cal Program ....................................................................... 1
The Patient-Centered Medical Home Model .......................................................... 2
Safety-Net Clinics ................................................................................................... 2
Senior Person with Disabilities ............................................................................... 3
Objectives ............................................................................................................... 5
Chapter 2: Literature Review ...............................................................................................6
Summary ........................................................................................................................6
Experiences from Community Care of North Carolina .................................................6
Experiences from New York State Medicaid ................................................................7
Experiences from the Rhode Island Chronic Care Sustainability Initiative Pilot
Program ..............................................................................................................8
Chapter 3: The Impact of Patient Centered Medical Homes on Safety-Net Clinics .........11
Abstract ........................................................................................................................11

vi

Introduction ......................................................................................................13
Methods........................................................................................................................15
Analysis Result ............................................................................................................20
Discussion ....................................................................................................................27
Chapter 4: Turn the Tide of Emergency Department Use in People with
Disabilities .............................................................................................................31
Abstract ........................................................................................................................31
Introduction ..................................................................................................................33
Methods........................................................................................................................36
Analysis Results ...........................................................................................................45
Discussion ....................................................................................................................47
Chapter 5: Conclusions ......................................................................................................50
References ..........................................................................................................................53
Appendix ............................................................................................................................59

vii

List of Tables
Table 3.1 Study cohorts derivation flow ........................................................................... 17
Table 3.2 The distribution of SPDs in PCMH Clinics ...................................................... 18
Table 3.3 Annual demographics and health resource utilization in all clinics in pre-
PCMH period ............................................................................................................ 20
Table 3.4 Annual demographics and health resource utilization in clinics with
<10% SPD Membership in pre-PCMH period ......................................................... 21
Table 3.5 Annual demographics and health resource utilization in clinics with
≥10% SPD Membership in pre-PCMH period ........................................................ 22
Table 3.6 Unadjusted difference in utilization rates between PCMH and non-PCMH
groups ........................................................................................................................ 24
Table 3.7 Adjusted difference in utilization rates between PCMH and non-PCMH
groups ........................................................................................................................ 25
Table 4.1 Population characteristics and health resource utilization based on
medicaid beneficiaries in 2011-PCMH vs. Non-PCMH clinics after matching
at clinic level ............................................................................................................. 38
Table 4.2 Population characteristics - PCMH vs. Non-PCMH clinics based on
medicaid beneficiaries with disabilities in 2012 after matching at member level ... 39
Table 4.3 Health resource utilization – PCMH vs. Non-PCMH in cohorts 1-7 ............... 46

viii

List of Figures
Figure 4.1 The study time line of the comparison of healthcare utilization between
PCMH and non-PCMH clinics…………………………………………………….36
Figure 4.2 The study cohort creation flow ……………………………………………...37

Figure 4.3 Steps of creating propensity score matching at clinic level……….…………42

Figure 4.4 Steps of creating propensity score matching at member level…....………….43

1

Chapter 1: Introduction
Background
The Problems of Medi-Cal Program
The California Medicaid program, also known as Medi-Cal is experiencing a
rapid growth in its membership. With the rollout of Medicaid expansion in 2014, it has
led to the explosion in Medi-Cal enrollment. As of October 2015 the Medi-Cal program
has reached about 12 million enrollees, which constitutes 30% of the state’s population,
and consumes 15% of state expenditures. Given its size and spending, one question
deserves all Californians’ attention is that “do we have sufficient resources to provide
care to all Medi-Cal beneficiaries”?
Prior studies have shown that the Medi-Cal program is associated with frequent
hospital admissions and heavy reliance on the emergency department (ED) in comparison
to commercially insured patients (McConville & Lee, 2008). According to the study
from Hsia et al., Medi-Cal beneficiaries had ED rate four times higher than the privately
insured, and 2.5 times higher than the uninsured. It also grew the fastest from year 2005
to 2010. In addition, Medicaid patients consistently had the highest rate of visits for
potentially preventable conditions (Hsia, Brownell, Wilson, Gordon, & Baker, 2013).
Studies on this surging Medicaid use in ED have pointed to unmet health needs
and lack of access to appropriate primary care (Sommers, Boukus, & Carrier, 2012). To
turn the tide of this growing ED use, as a result of the explosion of Medicaid enrollment,
this dissertation aimed to evaluate the impact of an innovative care model – Patient-
Centered Medical Home on reducing the ED use in a managed Medicaid plan.
2

The Patient-Centered Medical Home Model
Patient-centered medical home is a model of care where each patient has an
ongoing relationship with a primary care physician who leads a team that takes collective
responsibility for the patient’s care. The physician-led care team may include roles for
nurse practitioners or physician assistants and is responsible for providing all of the
patient's health care needs. When needed, the team also arranges for appropriate care with
other qualified physicians. A PCMH model emphasizes enhanced care through open
scheduling, expanded hours, and communication between patients, providers and staff.
Care is also facilitated by disease registries, information technology, health information
exchange among providers and other means to ensure that patients receive the proper care
in a culturally and linguistically appropriate manner (Stange et al., 2010).
Safety-Net Clinics
Safety-net clinics play a pivotal role in delivering both primary and specialty care
to millions of low-income Californians (Saviano & Powers, 2005). They are mainly
defined by their mission to maintain an open-door policy, providing health care services
to individuals and their families regardless of their ability to pay. These clinics comprise
licensed primary care clinics, clinics operated by governmental entities such as counties
and cities, and clinics operated by federally recognized Indian tribes or tribal
organizations.
Safety-net clinics may be operated by for-profit corporations, public agencies, or
private, nonprofit organizations. There is no legal definition of a safety-net clinic
(Bindman, Grumbach, Bernheim, Vranizan, & Cousineau, 2000).
3

Many safety-net clinics have specific legislative mandates to provide health care
services to the medically indigent as a condition of federal or state funding and/or
reimbursement from public health programs. For example, whether operated by public
agencies or private nonprofit organizations, federally qualified health centers (FQHCs)
and FQHC look-alikes (described further in Section VI) are required by federal law to
provide certain services. Similarly, counties operate clinics to provide services pursuant
to the Section 17000 mandate under state law.
Seniors and Persons with Disabilities
In 2011, Medicaid beneficiaries enrolled on the grounds of disability composed a
small fraction of the total enrollment: approximately 15% ("Medicaid Facts: Distribution
of Medicaid Enrollees by Enrollment Group," 2011). This small fraction of enrollees
disproportionately impacted spending, representing a whopping 42% of the total
spending Medicaid nationwide ("Medicaid Facts: Medicaid Spending by Enrollment
Group ", 2011). The statistic for the state of California was worse yet, with disabled
beneficiaries representing just 9% of all beneficiaries, nevertheless constituting 41% of
the spending ("Medicaid Facts: Distribution of Medicaid Enrollees by Enrollment
Group," 2011; "Medicaid Facts: Medicaid Spending by Enrollment Group ", 2011). The
trends in spending are hardly news though, as the proportion of expenditures on disabled
beneficiaries has only increased in the last decades (Vladeck, 2003). Naturally, cost
cutting methods have been a topic of interest among providers, payers and policy makers.
California, on the heels of the Patient Protection and Affordable Care Act (ACA), was
one of the first states to plan reform through submission of the Medicaid 1115 waiver in
4

2010. Notably, one component of this Bridge to Reform is a transition of Seniors and
Persons with Disabilities (SPD) from traditional Fee For Service (FFS) to managed care
in an effort to promote coordinated systems of care more cost efficiently. Another major
component is the support for reform of safety-net facilities (Harbage P, 2012).
Understanding the fundamental characteristics of this SPD population,
specifically those with disabilities, is crucial to understanding the implications of this
transition to managed care. Disabled beneficiaries often present the most complex cases
to a health system as compared to other Medicaid beneficiaries and require a wide array
of specialists and specialized facilities ("People with Disabilities and Medicaid Managed
Care: Key Issues to Consider," 2012). They qualify for Medicaid based on a wide variety
of handicaps, including serious mental illness and behavioral health diagnoses,
developmental disorders, severe chronic illness and disabling brain or spinal cord
injuries. Considering their low health and functional status, their proportionally higher
utilization of health services is not entirely unexpected. What’s more is that in all of the
aforementioned spending, there were still unmet needs under FFS. In a survey of
working-age disabled beneficiaries on FFS, more than 38% reported an unmet need,
commonly citing availability, accessibility and language as barriers (Coughlin, Long, &
Kendall, 2002). Transitioning such a special needs population to a managed care setting
requires a multifaceted approach that not only serves the needs sought but also addresses
the barriers this population faces in a manner more cost efficient than FFS.
5

Objectives
This study hones in on a previously untested population: safety-net clinics serving
the Greater Los Angeles Area (i.e., the largest urban area in California with a Hispanic
population approximately 50% of the total). Specifically, this dissertation evaluated the
impact of PCMH on safety-net clinics with the following two objectives.
1) To evaluate the impact of practice transformation to PCMH of safety-net
clinics in a non-elderly Medicaid population
2) To evaluate the impact of practice transformation to PCMH of safety-net
clinics in non-elderly Medicaid Beneficiaries with disabilities
6

Chapter 2: Literature Review
Summary
Encouraging results in improving the quality of patient care, reducing
hospitalization and ED visits were presented in some PCMH pilot projects mainly from
integrated delivery systems and multi-payer sponsored PCMH initiatives (Friedberg, Lai,
Hussey, & Schneider, 2009; Gilfillan et al., 2010; Grumbach & Grundy, 2010; Reid et
al., 2010; M. Takach, 2011). Results have been promising in Medicaid as well. Medicaid
provides insurance to underserved, minority, and low-income patients, a population most
susceptible to fragmented and uncoordinated care. As such, more than half of the states
implemented a variety of payment policy changes and other reforms in Medicaid to help
primary care providers function as PCMH (Grumbach & Grundy, 2010; Mary Takach,
2012). Thus far, Colorado, Minnesota, New York, North Carolina, Oklahoma, South
Carolina, and Vermont have reported reductions in ED visits, hospitalizations and costs.
In this literature review, the focus was on studies that had study population or
research method similar to this dissertation.
Experiences from Community Care of North Carolina
The Community Care of North Carolina is a systemic care management
intervention program (Community Care of North Carolina) with person-centered medical
homes throughout North Carolina. It covered non-elderly Medicaid beneficiaries with
disabilities (PCMH group: 537,336; Comparison group: 5,275,017) in the time period
from January 1, 2007 to September 30, 2011 (Fillmore, DuBard, Ritter, & Jackson,
2014).
7

