University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1

IMPACT OF PHARMACY-BASED TRANSITIONAL CARE ON
HEALTHCARE UTILIZATION AND COSTS

BY

WEIYI NI

A Dissertation Presented to The
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree of
DOCTOR OF PHILOSOPHY
(PHARMACEUTICAL ECONOMICS AND POLICY)

May 2017

2

DEDICATION
This dissertation is dedicated to my parents for their love, endless support and
encouragement. Also, this thesis is dedicated to my wife, Fang, who has been a great source of
motivation and inspiration.

3

ACKNOWLEDGMENTS
I would never have been able to finish my dissertation without the guidance of research committee
members, help from friends, and support from my family.

First and foremost, I would like to express my gratitude to my dissertation advisor, Dr. Jeffrey
McCombs for his supervision, advice, and guidance throughout my graduate study. Above all and
the most needed, Dr. McCombs provided me unflinching encouragement and support in various
ways.

I would like to thank all the members in my research committee, Dr. John Romley, Dr. Naomi
Florea, Dr. Jason Doctor, and Dr. Steven Fox. Their helpful insights, comments, and suggestions
truly shaped my research work.

I offer my thanks to all the current and former collaborators in my dissertation project. I am grateful
to Dr. Danielle Colayco, Dr. Jonathan Hashimoto, and Dr. Kevin Komoto for providing with me
ideas and assistance. My thanks are extended to all my colleagues and friends at USC. They made
my journey at USC enjoyable and delightful.

I would like to sincerely thank my parents for all their endless love and encouragement. They were
always supporting me and encouraging me with their best wishes.

Finally, I would like to thank my beloved wife, Fang Meng. She was always there cheering me up
and stood by me through the good times and bad.
4

TABLE OF CONTAINS
Acknowledgments ......................................................................................................................................... 3
Chapter 1 Introduction .................................................................................................................................. 8
References ............................................................................................................................................... 10
Chapter 2 Impact of a Pharmacy-Based Transitional Care Program on Hospital Readmissions ................ 12
Abstract ................................................................................................................................................... 12
2.1 Introduction ....................................................................................................................................... 14
2.2 Methods............................................................................................................................................. 15
2.2.1 Ambulatory Care Pharmacy-Based Transitional Care Program ................................................ 15
2.2.2 Intervention ................................................................................................................................ 16
2.2.3. Data Collection ......................................................................................................................... 16
2.2.4 Study Population ........................................................................................................................ 17
2.2.5 Outcome Measures ..................................................................................................................... 18
2.2.6 Statistical Analysis ..................................................................................................................... 18
2.3 Results ............................................................................................................................................... 19
2.4 Discussion ......................................................................................................................................... 21
2.4.1 Limitations ................................................................................................................................. 23
2.5 Conclusions ....................................................................................................................................... 24
2.6 References ......................................................................................................................................... 24
2.7 Tables ................................................................................................................................................ 28
2.8 Figures .............................................................................................................................................. 38
Chapter 3 Reducing Healthcare Costs through a Community Pharmacy-Based Transitional Care
Program ....................................................................................................................................................... 39
Abstract ................................................................................................................................................... 39
3.1 Introduction ....................................................................................................................................... 42
3.2 Research Methods ............................................................................................................................. 43
3.2.1 Study Design .............................................................................................................................. 43
3.2.2 Data ............................................................................................................................................ 43
3.2.3 Pharmacist-Provided Transitional Care Program ....................................................................... 43
3.2.4 Study Populations ...................................................................................................................... 44
3.2.5 Outcome Measures ..................................................................................................................... 45
3.2.6 Statistical Methods ..................................................................................................................... 45
3.3 Results ............................................................................................................................................... 47
5

3.4 Discussion ......................................................................................................................................... 49
3.5 Conclusion ........................................................................................................................................ 52
3.6 References ......................................................................................................................................... 52
3.7 Tables ................................................................................................................................................ 55
3.8 Figures .............................................................................................................................................. 59
Chapter 4 Budget Impact Analysis on Pharmacist-Provided Transitional Care Program .......................... 61
Abstract ................................................................................................................................................... 61
4.1 Introduction ....................................................................................................................................... 63
4.2 Methods............................................................................................................................................. 64
4.2.1 Interventions .............................................................................................................................. 64
4.2.2 Budget Impact Model................................................................................................................. 65
4.2.3 Data Inputs ................................................................................................................................. 66
4.2.4 Sensitivity Analyses ................................................................................................................... 67
4.3 Results ............................................................................................................................................... 68
4.4 Discussion ......................................................................................................................................... 70
4.5 Conclusions ....................................................................................................................................... 73
4.6 References ......................................................................................................................................... 73
4.7 Tables ................................................................................................................................................ 77
4.8 Figures .............................................................................................................................................. 80
4.9 Appendix - Data Analysis for Sensitivity Analytic Input Parameters .............................................. 82
4.9.1 Data Collection .......................................................................................................................... 82
4.9.2 Analytic Population.................................................................................................................... 82
4.9.3 Outcome Measures ..................................................................................................................... 83
4.9.4 Statistical Analysis ..................................................................................................................... 83
4.9.5 Results ........................................................................................................................................ 84
Chapter 5 Summary and Future Directions ................................................................................................. 85
References ............................................................................................................................................... 89