The study design was a pre-post, intervention/comparison group, difference-in-
differences mixed model, which directly matched cohort samples of enrolled and
unenrolled members on strata of pre-enrollment pharmacy use, race, age, year, months in
pre-post periods, health status, and behavioral health history. Hierarchical Models by
years were conducted to estimate the CCNC program effects on cost. To account for
potential clustering, variables for regional differences were included as fixed effects and
variables for physician group were included as random effects.
The study team found significant cost avoidance associated with program
enrollment for the non-elderly disabled population after the first years, savings that
increased with the length of time in the program. The impact of the program was greater
in persons with multiple chronic disease conditions.
Experiences from New York State Medicaid
In 2009, New York State approved a statewide medical home initiative program
that provided enhanced payments to primary care providers for providing primary care
services. This initiative also involved Managed Medicaid Plan (MMP), child health plus
(CHPlus) and Medicaid FFS recipients (M. Takach, 2011).
For the 2010 measurement year, there were 333,847 MMC members in the
matched study population of which 41% were less than 18 years old. For 2011, there
were 653,194 MMP members in the final matched study population.
A matched comparison study of two subsets of MMP members (PCMH group and
non-PCMH group) was used to assess differences in utilization between the PCMH and
non-PCMH groups. Because certain enrollee factors, such as health status, may impact
8

healthcare utilization, matched comparison studies allow for the selection of a
comparison group that is similar in terms of patient demographics and clinical
characteristics.
The comparison was made by evaluating the difference of utilization between
PCMH and non-PCMH group. In 2010 there were slightly higher rates of ED visit and
very similar rates of outpatient primary care visits in adult PCMH members compared to
adult non-PCMH members. There was very little difference in ED visit rates and slightly
lower rates of outpatient primary care visits in pediatric PCMH members compared to
pediatric non-PCMH members.
Experiences from the Rhode Island Chronic Care Sustainability Initiative Pilot
Program
This project evaluated the Rhode Island Chronic Care Sustainability Initiative
(CSI). This initiative was launched in 2008 with support from the 3 largest commercial
insurers in the state: Blue Cross Blue Shield of Rhode Island, Neighborhood Health Plan,
and United Health Care. These insurers cover approximately two-thirds of the patients
seen by participating practices (Rosenthal et al., 2013)
Five primary care practices, including 1 federally qualified health center, with a
total between them of 45 primary care providers, volunteered to participate in the pilot
program. The pre-intervention period began October 1, 2006, and ended September 30,
2008; the post-intervention period began October 1, 2008, and ended September 30,
2010.
9

An interrupted time series design with propensity score–matched comparison
practices was adopted. To adjust for residual confounding after matching and to address
the substantial number of patients who do not use services in a typical year, we estimated
the effect of the CSI practice-level measures of utilization using zero-inflated negative
binomial models. These 2-part negative binomial models separately estimate the
probability that a patient would have zero utilization for the entire period from a model
that estimates the number of events and are used where a large share of the population
has an extremely low risk of an event. All analyses included practice fixed effects and
accounted for clustering at the practice level. Marginal effects and their associated P
values were calculated using the Δ method.
After 2 years, a pilot program of a patient-centered medical home was associated
with a significant reduction in ambulatory care sensitive emergency department visits.
Conclusion
The only literature studying the impact of PCMH on people with disabilities was
from the experience of the Community Care of North Carolina. This was a program
supported by the state government and has medical home as one of the modules to
manage its Medicaid beneficiaries (Fillmore et al., 2014; Rich, Lipson, Libersky, Peikes,
& Parchman, 2012). Such a design was not applicable to states like California, where
insurers delegate responsibilities, such as provider credentialing, utilization management,
and chronic disease management, to a group of physicians or an IPA. Therefore, this
dissertation is unique in evaluating a PCMH model initiated at the clinic level rather than
from a state’s initiative.
10

Although the study population of New York State was similar to our objective
one, the rudimentary analysis (only univariate analysis is available) and with only one
year follow-up period hindered readers gaining insight from their implementation.
Besides a handful of states published their own evaluation, I also referenced Mary
Takach’s review paper where she summarized many states’ PCMH results through a
personal communication with states’ health service agency (Mary Takach, 2012).
Unfortunately, these studies couldn’t be generalized to our study population because the
majority of them was a state run program where a multi-payer model was adopted.
Moreover, none of them specifically looked into the impact of PCMH on the safety-net
clinics. As a result, our study is unique in addressing the following research questions:
1) Can PCMH effectively reduce ED use in a Managed Medicaid Plan?
2) Can PCMH effectively reduce ED use in safety-net clinics?
3) How did PCMH respond to an influx of SPD?
4) Can PCMH effectively reduce ED use in people with disabilities?
11

Chapter 3: The Impact of Patient Centered Medical Homes on Safety-Net Clinics
Abstract
Objective
The objective of the study was to evaluate the impact of practice transformation to
Patient-Centered Medical Homes (PCMH) of safety-net clinics in a Managed Medicaid
plan.
Study Design and Method
This study employed a quasi-experimental, difference-in-difference study design.
The PCMH group included seven safety-net clinics (22,870 members) recognized as
PCMH in late 2011 in the greater Los Angeles area. The comparison group consisted of
110 other safety-net clinics (143,530 members) in the same area. The study timeline
ranged from January 2011 to December 2013, running concurrently with a federal
waiver, effective June 1, 2011, under which California began transitioning Seniors and
Persons with Disabilities (SPDs) with fee-for-service (FFS) Medicaid into managed care
systems. This transition allowed us to examine whether the impact of transformation to
PCMH was influenced by the extent of increase in SPD population.
Results
Our findings suggest among clinics with less than 10% SPD membership,
transformation to PCMH was associated with increased utilization of office visits and
reduced use of emergency departments (ED). In particular, PCMH clinics (relative to
non-PCMH clinics) reduced ED visits by an average 70 visits per thousand members per
year (PTMPY) and avoidable ED visits by 20 visits PTMPY. No significant change in
12

office visits or reduction in ED were found in clinics with SPD membership greater than
10%.
Conclusions
The adoption of the patient centered medical home model in safety-net practices
can effectively reduce rates of ED and avoidable ED visits, and increase the use of office
visits among Medicaid patients. However, the beneficial effects of PCMH model can be
muted by a sudden influx of heavy utilizers.
13

Introduction
Safety-net clinics play a pivotal role in delivering both primary and specialty care
to millions of low-income Californians (Saviano & Powers, 2005). These clinics
comprise licensed primary care clinics, clinics operated by governmental entities such as
counties and cities, and clinics operated by federally recognized Indian tribes or tribal
organizations. In 2011, the State of California authorized a Section 1115 Medicaid waiver
that mandated enrollment of Medicaid-only seniors and people with disabilities (SPD)
into Managed Medicaid Plans. In turn, this led to an influx of patients with chronic
conditions into safety-net clinics.
The California Medicaid population is associated with frequent hospital
admissions and heavy reliance on the emergency department (ED) in comparison to
commercially insured patients (McConville & Lee, 2008). In light of this knowledge, a
Los Angeles local health plan initiated a pilot program to transform selected safety-net
clinics into Patient-Centered Medical Homes (PCMH). This practice transformation was
designed to improve patient care and alleviate the impact of the SPD influx through the
standards of the National Committee for Quality assurance (NCQA) PCMH.
Patient-centered medical home is a model of care where each patient has an
ongoing relationship with a primary care physician who leads a team that takes collective
responsibility for the patient’s care. The physician-led care team may include roles for
nurse practitioners or physician assistants and is responsible for providing all of the
patient's health care needs. When needed, the team also arranges for appropriate care with
other qualified physicians. A PCMH model emphasizes enhanced care through open
14

scheduling, expanded hours, and communication between patients, providers and staff.
Care is also facilitated by disease registries, information technology, health information
exchange among providers and other means to ensure that patients receive the proper care
in a culturally and linguistically appropriate manner (Stange et al., 2010).
Our research focused on a previously untested population: safety-net clinics
serving the Greater Los Angeles Area (i.e., the largest urban area in California with a
Hispanic population approximately 50% of the total). Specifically, our primary analysis
was to evaluate the impact of PCMH on selected safety-net clinics. In addition, the
transition to PCMH coincided with the state mandated switch of SPDs from Fee-for-
Service (FFS) to a Managed Medicaid Plan (MMP). This switch of SPDs, whose demand
on health services is five to ten times higher than regular Medicaid members, posed a
potential complication (Graham, Kurtovich, Ivey, & Neuhauser, 2011). The new, high-
utilizing SPD members could crowd out or delay the routine medical services for the non-
SPD Medicaid population at safety-net clinics. Such phenomenon, on the one hand,
posed a challenge to our study design; on the other hand, it led to our second research
question: Was the PCMH model less effective in clinics that experienced a larger influx
of heavy utilizers. Ultimately, the objectives of this study included the primary aim to
evaluate the impact of PCMH on health care utilization in a Managed Medicaid
population served by safety-net clinics; and the secondary aim to assess the response of
PCMH to a sudden influx of SPDs.
15

Methods
Study Cohorts
Eleven safety-net clinics, associated with a large MMP in the greater Los Angeles
area, were recognized as patient centered medical homes in late 2011 and early 2012
according to the NCQA PCMH standards ("Patient-Centered Medical Home
Recognition," 2014). Seven of them were selected into the treatment (PCMH) group after
applying the following selection criteria: 1) located within Los Angeles County; 2)
defined as stand-alone, not-for-profit, licensed free clinic, community clinic or federally
qualified health center that provides primary care services; 3) part of the Health Plan’s
contracted provider network; and 4) provide care to at least 300 of the plan’s members,
including adults and children.
110 other safety-net clinics that were contracted with the same MMP and located
in Los Angeles County but did not acquire PCMH recognition from the National
Committee for Quality Assurance (NCQA), Utilization Review Accreditation Committee
(URAC), and Joint Commission prior to 2013 were treated as the comparison (non-
PCMH) group.
Data Source
Administrative claims data from January 1, 2011 through December 31, 2013
from a local MMP were used in the analysis. The study timeframe encompassed pre- and
post- PCMH transformation. Under an MMP, all members are required to select or
otherwise assigned to a primary care physician. Members who were served by a primary
care provider (PCP) affiliated with a PCMH clinic was grouped into the PCMH cohort.
16

Members served by PCPs affiliated with non-PCMH clinics were grouped into the non-
PCMH cohort. Data elements drawn from these databases included: member
demographics (age, gender, race/ethnicity and enrollment history), service dates, episodes
(hospital admission and emergency department visit), ICD-9 diagnosis codes, procedure
codes and pharmacy claims.
Comorbidities were identified based on the Medicaid RX model, which is a
pharmacy based risk adjustment model used to adjust capitated payments to health plans
that enroll Medicaid beneficiaries (Gilmer, Kronick, Fishman, & Ganiats, 2001). The Top
twenty of the total forty five therapeutic classes in our study cohort included
Asthma/COPD, attention deficit disorder, cardiac, depression/anxiety, diabetes, EENT
(eye, ear, nose and throat), end stage renal disease, gastric acid disorder, gout, herpes,
hyperlipidemia, infections (high, medium, low), inflammatory/autoimmune,
malignancies, multiple sclerosis/paralysis, nausea, pain, seizure disorders and thyroid
disorders.
Study Design
This study employed a quasi-experimental, difference-in-difference study design.
The annual utilization rates were measured over a three-year period with 2011 as the
baseline representing pre-PCMH period, and 2012-2013 as post-PCMH period. Overall,
these rates were compared between the PCMH and non-PCMH groups. We focused on
the non-SPD population under 65 years old only as the SPD population did not complete
the transition to an MMP till mid-2012. Additionally, patients who switched between the
PCMH group and comparison group during the study period were also excluded from the
17

analysis. Lastly, a 10-month minimum Medicaid eligibility during each study year for
both PCMH and non-PCMH groups was required to ensure sufficient exposure or
interaction between patients and their primary care physicians.
Table 3.1
Study Cohorts Derivation Flow
PCMH Non-PCMH Total
Starting Total

37,066 100.0%

236,882 100.0%

273,948
Include Non-SPD only

35,213 95.0%

227,644 96.1%

262,857
Continuous Eligibility >= 10 months

24,297 65.6%

154,115 65.1%

178,412
Age <65

23,892 64.5%

147,984 62.5%

171,876
clinic size > 300 members

22,936 61.9%

145,024 61.2%

167,960
Stay in the same clinics during the study
period

22,870 61.7%

143,530 60.6%

166,400

In late 2011, California began transitioning SPDs from fee-for-service (FFS)
Medicaid into managed care systems. The health service consumption pattern of SPDs
varied dramatically from the non-SPD Medicaid members who were accustomed to a
managed care environment (Graham et al., 2011). Moreover, the sudden influx of high
health care utilizers in safety-net clinics could have potentially crowded out the existing
patients perhaps making the PCMH model less effective. Therefore, we further divided
the study cohorts into two clinic types or groups: those with less than 10% SPD
membership and those with more than 10% membership. The 10% cut off was found as a
natural break while examining the distribution of SPDs in seven PCMH clinics. (Table
3.2) The proportion of SPDs at each clinic was determined based on the year 2012.
18

Table 3.2
The Distribution of SPDs in PCMH Clinics
PCMH Clinic ID % of SPD members Total Members
1 7.60% 2502
2 8.80% 310
3 8.90% 2161
4 11.80% 5901
5 12.10% 2854
6 12.60% 1532
7 14.00% 8034

Outcome of Interest
Hospital admissions, ED visits and office visits were measured using NCQA
standard definitions and served as outcomes of interest in this study ("Healthcare
Effectiveness Data and Information Set," 2013).
Excessive ED visits have shown to be a driver of inefficiency and waste of health
care resources. To more thoroughly examine the impact of PCMH on ED visits,
avoidable ED visits, as defined by the California Department of Health Services
Collaborative, were also included ("Statewide Collaborative Quality Improvement
Project Reducing Avoidable Emergency Room Visits ", 2012).
All utilization rates were presented as per thousand members per year (PTMPY).
Statistical Analysis
Observable characteristics between the PCMH and non-PCMH groups were
compared using the Kruskal-Wallis and chi-square test for continuous and categorical
variables respectively.
19

For the main outcomes of interest, difference in difference analyses were
conducted by fitting generalized estimating equations with negative binomial
distributions and with robust standard errors to account for heteroscedasticity, and
clustering of patients within practices (Zeger & Liang, 1986). Independent variables
included indicators for year 2012 and 2013 respectively where year 2011 serves as a
reference, interactions between time periods and PCMH/non-PCMH status,
demographics, and comorbidities.
A 2-tailed p-value <0.05 was treated as significant in all statistical tests. All the
data management and analyses were conducted using SAS version 9.3 (SAS Institute,
Cary NC).
Semi-structured Interview
Data were collected using in-depth, semi-structured interviews with 3 PCMH and
3 non-PCMH clinics’ leaders (e.g., CEO, CMO and medical directors). The leaders were
asked related to clinic operations and approach to patient care including use of
information technology, involving patients in decision making, disease management,
measurement of quality and access to care. The goal of the interviews was to identify
differences in attributes between PCMH versus non-PCMH clinics and seek the opinion
of clinic leaders on the plausibility of our study findings as well as the potential
differences in clinic attributes that might explain the study results. The detailed survey
questions are presented in appendix 1.
20

Analysis Result
Across the three years included in the study, there were on average 23,662
members in the PCMH group and 138,152 in the non-PCMH group. The distribution of
members’ characteristics, including age categories, gender, and comorbidities, were
similar between PCMH and non-PCMH groups. However, the minor differences were
still statistically significant. This could be an artifact of the very large sample size.
Notably, though, the proportion of Hispanics was 6-7% higher in the non-PCMH group
as compared to the PCMH group (Table 3.3 – 3.5). Similar distribution were found in
clinics with SPD membership less than 10%, and 10% and greater.
Table 3.3
Annual demographics and health resource utilization in all clinics in pre-PCMH period
All Clinics
PCMH
N=22,870
Non-PCMH
N=143,530
Total
N=166,400 p-value
Population Characteristics

Age
a
<0.0001
≤11 11574
(50.6)
71009
(49.5)
82583
(49.6)

12-17 4826
(21.1)
28992
(20.2)
33818
(20.3)

18-29 3301
(14.4)
19138
(13.3)
22439
(13.5)

30-44 2053
(9)
15076
(10.5)
17129
(10.3)

45-64 1116
(4.9)
9315
(6.5)
10431
(6.3)

Female
a
12785
(55.9)
80041
(55.8)
92826
(55.8)
0.6987
Hispanic
a
15228
(66.6)
105666
(73.6)
120894
(72.7)
<0.0001
Mean Comorbidity
b
0.3
(0.37)
0.3
(0.38)
0.3
(0.37)
0.0085
Comorbidity Groups
a,c
<0.0001
1 3658
(16)
24961
(17.4)
28619
(17.2)

2 5475
(23.9)
33554
(23.4)
39029
(23.5)

3 8778
(38.4)
54193
(37.8)
62971
(37.8)

4 4959
(21.7)
30822
(21.5)
35781
(21.5)

a,c
Count (%);
b
Mean (SD); ED: Emergency Dept; PCMH: Patient-Centered Medical Home;
c
Comorbidity Groups were based on the
breakdown of comorbidity scores calculated through Medicaid MRX model. Group 1: 0-0.03; group2: 0.04-0.09; group3:0.1-0.39;
group4: >=0.4
21

Table 3.4
Annual demographics and health resource utilization in clinics with <10% SPD
Membership in pre-PCMH period
PCMH
N=4,676
Non-PCMH
N=62,481
Total
N=67,157 p-value
Population Characteristics

Age
a
<.0001
≤11 2461
(52.6)
32814
(52.5)
35275
(52.5)

12-17 968
(20.7)
13109
(21)
14077
(21)

18-29 646
(13.8)
7617
(12.2)
8263
(12.3)

30-44 380
(8.1)
5318
(8.5)
5698
(8.5)

45-64 221
(4.7)
3623
(5.8)
3844
(5.7)

Female
a
2561
(54.8)
34366
(55)
36927
(55)
0.7571
Hispanic
a
4417
(94.5)
49957
(80)
54374
(81)
<0.0001
Mean Comorbidity
b
0.3
(0.38)
0.3
(0.38)
0.3
(0.38)
0.1993
Comorbidity Groups
a,c
0.6743
1 716
(15.3)
9541
(15.3)
10257
(15.3)

2 1116
(23.9)
14834
(23.7)
15950
(23.8)

3 1841
(39.4)
24233
(38.8)
26074
(38.8)

4 1003
(21.4)
13873
(22.2)
14876
(22.2)

a,c
Count (%);
b
Mean (SD); ED: Emergency Dept; PCMH: Patient-Centered Medical Home;
c
Comorbidity Groups were based on the
breakdown of comorbidity scores calculated through Medicaid MRX model. Group 1: 0-0.03; group2: 0.04-0.09; group3:0.1-0.39;
group4: >=0.4
22

Table 3.5
Annual demographics and health resource utilization in clinics with ≥10% SPD
Membership in pre-PCMH period
PCMH
N=18,194
Non-PCMH
N=81,049
Total
N=99,243 p-value
Population Characteristics

Age
a
<0.0001
≤11 9113
(50.1)
38195
(47.1)
47308
(47.7)

12-17 3858
(21.2)
15883
(19.6)
19741
(19.9)

18-29 2655
(14.6)
11521
(14.2)
14176
(14.3)

30-44 1673
(9.2)
9758
(12)
11431
(11.5)

45-64 895
(4.9)
5692
(7)
6587
(6.6)

Female
a
10224
(56.2)
45675
(56.4)
55899
(56.3)
0.6933
Hispanic
a
10811
(59.4)
55709
(68.7)
66520
(67)
<0.0001
Mean Comorbidity
b
0.3
(0.37)
0.3
(0.38)
0.3
(0.37)
<0.0001
Comorbidity Groups
a,c
<0.0001
1 2942
(16.2)
15420
(19)
18362
(18.5)

2 4359
(24)
18720
(23.1)
23079
(23.3)

3 6937
(38.1)
29960
(37)
36897
(37.2)

4 3956
(21.7)
16949
(20.9)
20905
(21.1)

a,c
Count (%);
b
Mean (SD); ED: Emergency Dept; PCMH: Patient-Centered Medical Home;
c
Comorbidity Groups were based on the
breakdown of comorbidity scores calculated through Medicaid MRX model. Group 1: 0-0.03; group2: 0.04-0.09; group3:0.1-0.39;
group4: >=0.4

Difference-in-difference (DID) analyses were conducted to identify the effect of
PCMH on health care utilization. The DID analysis compared the difference in utilization
between PCMH and non-PCMH clinics in the pre-PCMH compared to the post-PCMH
period. The results presented in both Table 3.3 and 3.4 include the analyses on primary
aim (i.e., the impact of PCMH on all clinics), and the secondary aim (i.e., the response of
PCMH clinics to different proportions of SPD membership).
In the pre-PCMH period, PCMH clinics had 38 fewer ED visits and 26 fewer
avoidable ED visits per thousand members per year (PTMPY) (Table 3.6). Following
23

implementation of PCMH, utilization of ED visits declined much faster in PCMH clinics
compared to non-PCMH clinics. As a result, by 2013 PCMH clinics had 84 fewer ED
visits and 32 fewer avoidable ED visits PTMPY. We find no evidence of differential
trends in use of inpatient hospital care by PCMH status. In contrast to trends in ED visits,
PCMH clinics experienced more rapid increase in use of office visits. In the pre-PCMH
period PCMH clinics had 96 fewer office visits PTMPY (Table 3.7). In 2013, this
differential use of office visits reversed and PCMH clinics had 175 more office visits
PTMPY. Overall the trends in utilization suggest that increased access to primary care in
PCMH clinics might have resulted in less frequent use of the ED.
Parts B and C of Table 3.6 shows the trends in utilization stratified by the
proportion of SPD membership: less than 10% versus 10% and greater. Our hypothesis
was that the influx of SPDs would constrain the capacity of clinics and reduce the effects
of PCMH on access to care and use of ED visits. The results were consistent with this
hypothesis. We see greater increase in office visits and larger declines in ED visits due to
adoption of PCMH in clinics that experienced a smaller influx of SPDs (Table 3.6).
The adjusted difference-in-difference analyses are presented in Table 3.7. The
results were consistent with the unadjusted analysis presented in Table 3.6 We found a
large decrease in ED visits and increase in office visits at PCMH clinics. Also, the effect
of PCMH status on decrease in ED visits was larger in clinics that experienced a smaller
influx of SPDs.
24