6

LIST OF TABLES
Table 2-1. Descriptive Statistics of ITT Population.................................................................................... 28
Table 2-2. Logistic Regression Analysis for Effect of TOC Intervention on 30-Day Readmissions, IIT
Population ................................................................................................................................................... 29
Table 2-3. Logistic Regression Analysis for Effect of TOC Intervention on 180-Day Readmissions, IIT
Population ................................................................................................................................................... 32
Table 2-4. Cox Proportional Hazards Analysis on Time-to-Readmission .................................................. 35
Table 3-1. Descriptive Statistics of 30-Day ITT Population ....................................................................... 55
Table 3-2. Descriptive Statistics of 180-Day ITT Population ..................................................................... 56
Table 3-3. Effects of TOC Program on Healthcare Costs, OLS and GLM results ..................................... 57
Table 3-4. Modified Park Test Results ....................................................................................................... 58
Table 3-5. Effects of TOC Program on Healthcare Costs, TPM results ..................................................... 58
Table 4-1. Budget Model Input Parameters ................................................................................................ 77
Table 4-2. Budget Impact Analysis on Total Healthcare Costs .................................................................. 78
Table 4-3. Budget Impact Analysis on Inpatient Costs............................................................................... 78
Table 4-4. Sensitivity Analysis on Budget Impact of Inpatient Costs ........................................................ 79
Appendix Table 4-1. Impact of TOC Program on Hospitalizations ........................................................... 84

7

LIST OF FIGURES
Figure 2-1. Study Population Selection Intervention group........................................................................ 38
Figure 3-1. Flow Chart ................................................................................................................................ 59
Figure 3-2. Patient Disposition ................................................................................................................... 60
Figure 4-1. Budget Impact Model ............................................................................................................... 80
Figure 4-2. One-way Sensitivity Analysis of Budget Impact Model .......................................................... 81

8

CHAPTER 1
Introduction

Among many interventions proposed to control healthcare utilization and costs, the topics of
current proposal are to evaluate the impact of a pharmacist-based transitional care on healthcare
utilizations using econometric models.

Suboptimal medication therapy during the transitional care period following hospital discharge is
a major contributory factor to hospital readmissions and increased healthcare utilization [1].
Forster et al. estimated that 11% of patients experienced an adverse drug event after discharge
from inpatient services [2].Twenty-seven percent of readmissions were considered to be
preventable if the patient had received appropriate post-discharge medication monitoring [2].

Several types of services have been shown to impact hospital readmissions, including patient
education, medication adherence counseling, and medication reconciliation [3-9]. Pharmacist
involvement in discharge counseling, medication reconciliation, and telephone follow-up has
resulted in a lower incidence of preventable adverse drug events [5-8]. Consequently, it is
reasonable to expect that pharmacist-led transitional care services may decrease readmission rates
as well as healthcare costs.

Synergy Pharmacy Solutions (SPS) in Bakersfield, CA initiated an ambulatory care pharmacy-
based transitional care service in 2013 for recently discharged members of Kern Health Systems
(KHS) managed Medicaid health plan which were classified as high risk based on their healthcare
9

utilization and medical history. High-risk members admitted to a single local hospital were referred
to SPS for post discharge services. Over 1,100 members were referred to SPS between April 2013
and March 2015.

The first part of this study is to evaluate the effect of the transition of care (TOC) services on 30-
day and 180-day readmissions. The intervention group of TOC patients was compared against a
control group of matched KHS members discharged from neighboring hospitals. Logistic
regressions controlling for demographics, comorbidities, and prior healthcare utilization were
performed. A Cox proportional hazards model was used to analyze the impact on time to
readmission.

The second part of this project focuses on the impact of the TOC services on healthcare costs
including inpatient, outpatient, medication, emergency room(ER), and total healthcare costs.
Ordinary least squares (OLS), generalized linear model (GLM), and two part model (TPM) were
applied to estimate the marginal effects of the TOC program on these healthcare costs.

The third part of this study is to create a budget impact model to simulate the economic impact of
the pharmacy-based transitional care program. Currently, this transitional program is restrictively
provided to high risk KHS members discharged from Bakersfield Memorial Hospital. Kern Health
System has more than 240,000 members and is associated with more 10 acute care hospitals. The
current pharmacy-based transitional care model could be expanded to the qualified members who
are discharged from other hospitals rather than Bakersfield Memorial Hospital. It is deemed a
10

budget impact analysis to demonstrate the cost savings when the model is provided to a larger
scope of population.