Table 3.6
Unadjusted difference in utilization rates between PCMH and non-PCMH groups
Type of Health Care
Utilization PCMH 2011 2012 2013
Part A - All Clinics
Acute
hospitalization
Yes 27.74 (4.12) 27.40 (3.99) 26.35 (3.79)
Diff (PCMH - Non-
PCMH)
2.81 (4.38) 1.66 (4.29) 1.51 (4.10)
DID -1.14 (6.13) -1.30 (6.00)
ED Visit Yes 482.79
(16.49)
467.87 (15.65) 464.95 (15.41)
Diff (PCMH - Non-
PCMH)
-38.21
(17.96)*
-62.20 (17.33)* -84.85 (17.27)*
DID -23.99 (24.96) -46.64 (24.92)
Avoidable ED Visit Yes 128.42 (7.97) 112.76 (6.85) 123.62 (7.44)
Diff (PCMH - Non-
PCMH)
-26.07 (8.85)* -31.20 (7.78)* -32.02 (8.46)*
DID -5.13 (11.78) -5.96 (12.24)
Office Visit Yes 1771.91
(33.24)
1832.25 (33.71) 2058.71 (36.96)
Diff (PCMH - Non-
PCMH)
-96.41
(36.17)*
190.41 (36.07)* 175.57 (40.35)*
DID 286.81 (51.08)* 271.98 (54.19)*
Part B - Clinics with <10% SPD membership
Acute
hospitalization
Yes 24.13 (8.40) 22.32 (7.65) 22.37 (7.88)
Diff (PCMH - Non-
PCMH)
2.43 (8.66) -1.78 (8.07) -0.44 (8.27)
DID -4.21 (11.84) -2.87 (11.97)
ED Visit Yes 415.92
(31.44)
352.90 (26.26) 388.28 (29.79)
Diff (PCMH - Non-
PCMH)
-92.20
(33.15)*
-190.00 (29.08)* -166.76 (32.36)*
DID -97.80 (44.10)* -74.56 (46.33)
Avoidable ED Visit Yes 133.37
(17.68)
97.37 (12.71) 121.37 (16.33)
Diff (PCMH - Non-
PCMH)
-25.27 (18.60) -63.31 (14.27)* -43.07 (17.60)*
DID -38.04 (23.44) -17.80 (25.61)
Office Visit Yes 1884.58
(76.59)
2237.50 (89.51) 3538.12 (145.94)
Diff (PCMH - Non-
PCMH)
4.03 (79.39) 486.74 (92.10)* 1750.90 (147.59)*
DID 482.71 (121.60)* 1746.87 (167.58)*

25

Part C - Clinics with at least 10% SPD membership
Acute
hospitalization
Yes 28.66 (4.63) 28.68 (4.54) 27.26 (4.23)
Diff (PCMH - Non-
PCMH)
1.24 (5.09) 1.99 (4.96) 1.15 (4.68)
DID 0.75 (7.11) -0.09 (6.92)
ED Visit Yes 499.82
(19.15)
496.81 (18.63) 482.43 (17.75)
Diff (PCMH - Non-
PCMH)
-31.09 (21.46) -25.76 (20.80) -64.11 (20.31)*
DID 5.33 (29.89) -33.02 (29.55)
Avoidable ED Visit Yes 127.16 (9.05) 116.63 (8.12) 124.14 (8.45)
Diff (PCMH - Non-
PCMH)
-24.13
(10.41)*
-17.53 (9.25) -26.05 (9.87)*
DID 6.60 (13.92) -1.92 (14.34)
Office Visit Yes 1743.22
(37.32)
1730.27 (36.25) 1721.39 (35.19)
Diff (PCMH - Non-
PCMH)
-115.7
(41.82)*
152.23 (39.46)* -221.29 (40.44)*
DID 267.92 (57.49)* -105.59 (58.18)
*significant at p<0.05; all the results were presented as mean (standard error); DID: Difference-in-
difference; ED: Emergency department; PCMH: Patient-Centered Medical Home.

Table 3.7
Adjusted§ difference in utilization rates between PCMH and non-PCMH groups
Type of Health Care
Utilization PCMH 2011 2012 2013
Part A - All Clinics
Acute
hospitalization
Yes 22.51 (3.29) 20.96 (3.01) 18.60 (2.64)
Diff (PCMH - Non-
PCMH)
4.13 (3.43) 3.40 (3.15) 1.64 (2.80)
DID -0.73 (4.65) -2.49 (4.42)
ED Visit Yes 429.50
(14.64)
420.03 (14.03) 408.26 (13.51)
Diff (PCMH - Non-
PCMH)
-32.40 (15.82) -47.58 (15.37)* -70.32 (15.00)*
DID -15.18 (22.03) -37.92 (21.77)*
Avoidable ED Visit Yes 98.86 (6.11) 88.89 (5.38) 95.64 (5.73)
Diff (PCMH - Non-
PCMH)
-21.82 (6.74)* -23.34 (6.04)* -23.62 (6.44)*
DID -1.52 (9.03) -1.8 (9.31)
Office Visit Yes 1536.83
(28.96)
1600.84 (29.55) 1737.45 (31.88)
Diff (PCMH - Non-
PCMH)
-76.91 (31.10) 242.03 (31.06)* 163.17 (33.72)*
DID 318.94 (43.95)* 240.08 (45.89)*

26

Part B - Clinics with <10% SPD membership
Acute
hospitalization
Yes 22.08 (7.66) 17.47 (6.01) 18.57 (6.58)
Diff (PCMH - Non-
PCMH)
5.42 (7.70) -0.20 (6.14) 2.21 (6.68)
DID -5.62 (9.82) -3.21 (10.14)
ED Visit Yes 394.53
(30.13)
336.76 (25.38) 367.38 (28.45)
Diff (PCMH - Non-
PCMH)
-62.23
(30.97)
-144.14 (27.12)* -120.76 (30.01)*
DID -81.91 (41.16)* -58.53 (43.11)
Avoidable ER Visit Yes 107.75
(14.35)
77.41 (10.21) 95.74 (12.95)
Diff (PCMH - Non-
PCMH)
-16.56
(14.76)
-45.31 (11.09)* -28.44 (13.61)*
DID -28.75 (18.46) -11.88 (20.06)
Office Visit Yes 1564.93
(64.34)
1882.53 (76.33) 2957.14 (123.27)
Diff (PCMH - Non-
PCMH)
-67.32
(65.85)
421.56 (77.32)* 1457.97 (123.55)*
DID 488.88 (101.35)* 1525.29 (139.71)*
Part C - Clinics with at least 10% SPD membership
Acute
hospitalization
Yes 22.55 (3.59) 21.85 (3.41) 18.54 (2.84)
Diff (PCMH - Non-
PCMH)
3.02 (3.84) 4.57 (3.60) 1.40 (3.08)
DID 1.55 (5.25) -1.62 (4.91)
ED Visit Yes 437.09
(16.70)
439.35 (16.44) 416.09 (15.29)
Diff (PCMH - Non-
PCMH)
-28.51
(18.61)
-19.90 (18.21) -55.93 (17.37)*
DID 8.61 (25.99) -27.42 (25.41)
Avoidable ER Visit Yes 96.66 (6.85) 91.67 (6.36) 95.53 (6.51)
Diff (PCMH - Non-
PCMH)
-21.90
(7.87)*
-14.78 (7.19)* -21.09 (7.52)*
DID 7.12 (10.64) 0.81 (10.86)
Office Visit Yes 1533.18
(32.83)
1534.42 (32.13) 1460.98 (30.05)
Diff (PCMH - Non-
PCMH)
-73.86
(36.48)
231.97 (34.47)* -162.35 (34.08)*
DID 305.83 (50.14)* -88.49 (49.89)
§
Adjusted for age, gender, race and comorbidities; *significant at p<0.05; DID: Difference-in-difference;
all the results were presented as expected value (standard error); ED: Emergency department; PCMH:
Patient-Centered Medical Home.

27

Discussion
Medicaid beneficiaries utilize the ED at an almost two-fold higher rate than the
privately insured (Sommers et al., 2012). Safety-net clinics that stand on the front lines to
provide care for the majority of Medicaid and uninsured patients, can play an important
role in reducing the use of ED. This study shows that implementing the PCMH model in
safety-net clinics can have a meaningful impact on reducing the use of ED. However, the
extent to which PCMH can be successful in reducing ED visits may depend on the
capacity of clinics to increase access to primary care. We found the effects of PCMH on
reducing ED visits were smaller in clinics that experienced a greater increase of SPDs.
Previous studies on the impact of the PCMH model have been largely
concentrated in integrated health systems and multi-payer models (Gilfillan et al., 2010;
Grumbach & Grundy, 2010; Reid et al., 2010; Mary Takach, 2012). A few Medicaid only
studies, including those from New York (Priority Community healthcare Center
program), North Carolina (Community Care), Oklahoma, and Vermont, have reported an
improvement in ED visit rates ranging from 7.5% to 31% ("Summary of Patient-Centered
Medical Home Cost and Quality Results, 2010 – 2013," 2013). Our study, which features
a Hispanic dominant population, shows results congruent with these studies. This is an
exciting finding because being able to replicate the favorable findings of the PCMH
model in populations with different cultures and lifestyles indicates that the standards for
PCMH as recognized by the NCQA are effective beyond geographic and population
boundaries.
28

According to Sommers et al, widespread, inappropriate use of the ED amongst
Medicaid beneficiaries can be attributed in part to unmet health needs and lack of access
to appropriate primary care (Sommers et al., 2012). Such finding was supported by our
semi-structured interviews where extended hours, weekend hours and helpline were
available in the PCMH clinics, and only one non-PCMH clinic had extended hours.
Moreover, when we asked about the plausibility of our study finding, the dominant
response was positive. For the same question, the interviewees also stated that access to
care was the key for the reduction of the ED use.
In regards to other attributes, the interview results show that the PCMH clinics
tend to offer more disease management programs. One PCMH clinic also stated that
health information technology was useful in informing and improving decision. There
were no conclusive findings on the attributes related to quality of care and patient
engagement.
In addition to DID results, the PCMH clinics tended to begin with lower ED visit
rates as compared to the non-PCMH clinics in the pre-implementation period. One
possible explanation is that the cut-off date of pre- and post- period is somewhat
arbitrary. Some PCMH clinics engaged in the preparation of PCMH as early as the
beginning of 2011. It was very likely that the training clinics received exerted its effect
before official NCQA recognition. We also cannot rule out the possibility that clinics
selected to participate might have had better infrastructure to serve their patients or had
already implemented some PCMH elements before official recognition. The adoption of
DID study design minimizes the impact from the difference in clinics infrastructure
29

between PCMH and non-PCMH groups, and focus on the differential results from pre-
PCMH to post-PCMH period.
According to 1115 wavier, a group of heavy health service utilizers - seniors and
people with disabilities (SPDs) - was transitioned from fee for service (FFS) Medicaid to
a managed care setting starting from June 2011 to May 2012. This transition created a
unique feature in this study such that we could examine how the PCMH model would
respond to a sudden influx of a new population of heavy utilizers. Although it is arguable
that SPDs might have been served by the same physician from FFS to a MMP, the
eligibility data shows that only less than 4% of SPDs kept prior affiliation and 7-8%
chose their own PCP. That said the majority of SPDs were assigned to a PCP for whom
they did not visit before through an auto-assignment algorithm which considers solely on
the proximity to a clinic, member’s age and primary language.
As expected, the PCMH clinics less impacted by this transition had better results
in reducing both ED and avoidable ED visits; while other PCMH clinics receiving more
than 10% of SPDs had more consistent rates of ED use in the first year of the post-PCMH
period and a minor dip in the second year. Our results, with the stratification on high and
low proportion of SPD membership in PCMH clinics suggest a potential crowding-out
effect, where the introduction of a new population constrains the resource which would
otherwise be allocated to the existing non-SPD Medicaid beneficiaries.
Lacking quality measures was a limitation in this study. As only four of the seven
PCMH safety-net clinics received the assistance from the local managed Medicaid plan,
there was a lack of complete information for analysis on their quality measures. In an
30

effort to account for this limitation, through a personal communication with a
representative from the vendor who conducted coaching and evaluation for these four
clinics, the quality measures did not show any significant improvement after the first year
of implementation. However, the analysis did not take into account the confounding
effect of the SPD transition. In addition, this study only examined the outcomes from
members of a local MMP, not a complete panel served by safety-net clinics. Thus, the
results may not be generalizable to the uninsured or other subset of populations.
Patient centered medical homes have been recognized as an important strategy to
address the primary care crisis in the United States. Our study results from a large urban
Medicaid population suggest that adoption of the patient centered medical home model in
safety-net practices can effectively reduce rates of ED and avoidable ED visits, and
increase the use of office visits among Medicaid patients. We will continue to follow up
with this PCMH pilot cohort and expand our research to focus on the impact to the newly
transitioned SPD members. Our findings support the effectiveness of the PCMH model
that can work across geographic and race/ethnicity boundaries, while also revealing the
potential limitations of the PCMH model in response to a sudden influx of heavy
utilizers.
31

Chapter 4: Turn the Tide of Emergency Department Use in People with Disabilities
Abstract
Objective
The objective of the study was to assess the emergency department (ED) use in
safety-net clinics adopting the patient-centered medical home (PCMH) model in
Medicaid beneficiaries with disabilities.
Study Design and Method
With the launch of 1115 wavier, the transition of Seniors and Persons with
Disabilities (SPD) from Fee-for-Service to Managed Medicaid Plans took place in the
time period from 2011 to 2012 in California. Concurrently, the patient-centered medical
home model was adopted by twelve safety-net clinics which mainly provide primary care
to the medically underserved population in Los Angeles County in California.
We compared ED use between PCMH and non-PCMH clinics. The propensity
score matching was applied to match PCMH and non-PCMH groups based on clinic and
patient characteristics to minimize potential selection bias.
We also conducted a semi-structured interview with the goal to identify
differences in attributes between PCMH versus non-PCMH clinics. This interview
allowed us to seek the opinion of clinic leaders on the plausibility of our study findings as
well as the potential differences in clinic attributes that might explain the study results.
32

Results
After improving the comparability by matching on the characteristics of clinics
and members (cohort 2 and 3), the adjusted odds ratio (OR) of having more than one ED
visit was found 28-33% lower (p-value < 0.05) in the PCMH group than the non-PCMH.
In the subsets of cohort 2 and 3 where only members with at least one office visit were
included, the OR of having at least one ED visit drops around 22-24% (p-value < 0.05)
when comparing the PCMH clinics with the non-PCMH clinics. Similarly, the OR of
having more than one ED visits drops around 40% (p<0.05).
Conclusion
Our finding highlights that the adoption of the PCMH model in safety-net clinics
can effectively reduce ED use. The interview result agrees with this finding and also
points out that improved access to care is the key to bend the growing ED use.
33

Introduction
In 2011, Medicaid beneficiaries enrolled on the grounds of disability composed a
small fraction of the total enrollment: approximately 15% ("Medicaid Facts: Distribution
of Medicaid Enrollees by Enrollment Group," 2011). This small fraction of enrollees
disproportionately impacted spending, representing a whopping 42% of the total
spending Medicaid nationwide ("Medicaid Facts: Medicaid Spending by Enrollment
Group ", 2011). The statistic for the state of California was worse yet, with disabled
beneficiaries representing just 9% of all beneficiaries, nevertheless constituting 41% of
the spending ("Medicaid Facts: Distribution of Medicaid Enrollees by Enrollment
Group," 2011; "Medicaid Facts: Medicaid Spending by Enrollment Group ", 2011). The
trends in spending are hardly news though, as the proportion of expenditures on disabled
beneficiaries has only increased in the last decades (Vladeck, 2003). Naturally, cost
cutting methods have been a topic of interest among providers, payers and policy makers.
California, on the heels of the Patient Protection and Affordable Care Act (ACA), was
one of the first states to plan reform through submission of the Medicaid 1115 waiver in
2010. Notably, one component of this Bridge to Reform is a transition of Seniors and
Persons with Disabilities (SPD) from traditional Fee For Service (FFS) to managed care
in an effort to promote coordinated systems of care more cost efficiently. Another major
component is the support for reform of safety-net facilities (Harbage P, 2012).
Understanding the fundamental characteristics of this SPD population,
specifically those with disabilities, is crucial to understanding the implications of this
transition to managed care. Disabled beneficiaries often present the most complex cases
34

to a health system as compared to other Medicaid beneficiaries and require a wide array
of specialists and specialized facilities ("People with Disabilities and Medicaid Managed
Care: Key Issues to Consider," 2012). They qualify for Medicaid based on a wide variety
of handicaps, including serious mental illness and behavioral health diagnoses,
developmental disorders, severe chronic illness and disabling brain or spinal cord
injuries. Considering their low health and functional status, their proportionally higher
utilization of health services is not entirely unexpected. What’s more is that in all of the
aforementioned spending, there were still unmet needs under FFS. In a survey of
working-age disabled beneficiaries on FFS, more than 38% reported an unmet need,
commonly citing availability, accessibility and language as barriers (Coughlin et al.,
2002). Transitioning such a special needs population to a managed care setting requires a
multifaceted approach that not only serves the needs sought but also addresses the
barriers this population faces in a manner more cost efficient than FFS.
A budding method to manage the needs of SPDs has been the Patient-Centered
Medical Home (PCMH) model. As the name suggests, with each patient at the center of
care, the model strives to provide coordinated, accessible and quality care tailored to
individual needs ("Joint Principles of the Patient-Centered Medical Home," 2007). This
model has been adapted to various health care systems and populations (e.g., private
insurance, Medicare) with mixed results (Cunningham, 2015). Despite the somewhat
variable evidence, 46 states have adopted the PCMH model to enhance Medicaid and/or
CHIP programs as of March 2015 ("Medical homes and patient-centered care," 2015).
Also notable is the adoption of the PCMH model by safety-net clinics with high Medicaid
35

enrollees (Nocon, 2014). Safety-net clinics are composed of providers who deliver a wide
range of services to medically underserved and uninsured populations regardless of their
ability to pay (Saviano & Powers, 2005). Medicaid beneficiaries constitute a significant
proportion of the patients who utilize these safety-net clinics, which only stresses the
importance of having coordinated, effective and efficient methods at these safety-net
clinics (Adashi, Geiger, & Fine, 2010).
With such widespread adoption of the PCMH model through Medicaid programs,
evaluation of its efficacy is critical. Literature is thin on the effectiveness of the PCMH
model in regards generally to the SPD population and specifically to the disabled. Only
recently, one study showed promising results of a care management intervention aligned
with the principles of PCMH on a North Carolina based Medicaid program. Looking
specifically at non-elderly, disabled Medicaid beneficiaries, the program showed
significant savings for this high-risk population, particularly among those with chronic
conditions (Fillmore et al., 2014). Following the lead of the North Carolina example, the
objective of this study is to evaluate the impact of the PCMH model on the utilization
trends of non-elderly, disabled Medicaid beneficiaries enrolled in a Los Angeles
Medicaid Managed Care Plan (MMP).
36

Methods
Study Setting
The SPD transition took place over a 12-month period from June, 2011 to May,
2012. Concurrently, twelve safety-net clinics in Los Angeles County were undergoing
practice transformation into patient-centered medical home and received recognition
from NCQA as PCMH in early 2012 (Figure 1). Together with the implementation of
PCMH and the completion of the transition of SPD, it creates a unique experiment to
examine the impact of PCMH on a population with complex health conditions.
Figure 4.1
The study time line of the comparison of healthcare utilization between PCMH and non-
PCMH clinics

Through a partnership with a local initiative MMP in Los Angeles County, we
were able to access data of non-elderly Medicaid beneficiaries with disabilities from 12
PCMH clinics and 110 non-PCMH clinics respectively. By limiting the study population
to non-elderly and Medicaid only members, we ensured that Medicaid was the sole
source of insurance coverage.
37

One challenge of this analysis was that neither the application to become a PCMH
clinic nor the circumstance that a member who chose to visit a PCMH clinic was a
random process. To account for such selection bias from clinics and members, we
adopted a quasi-experimental study design and applied a series of matching and
stratification schemes to isolate the effect of PCMH in seven different comparison groups
(Figure 2)
Figure 4.2
The study cohort creation flow

A few exclusions were applied when constructing the cohort one. First, patients
who switched between the PCMH group and comparison group during the study period
were excluded (0.2%). Second, at least ten months continuous Medicaid eligibility was
required to prevent compositional change such as the introduction of a new population to
the study cohort. It is also known that the Medicaid population is subject to considerable
movement in and out of Medicaid status. The requirement in continuous eligibility
38

retained members who were reinstated with retroactivity (10.2% members were
excluded).
This study was approved by the Institutional Review Board at the University of
Southern California.
Study Design
The first set of comparison groups (cohort 1) was composed of the non-elderly,
disabled members assigned to safety-net clinics from the local MMP. The second set of
comparison groups (cohort 2) was constructed by identifying the non-PCMH clinics that
have similar propensity scores to the PCMH clinics (described in the following Statistical
Analysis session). In total, there were twelve comparison clinics selected to match twelve
PCMH clinics. (Table 4.1)
Table 4.1
Population Characteristics and Health Resource Utilization Based on Medicaid
Beneficiaries in 2011-PCMH vs. Non-PCMH Clinics after Matching at Clinic Level
PCMH
N=24,284
Non-PCMH
N=12,980
Total
N=37,264

p-value
Population Characteristics
Age, mean (SD), y 15.3 (13.82) 16.6 (14.56) 15.8 (14.09) 0.01
Female, % 55.6 56.1 55.8 0.67
Hispanic, % 64.2 69.1 65.9 0.01
CRG Risk Category, %
1 23.7 23.3 23.5 0.17
2 37.8 37.6 37.7
3 22.3 21.3 22
4 16.2 17.8 16.8
Income Category, %
<$30,000 12.8 4.5 9.9 <0.01
$30,000 - $40,000 46.6 34.1 42.2
$40,000 - $50,000 31.2 21.5 27.8
>$50,000 9.4 39.9 20.1
Health Resource Utilization
Acute hospitalization , % 5 4.7 4.9 0.78
ED Visits, % 0.25
1 18.2 18.6 18.3
2 9.4 9.6 9.5
3+ 13.6 14.2 13.8
Office Visits, mean (SD) 3.2 (3.49) 3.1 (3.24) 3.2 (3.42) 0.17
PCMH: Patient-Centered Medical Home; CRG: Clinical Risk Groups; ED: Emergency Department
39