References
1. O'Sullivan, D., et al., The impact of a structured pharmacist intervention on the
appropriateness of prescribing in older hospitalized patients. Drugs Aging, 2014. 31(6):
p. 471-81.
2. Forster, A.J., et al., Adverse drug events occurring following hospital discharge. J Gen
Intern Med, 2005. 20(4): p. 317-23.
3. Schnipper, J.L., et al., Role of pharmacist counseling in preventing adverse drug events
after hospitalization. Arch Intern Med, 2006. 166(5): p. 565-71.
4. Kucukarslan, S.N., et al., Pharmacists on rounding teams reduce preventable adverse
drug events in hospital general medicine units. Arch Intern Med, 2003. 163(17): p. 2014-
8.
5. Al-Rashed, S.A., et al., The value of inpatient pharmaceutical counselling to elderly
patients prior to discharge. Br J Clin Pharmacol, 2002. 54(6): p. 657-64.
6. Dudas, V., et al., The impact of follow-up telephone calls to patients after hospitalization.
Am J Med, 2001. 111(9B): p. 26S-30S.
7. Szkiladz, A., et al., Impact of pharmacy student and resident-led discharge counseling on
heart failure patients. J Pharm Pract, 2013. 26(6): p. 574-9.
8. Walker, P.C., et al., Impact of a pharmacist-facilitated hospital discharge program: a
quasi-experimental study. Arch Intern Med, 2009. 169(21): p. 2003-10.
11

9. Warden, B.A., et al., Pharmacy-managed program for providing education and
discharge instructions for patients with heart failure. Am J Health Syst Pharm, 2014.
71(2): p. 134-9.
12

CHAPTER 2
Impact of a Pharmacy-Based Transitional Care Program on Hospital
Readmissions

Abstract
Objectives
Avoidable readmissions of patients discharged from hospitals are a major concern. This study
evaluates the impact of pharmacist-provided post-discharge services on hospital readmissions for
members of a US managed Medicaid health plan.

Methods
Synergy Pharmacy Solutions (SPS) initiated a transitional care service for high risk members of
Kern Health Systems (KHS) managed Medicaid plan in 2013. Over 1,100 patients were referred
to SPS between April, 2013 and March, 2015. KHS classified hospitalized members as at high
risk of readmission based on prior healthcare utilization and a health risk assessment questionnaire.
This study compares SPS transitional care recipients to a matched sample of KHS members
discharged from non-intervention hospitals. Thirty-day and 180-day readmissions and time-to-
readmission were defined as outcomes. Logistic regression and cox model were estimated
controlling for demographics, diagnostic and drug profiles, and prior hospital utilization.

13

Results
KHS identified 1,463 high risk patients discharged from non-intervention hospitals of which 1,005
were matched to 830 SPS patients in the 30-day population, and 669 were matched to 558 SPS
patients in the 180-day population, respectively. The SPS post-discharge intervention reduced the
risk of readmission within 30 days by 28.0% [OR= 0.720 (95% CI 0.526-0.985)] and reduced 180-
day readmission risk by 31.9% [OR=0.681 (95% CI 0.507-0.914]. The estimated effect of the SPS
intervention from the Cox model was a reduction in risk of 25% [H.R.=0.749, (95% CI 0.566-
0.992)].

Conclusions
A community pharmacy-based post-discharge transition of care program significantly reduced
readmission rates at 30 and 180 days.

Key Words: Pharmacy-Based Transition of Care, Readmissions, Ambulatory Care

14

2.1 Introduction
Patients with complex medical histories and medication regimens who are admitted to the hospital
are at risk for readmission due to a number of factors, including lapses in the continuity of care.
The average 30-day readmission rate in the United States is around 16% [1]. High readmission
rates have imposed a significant clinical and economic burden to the US healthcare system. As a
result, the Hospital Readmissions Reduction Program, enacted in October 2012 directed that the
Centers for Medicare and Medicaid Services reduce payments to hospitals with excess 30-day
readmission rates for conditions such as acute myocardial infarction (AMI), heart failure,
pneumonia, chronic obstructive pulmonary disease (COPD), total hip arthroplasty, and total knee
arthroplasty [2]. Thus, it is critical to identify the reasons for readmissions and to implement
programs to decrease the risk of readmissions.

Suboptimal medication therapy during the transitional care period following hospital discharge is
a major contributory factor to hospital readmissions and increased healthcare utilization [3].
Forster et al. estimated that 11% of patients experienced an adverse drug event after discharge
from inpatient services [4].Twenty-seven percent of readmissions were considered to be
preventable if the patient had received appropriate post-discharge medication monitoring [4].

Several services have been shown to impact hospital readmissions, including patient education,
medication adherence counseling, and medication reconciliation [5-11]. Pharmacist involvement
in discharge counseling, medication reconciliation, and telephone follow-up has resulted in a lower
incidence of preventable adverse drug events [7-10]. Consequently, it is reasonable to expect that
pharmacist-led transitional care services may decrease readmission rates.
15

Synergy Pharmacy Solutions (SPS) in Bakersfield, CA initiated an ambulatory care pharmacy-
based transitional care service in 2013 for recently discharged members of Kern Health Systems
(KHS) managed Medicaid health plan that were classified as high risk based on their healthcare
utilization, medical history, and social history. High-risk members admitted to a