Cohort 3 was constructed through another round of propensity score matching.
This time matching was done at member level. There were 1,283 members in PCMH and
non-PCMH clinics respectively. Although age, gender, Hispanic and CRG category were
comparable between two groups, unbalanced distribution on the average household
income was found (Table 4.2). A subset of cohort 3, cohort 4 was then created by
limiting the members that have the average household income below $40,000.
Table 4.2
Population Characteristics - PCMH vs. Non-PCMH Clinics Based on Medicaid
Beneficiaries with Disabilities in 2012 after Matching at Member Level
PCMH
N=1,283
Non-PCMH
N=1,283
Total
N=2,566

p-value
Population Characteristics

Age, mean (SD), y
34.3 (19.61) 37.0 (18.46) 35.7 (19.09)
0.74
Female, %
46.8 47.2 47
0.92
Hispanic, %
28.1 30.9 29.5
0.38
CRG Risk Category, %

1
22.9 25.8 24.4
0.73
2
29.1 29.4 29.2

3
25.5 26.2 25.8

4
22.5 18.6 20.6

Income Category, %

<$30,000
4.6 3.6 4.1
<0.01
$30,000 - $40,000
31.5 31.5 31.5

$40,000 - $50,000
45.9 17.4 31.6

>$50,000
18 47.5 32.8

PCMH: Patient-Centered Medical Home; CRG: Clinical Risk Groups

Three additional cohorts - 5, 6 and 7 - were subsets of cohorts 2, 3 and 4
respectively, where only those members who had at least one office visit were
considered. We suspected that members who interacted with their healthcare providers
through office visits might have benefitted from the PCMH model more than those
40

without any office visits. Thus, the impact of PCMH on reducing ED visit was
hypothesized to be greater in cohort 5, 6, and 7 when comparing to cohort 2, 3, and 4.
Data
This study considered administrative claims data from January 1, 2011 through
December 31, 2013 from a local MMP in Los Angeles County. The 12 PCMH clinics
received certification in late 2011 and early 2012. Year 2011 was considered the pre-
PCMH period and after 2011 was considered the post-PCMH period. The non-PCMH
group was composed of 110 safety-net clinics contracted with the same MMP but did not
acquire PCMH recognition from the National Committee for Quality Assurance (NCQA),
Utilization Review Accreditation Committee (URAC), and Joint Commission prior to
2013. Since all members were required to have a primary care physician (PCP), members
assigned to a PCP affiliated with a PCMH clinic were grouped into the PCMH cohort.
Data elements drawn from these databases included: member demographics (age,
gender, race/ethnicity and enrollment history), service dates, episodes (hospital admission
and emergency department visit), ICD-9 diagnosis codes, procedure codes and pharmacy
claims. Gender and Hispanic were each coded as binary variables. The zip code of
member’s primary residence was used to map to 2013 census data (purchased through
zip-codes.com) and derive the average household income.
Hospital admissions, readmission and ED visits were measured using NCQA
standard definitions and served as outcomes of interest in this study ("Healthcare
Effectiveness Data and Information Set," 2013). Excessive ED visits were also
considered and defined in this study as having more than one ED visit in one year.
41

Members’ underlying health conditions were estimated through 3M Clinical Risk
Groups (CRG), a claim-based disease burden model (Murphy, McGready, Griswold, &
Sylvia, 2013). We fit the model with medical and pharmacy claims from the second half
of 2012. Each member was assigned to mutually exclusive and hierarchically ranked risk
groups. Then the groups were further classified into aggregated level 3 CRG (ACRG
level III) groups. Since the cost data was not provided by the MMP, a corresponding
weight calculated based on New York State adult Medicaid program (provided by the 3M
representative) was assigned to the ACRG level III. After examining the distribution of
ACRG weights, we categorized the population into 4 groups of similar membership count
(CRG group 1-4) based on the cutoff points at 0.2, 0.6 and 1.3.
Statistical Analysis
In the process of identifying contemporaneous groups of comparison clinics and
members, we applied propensity score matching method to create cohort 2 and 3
(Rosenbaum & Rubin, 1983). For cohort 2, the baseline data was collected from
Medicaid beneficiaries in 2011 when PCMH had not been implemented yet. The purpose
of using pre-PCMH data was to identify clinics where their performance and patient
profile were comparable. When fitting the logistic regression model, the dependent
variable was the PCMH status, and independent variables included age, gender,
race/ethnicity, average household income, CRG category, inpatient admission (yes/no),
number of ED visits and office visits. A propensity score, which was the conditional
probability of a member assigned to a PCMH clinic was obtained in the regression
output. Then clinic’s propensity score was calculated by averaging the propensity scores
42

of members assigned to the clinic. A comparison clinic was selected when it had the
average propensity score closest to a PCMH clinic. In the end, there were twelve clinics
selected for both PCMH and non-PCMH groups. The propensity score matching process
was detailed in Figure 4.3.
Figure 4.3
Steps of Creating Propensity Score Matching at Clinic Level

Cohort 3 was also constructed through propensity score matching based on the
data from non-elderly and disabled Medicaid beneficiaries in the second half of 2012.
When fitting the model, the dependent variable was PCMH status and independent
variables included age, gender, race/ethnicity, average household income (based on the
zip code), and CRG category. After one to one match at member level, the PCMH and
comparison group ended with 1,283 members in 7 and 5 clinics respectively. The
propensity score matching process was detailed in Figure 4.4.

43

Figure 4.4
Steps of Creating Propensity Score Matching at Member Level

Odds ratios (OR) derived from the logistic regression model were used to describe
the association between PCMH status (binary variable) and various utilization outcomes
(i.e., having at least one ED visit, having at least two ED visits, any acute hospitalization,
and any readmission) in 2013. The model was adjusted for gender, race/ethnicity, average
household income (based on the zip code), and CRG category.
A 2-tailed p-value <0.05 was treated as significant in all statistical tests. All the
data management and analyses conducted using SAS® version 9 (SAS, Cary, NC, USA)
and Stata® version 9 (Stata Corp., College Station, TX, USA).
Semi-structured Interview
Data were collected using in-depth, semi-structured interviews with 3 PCMH and
3 non-PCMH clinics’ leaders (e.g., CEO, CMO and medical directors). The leaders were
asked questions related to clinic operations and approaches to manage patient care
including use of information technology, involving patients in decision making, disease
management, measurement of quality and access to care. The goal of the interviews was
to identify differences in attributes between PCMH versus non-PCMH clinics and to seek
44

the opinion of clinic leaders on the plausibility of our study findings as well as the
potential differences in clinic attributes that might explain the study results. The detailed
survey questions are presented in appendix 1.
45

Analysis Results
Prior to matching, cohort 1 was composed of 2,269 non-elderly Medicaid
beneficiaries with disabilities in the PCMH clinics and 21,897 in the non-PCMH clinics.
In the unadjusted model, having at least two ED visits was 12.58% lower in the PCMH
group as compared to the non-PCMH group. The adjusted difference was even greater,
with a 25% lower odds ratio (OR) in the PCMH group as compared to the non-PCMH.
Having at least two ED visits was 16.67% lower in cohort 2 when comparing the PCMH
with the non-PCMH group. The adjusted OR was 33% lower. Similarly, in cohort 3 the
unadjusted reduction was 16.12% while the adjusted reduction was 28%. (Table 4.3)
When limiting the study population to the zip codes where the average household
income was lower than $40,000, a significant reduction in having at least one ED use
(19.59%) and having at least two ED uses (36.53%) was found in the PCMH group
compared to the non-PCMH group. The adjusted OR was 33% (at least one ED use) and
47% (at least two ED use) lower in the PCMH group compared to the non-PCMH group.
(Table 4.3)
For cohort 5 and 6, the reduction of having at least one ED use when comparing
the PCMH group with the non-PCMH group reached 13-15% with p-value less than 0.05,
while the cohort 2 and 3 showed no difference. Similarly, the OR dropped 22-24% in
cohort 5 and 6, while only around 10% reduction found in cohort 2 and 3. Similar pattern
was found in the comparison of having at least two ED visits. In cohort 7 where patients
need to have at least one office visit and reside in areas with income under $40,000, the
46

adjusted OR had a significant drop around 60% when comparing the PCMH group with
the non-PCMH group.
Table 4.3
Health Resource Utilization – PCMH vs. Non-PCMH in Cohorts 1-7
PCMH Non-PCMH ∆ % ∆ OR* (95%CI)
Cohort 1 N=2,269 N=21,897
ED visits, %
1+ 33.8 33.7 0.1 0.18 0.90 (0.82, 0.99)
2+ 15.6 18.0 -2.4 -12.85 0.75 (0.67, 0.85)
Readmission, % 1.5 1.6 -0.1 -9.01 0.98 (0.68, 1.42)
Acute Hospitalization, % 9.6 9.7 -0.1 -1.59 1.01 (0.87, 1.18)
Cohort 2 N=2,269 N=1,422
ED visits, %
1+ 33.8 34.2 -0.5 -1.43 0.91 (0.78, 1.06)
2+ 15.6 18.8 -3.1 -16.67 0.67 (0.55, 0.81)
Readmission, % 1.5 1.7 -0.2 -13.83 0.82 (0.45, 1.49)
Acute Hospitalization, % 9.6 9.8 -0.2 -2.16 0.92 (0.71, 1.19)
Cohort 3 N=1,283 N= 1,283
ED visits, %
1+ 33.7 34.7 -1.0 -2.88 0.93 (0.79, 1.11)
2+ 15.8 18.9 -3.0 -16.12 0.72 (0.58, 0.90)
Readmission, % 1.7 1.8 -0.1 -4.35 0.88 (0.48, 1.63)
Acute Hospitalization, % 10.4 9.6 0.8 8.13 1.02 (0.77, 1.35)
Cohort 4 N=463 N=450
ED visits, %
1+ 30.0 37.3 -7.3 -19.59 0.67 (0.50, 0.89)
2+ 13.8 21.8 -8.0 -36.53 0.53 (0.37, 0.77)
Readmission, % 2.0 2.7 -0.7 -25.93 0.73 (0.26, 2.02)
Acute Hospitalization, % 13.6 14.2 -0.6 -4.23 0.89 (0.58, 1.39)
Cohort 5 (Having at least one office visit) N=1,576 N=846
ED visits, %
1+ 41.1 47.9 -6.8 -14.11 0.78 (0.65, 0.94 )
2+ 20.6 28.4 -7.8 -27.53 0.60 (0.48, 0.74 )
Readmission, % 2.1 2.8 -0.7 -26.19 0.78 (0.43, 1.43 )
Acute Hospitalization, % 13.7 16.4 -2.7 -16.58 0.84 (0.64, 1.09 )
Cohort 6 (Having at least one office visit) N=870 N=748
ED visits, %
1+ 43.0 49.5 -6.5 -13.09 0.76 (0.62, 0.93 )
2+ 20.9 28.9 -8.0 -27.56 0.62 (0.49, 0.79 )
Readmission, % 2.5 3.1 -0.5 -17.76 0.83 (0.45, 1.54 )
Acute Hospitalization, % 15.2 16.4 -1.3 -7.73 0.90 (0.68, 1.20 )
Cohort 7 (Having at least one office visit) N=308 N=257
ED visits, %
1+ 37.0 54.1 -17.1 -31.57 0.48 (0.34, 0.69 )
2+ 17.2 34.2 -17.0 -49.75 0.40 (0.27, 0.59 )
Readmission, % 2.3 3.5 -1.20 -35.10 0.65 (0.24, 1.82 )
Acute Hospitalization, % 16.2 19.5 -3.20 -16.56 0.76 (0.49, 1.20 )
PCMH: Patient-Centered Medical Home; ED: Emergency Department;
*OR: Odds Ratio adjusted for age, gender, race/ethnicity, average household income and comorbidity

47

Discussion
Prior studies have shown that Medi-Cal program is associated with frequent
hospital admissions and heavy reliance on the emergency department (ED) in comparison
to commercially insured patients (McConville & Lee, 2008). A recent report shows that
the ED rate increased from 572 to 651 visits per 1000 enrollees from 2005 to 2010. This
is four times higher than the privately insured, and 2.5 times higher than the uninsured. In
addition, Medicaid patients consistently had the highest rate of visits for potentially
preventable conditions (Hsia, Brownell, Wilson, Gordon, & Baker, 2013). Naturally, it is
expected to be higher in a disabled population.
Our result shows that safety-net clinics operating under a PCMH model avoided
ED use by 30% to 50% in people with disabilities. Such an encouraging finding indicates
that the PCMH model can be an effective strategy to reduce ED use, particularly for the
frequent ED users.
The PCMH model has been shown some promising results in managing health
care utilizations in a wide range of populations (e.g., commercially insured people, the
elderly, patients with chronic conditions and children) (Nutting et al., 2009; Raskas et al.,
2012; Reid et al., 2010; Rosenberg, Peele, Keyser, McAnallen, & Holder, 2012;
Rosenthal et al., 2013; Silow-Carroll, Edwards, & Rodin, 2013; Mary Takach, 2012).
People with disabilities, who tend to have higher demand on health care service,
however, has not been extensively studied or reported on. The only related literature we
found was the Community Care of North Carolina’s (CCNC) experience, a program that
was supported by the state government and has medical home as one of the modules to
48

manage its Medicaid beneficiaries (Fillmore et al., 2014; Rich, Lipson, Libersky, Peikes,
& Parchman, 2012). Such a program was not applicable to states like California, where
insurers delegate responsibilities, such as provider credentialing, utilization management,
and chronic disease management, to a group of physicians or an independent practice
association (IPA). Therefore, our study is unique to a health care environment where a
delegated model is adopted.
This encouraging finding prompted us to turn to experts for assistance in its
interpretation. We accomplished this through a semi-structured interview with clinic
leaders. The purposes of these interviews were twofold: Firstly, to evaluate the
plausibility of our findings and secondly to understand the key differences between the
PCMH and non-PCMH clinics in regards to the availability of the PCMH attributes.
Among all the responses, “access to care” was most frequently cited by experts. By
operating with extended office hours, including weekends, and a helpline available to
their patients, the PCMH clinics offer accessibility to those who would otherwise not
have access to care.
Additionally, PCMH clinics often offer broader disease management programs
that cover more chronic conditions than non-PCMH clinics. One PCMH clinic also stated
that health information technology was useful in informing and improving decision.
There were no definitive expert opinions on the attributes related quality of care and
patient engagement. Ultimately this points to the importance of access to care.
There are a few limitations in this study. Although the potential selection bias at
clinic and member level were mitigated through a series of matching schemes, the
49

selection of comparable clinics was indirectly identified through the outcomes from their
patients at the pre-PCMH period. As a result, the comparison clinics might not be similar
in size, infrastructure, and/or capacity. Also, the study population was limited to people
with disabilities, these results may not be generalizable to the uninsured or other non-
disabled populations. However, we are in the process of preparing a draft where
reduction in ED use was found in non-disabled Medicaid beneficiaries after the
implementation of the PCMH.
People with disabilities are an understudied population and require more
healthcare resource than most other people. With a focus on non-elderly Medicaid
beneficiaries with disabilities, our study suggests that the adoption of the PCMH model in
safety-net clinics can effectively reduce the ED use. Furthermore, our survey result
highlights that improved access to care is the key to reduce ED use.
50

Chapter 5: Conclusions
Medicaid accounts for 16 cents of every U.S. healthcare dollar, 24 cents of every
State budget dollar. Of the 60 million Medicaid beneficiaries in the U.S., Seniors and
Persons with Disabilities (SPD) who represent just one-fourth of program enrollees yet
account for 70% of overall Medicaid cost. As California (Medi-Cal) is the largest
Medicaid program in the nation with total 12 million beneficiaries as of Jan 2015, almost
one in three Californians are in the Medi-Cal program, seeking solutions to manage and
ensure the efficiency of care delivery has an extensive impact to the State’s budget and
resource allocation.
As a result of the launch of Medicaid expansion in 2014, the already high ED use
in the Medi-Cal population is expected to grow faster than other types of health insurance
programs. How to bend the curve of ED use, and how to ensure the quality of care to the
staggering number of Medi-Cal beneficiaries become two urgent questions to the policy
makers, care providers and all Medi-Cal enrollees.
Patient-centered medical home is a model of care where each patient has an
ongoing relationship with a personal physician who leads a team that takes collective
responsibility for patient care. By integrating patient engagement, coordination of care,
quality of care, adopting health information technology and access to care, encouraging
results in improving the quality of patient care, reducing hospitalization and ED visits
were presented in some pilot projects mainly from integrated delivery systems and multi-
payer sponsored PCMH initiatives. As the evidence is thin for the Medicaid population,
especially for the people with disabilities, this dissertation presents timely and essential
51

evaluation of the potential of the PCMH model in guarding against the explosion of
Medi-Cal enrollment.
Safety-net clinics which stand on the front lines to take care of the majority of
Medicaid and uninsured population have an indispensable role in the success of a
Medicaid program. This dissertation with a special focus on safety-net clinics examined
the effectiveness of the PCMH model in managing healthcare utilization in two separate
populations – nonelderly and nondisabled Medicaid beneficiaries, and nonelderly
Medicaid beneficiaries with disabilities.
In the first part of the dissertation, the result shows that implementing the PCMH
model in safety-net clinics can have a meaningful impact on reducing ED use in
nonelderly and nondisabled Medicaid beneficiaries. However, the extent to which PCMH
can be successful in reducing ED visits will depend on the capacity of clinics to increase
access to primary care. Another finding from objective one shows that the effect of
PCMH on reducing the use of ED is smaller in clinics that experienced an increase in
patients that were senior or disabled. In the second part of the dissertation, the result
shows that safety-net clinics operating under a PCMH model avoided ED use by 50% in
nonelderly Medicaid beneficiaries with disabilities when patients had at least one office
visit per year.
Together with both findings, this study based on a local Managed Medicaid Plan
suggests that adoption of the PCMH model in safety-net practices can effectively reduce
ED use and avoidable ED use, and increase office visits among Medicaid patients. The
semi-structured interview result demonstrated the validity of the quantitative findings
52

from data analysis and at the same time highlighted the importance of access to care in
regards to the success of the PCMH model.
53

References
Adashi, E. Y., Geiger, H. J., & Fine, M. D. (2010). Health care reform and primary care--
the growing importance of the community health center. N Engl J Med, 362(22),
2047-2050. doi: 10.1056/NEJMp1003729
Bindman, A. B., Grumbach, K., Bernheim, S., Vranizan, K., & Cousineau, M. (2000).
Medicaid managed care's impact on safety-net clinics in California. Health Aff
(Millwood), 19(1), 194-202.
Coughlin, T. A., Long, S. K., & Kendall, S. (2002). Health care access, use, and
satisfaction among disabled Medicaid beneficiaries. Health Care Financ Rev,
24(2), 115-136.
Cunningham, P. J. (2015). Many Medicaid Beneficiaries Receive Care Consistent With
Attributes Of Patient-Centered Medical Homes. Health Aff (Millwood), 34(7),
1105-1112. doi: 10.1377/hlthaff.2015.0141
Fillmore, H., DuBard, C. A., Ritter, G. A., & Jackson, C. T. (2014). Health care savings
with the patient-centered medical home: Community Care of North Carolina's
experience. Popul Health Manag, 17(3), 141-148. doi: 10.1089/pop.2013.0055
Friedberg, M. W., Lai, D. J., Hussey, P. S., & Schneider, E. C. (2009). A guide to the
medical home as a practice-level intervention. Am J Manag Care, 15(10 Suppl),
S291-299.
Gilfillan, R. J., Tomcavage, J., Rosenthal, M. B., Davis, D. E., Graham, J., Roy, J. A., . . .
Steele, G. D., Jr. (2010). Value and the medical home: effects of transformed
primary care. Am J Manag Care, 16(8), 607-614.
54

Gilmer, T., Kronick, R., Fishman, P., & Ganiats, T. G. (2001). The Medicaid Rx model:
pharmacy-based risk adjustment for public programs. Med Care, 39(11), 1188-
1202.
Graham, C. L., Kurtovich, E., Ivey, S. L., & Neuhauser, L. (2011). Fee-for-service and
managed care for seniors and people with disabilities on Medicaid: implications
for the managed care mandate in California. J Health Care Poor Underserved,
22(4), 1413-1423. doi: 10.1353/hpu.2011.0141
Grumbach, K., & Grundy, P. (2010). Outcomes of implementing patient centered medical
home interventions. Washington, DC: Patient-Centered Primary Care
Collaborative.
Harbage P, K. M. (2012). A Bridge to Reform: California’s Medicaid Section 1115
Waiver. Retrieved 09/01/2015, 2015, from
http://www.chcf.org/~/media/MEDIA%20LIBRARY%20Files/PDF/B/PDF%20B
ridgeToReform1115Waiver.pdf
Healthcare Effectiveness Data and Information Set. (2013). Retrieved January 11, 2014,
from http://www.ncqa.org/HEDISQualityMeasurement.aspx
Hsia, R. Y., Brownell, J., Wilson, S., Gordon, N., & Baker, L. C. (2013). Trends in adult
emergency department visits in California by insurance status, 2005-2010. JAMA,
310(11), 1181-1183. doi: 10.1001/jama.2013.228331
Joint Principles of the Patient-Centered Medical Home. (2007). Retrieved 08/01, 2015,
from
https://www.acponline.org/running_practice/delivery_and_payment_models/pcm
55

h/demonstrations/jointprinc_05_17.pdf
McConville, S., & Lee, H. (2008). Emergency department care in California. Public
Policy Institute of California, 10(1), 1-23.
Medicaid Facts: Distribution of Medicaid Enrollees by Enrollment Group. (2011).
Distribution of Medicaid Enrollees by Enrollment Group. Retrieved 10/04/2015,
2015, from http://kff.org/medicaid/state-indicator/distribution-of-medicaid-
enrollees-by-enrollment-group/
Medicaid Facts: Medicaid Spending by Enrollment Group (2011). Retrieved
10/04/2015, 2015, from http://kff.org/medicaid/state-indicator/medicaid-
spending-by-enrollment-group/
Medical homes and patient-centered care. (2015). Retrieved 08/03, 2015, from
http://www.nashp.org/med-home-map
Murphy, S. M., McGready, J., Griswold, M. E., & Sylvia, M. L. (2013). A method for
estimating cost savings for population health management programs. Health Serv
Res, 48(2 Pt 1), 582-602. doi: 10.1111/j.1475-6773.2012.01457.x
Nocon, R. (2014). Association between primary care provider medical home
characteristics and emergency department visits and hospitalizations among
Medicaid patients. Paper presented at the 142nd APHA Annual Meeting and
Exposition 2014.
Nutting, P. A., Miller, W. L., Crabtree, B. F., Jaen, C. R., Stewart, E. E., & Stange, K. C.
(2009). Initial lessons from the first national demonstration project on practice
transformation to a patient-centered medical home. Ann Fam Med, 7(3), 254-260.
56

doi: 10.1370/afm.1002
Patient-Centered Medical Home Recognition. (2014). Retrieved July 2, 2014, from
http://www.ncqa.org/Programs/Recognition/Practices/PatientCenteredMedicalHo
mePCMH.aspx
People with Disabilities and Medicaid Managed Care: Key Issues to Consider. (2012).
Retrieved 08/29, 2015, from
https://kaiserfamilyfoundation.files.wordpress.com/2013/01/8278.pdf
Raskas, R. S., Latts, L. M., Hummel, J. R., Wenners, D., Levine, H., & Nussbaum, S. R.
(2012). Early Results Show WellPoint’s Patient-Centered Medical Home Pilots
Have Met Some Goals For Costs, Utilization, And Quality. Health Aff
(Millwood), 31(9), 2002-2009. doi: 10.1377/hlthaff.2012.0364
Reid, R. J., Coleman, K., Johnson, E. A., Fishman, P. A., Hsu, C., Soman, M. P., . . .
Larson, E. B. (2010). The Group Health medical home at year two: cost savings,
higher patient satisfaction, and less burnout for providers. Health Aff (Millwood),
29(5), 835-843. doi: 10.1377/hlthaff.2010.0158
Rich, E. C., Lipson, D., Libersky, J., Peikes, D. N., & Parchman, M. L. (2012).
Organizing care for complex patients in the patient-centered medical home. Ann
Fam Med, 10(1), 60-62. doi: 10.1370/afm.1351
Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in
observational studies for causal effects. Biometrika, 70(1), 41-55.
Rosenberg, C. N., Peele, P., Keyser, D., McAnallen, S., & Holder, D. (2012). Results
From A Patient-Centered Medical Home Pilot At UPMC Health Plan Hold
57

Lessons For Broader Adoption Of The Model. Health Aff (Millwood), 31(11),
2423-2431. doi: 10.1377/hlthaff.2011.1002
Rosenthal, M. B., Friedberg, M. W., Singer, S. J., Eastman, D., Li, Z., & Schneider, E. C.
(2013). Effect of a multipayer patient-centered medical home on health care
utilization and quality: the Rhode Island chronic care sustainability initiative pilot
program. JAMA Intern Med, 173(20), 1907-1913. doi:
10.1001/jamainternmed.2013.10063
Saviano, E. C., & Powers, M. (2005). California's safety-net clinics: A primer: California
HealthCare Foundation.
Silow-Carroll, S., Edwards, J. N., & Rodin, D. (2013). How Colorado, Minnesota, and
Vermont are reforming care delivery and payment to improve health and lower
costs. Issue Brief (Commonw Fund), 10, 1-9.
Sommers, A. S., Boukus, E. R., & Carrier, E. (2012). Dispelling myths about emergency
department use: majority of Medicaid visits are for urgent or more serious
symptoms. Res Brief(23), 1-10, 11-13.
Stange, K. C., Nutting, P. A., Miller, W. L., Jaen, C. R., Crabtree, B. F., Flocke, S. A., &
Gill, J. M. (2010). Defining and measuring the patient-centered medical home. J
Gen Intern Med, 25(6), 601-612. doi: 10.1007/s11606-010-1291-3
. Statewide Collaborative Quality Improvement Project Reducing Avoidable Emergency
Room Visits (D. o. H. C. services, Trans.). (2012) (pp. 11): California Department
of Health Care services
. Summary of Patient-Centered Medical Home Cost and Quality Results, 2010 – 2013.
58

(2013). The Patient-Centered Primary Care Collaborative
Takach, M. (2011). Reinventing Medicaid: state innovations to qualify and pay for
patient-centered medical homes show promising results. Health Aff (Millwood),
30(7), 1325-1334. doi: 10.1377/hlthaff.2011.0170
Takach, M. (2012). About Half Of The States Are Implementing Patient-Centered
Medical Homes For Their Medicaid Populations. Health Aff (Millwood), 31(11),
2432-2440. doi: 10.1377/hlthaff.2012.0447
Vladeck, B. C. (2003). Where the action really is: Medicaid and the disabled. Health Aff
(Millwood), 22(1), 90-100.
Zeger, S. L., & Liang, K. Y. (1986). Longitudinal data analysis for discrete and
continuous outcomes. Biometrics, 42(1), 121-130.
59

Appendix
Appendix 1. Semi-Structured Interview Questionnaire
Interview Date: ______________________________
Name of Interviewee: _________________________
I am a researcher from the Sol Price School of Policy at USC. I am calling today
regarding a research project that will be presented in a health policy forum or for
publication in a health policy journal. May I speak with you briefly to complete an
8 question survey?
The purpose of the project is to better identify key drivers of a successful care
model based on collected feedback from experts in the field.
Your responses are voluntary and will be confidential. Responses will not be
identified by the individual, but rather compiled together and analyzed as a
group.

1. Does your clinic involve patients in decisions about their health care? If yes, can you
describe the practices or protocols you use?

2. Does your clinic use clinical decision support tools? If yes, please describe the tools.

3. Does your clinic monitor quality of care? If yes, what metrics do you use? Are there
incentives tied to meeting quality thresholds?

4. Does your clinic have any panel management or disease management programs for
all patients or high risk patients? If so, please describe.

5. Does your clinic have electronic health records? (Is the EHR portable across other
providers?) Is it helpful in improving decisions?

6. What are your regular hours? Do you have a helpline or extended hours? Do you
use email or phone consultation? Who is paying for the extended hours?

7. Can you describe who is in the team of health care providers who routinely take care
of patients in the clinic?

8. My research shows that Patient-Centered Medical Home clinics have lower ER use.
Do you think this is plausible result?
If yes, what are the key drivers of lower ER use?

Abstract (if available)

Abstract The California Medicaid program, also known as Medi-Cal is the largest Medicaid program in the nation. With the rollout of Medicaid expansion, the total Medi-Cal beneficiaries have reached about 12 million, constituting nearly 30% of the state's population. Knowing that the resource and capacity at the provider side can hardly keep up with the growth of the Medi-Cal population, the identification of a model of care to direct patients to the proper setting for care has emerged as a top priority for the state’s health policy. ❧ Safety-net clinics play a pivotal role in delivering both primary and specialty care to millions of low-income people, and yet we know little about their performance under different health care delivery models. With the implementation of Patient-Centered Medical Home (PCMH) in early 2012, where patient engagement, health information technology, coordination of care, quality of care and access to care was integrated into clinics’ daily practice, it would be interesting to examine the impact of this model on the healthcare use of Medi-Cal beneficiaries. ❧ This dissertation included two main objectives. The first objective evaluated the impact of PCMH on non-disabled Medi-Cal beneficiaries. The analysis shows that among clinics with less than 10% Seniors and Persons with Disabilities (SPD) membership, transformation to PCMH was associated with increased use of office visits and reduced use of emergency departments (ED). In particular, PCMH clinics (relative to non-PCMH clinics) reduced ED visits by an average 70 visits per thousand members per year (PTMPY) and avoidable ED visits by 20 visits PTMPY. No significant change in office visits or reduction in ED use was found in clinics with SPD membership greater than 10% suggesting that the beneficial effects of the PCMH model in safety net clinics can be muted by a sudden influx of heavy utilizers. ❧ The second objective evaluated the impact of PCMH on Medi-Cal beneficiaries with disabilities. The finding shows that among patients who had at least one office visit in a year, the odds ratio (OR) of having at least one ED use drops around 22-24% (p-value < 0.05) when comparing PCMH clinics with non-PCMH clinics. Similarly, the OR of having at least two ED visit drops around 40% (p<0.05). ❧ Both findings led to the same conclusion that the adoption of the PCMH model in safety-net clinics can effectively reduce ED use. Improved access to care was recognized as the key attribute of the success of the PCMH model by the leaders from safety-net clinics.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Delivering better care for children with special health care needs: analyses of patient-centered medical home and types of insurance

PDF

Effects of a formulary expansion on the use of atypical antipsychotics and health care services by patients with schizophrenia in the California Medicaid Program

PDF

Discriminating changes in health using patient-reported outcomes

PDF

The impact of treatment decisions and adherence on outcomes in small hereditary disease populations

PDF

Burden of illness in hemophilia A: taking the patient’s perspective

PDF

Impact of pharmacy-based transitional care on healthcare utilization and costs

PDF

Understanding primary nonadherence to medications and its associated healthcare outcomes: a retrospective analysis of electronic medical records in an integrated healthcare setting

PDF

An evaluation of risk predictors for disease progression in Veteran Affairs patients with chronic hepatitis C

PDF

Essays on the use of microsimulation for health and economic policy analysis

PDF

Economic aspects of obesity

PDF

A series of longitudinal analyses of patient reported outcomes to further the understanding of care-management of comorbid diabetes and depression in a safety-net healthcare system

PDF

Three essays on estimating the effects of government programs and policies on health care among disadvantaged population

PDF

Total systems engineering evaluation of invasive pediatric medical therapies conducted in non-clinical environments

PDF

Assessment of the impact of second-generation antipscyhotics in Medi-Cal patients with bipolar disorder using panel data fixed effect models

PDF

Three essays on comparative economic and clinical outcomes of antipsychotics using retrospective claims data

PDF

Investigation of health system performance: effects of integrated triple element method of high reliability, patient safety, and care coordination

PDF

Essays in pharmaceutical and health economics

PDF

The causal-effect of childhood obesity on asthma in young and adolescent children

PDF

Economic, clinical, and behavioral outcomes from medical and pharmaceutical treatments

PDF

Three essays on emerging issues in hemophilia care

Asset Metadata

Creator Chu, Li-Hao (author)

Core Title The impact of Patient-Centered Medical Home on a managed Medicaid plan

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Pharmaceutical Economics and Policy

Publication Date 02/08/2016

Defense Date 10/22/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Medicaid,OAI-PMH Harvest,Patient Centered Medical Home,safety-net clinics

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sood, Neeraj (committee chair), Joyce, Geoffrey (committee member), Sayles, Jennifer N. (committee member)

Creator Email chulihao@gmail.com,lihaochu@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-205815

Unique identifier UC11277419

Identifier etd-ChuLiHao-4084.pdf (filename),usctheses-c40-205815 (legacy record id)

Legacy Identifier etd-ChuLiHao-4084.pdf

Dmrecord 205815

Document Type Dissertation

Format application/pdf (imt)

Rights Chu, Li-Hao

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA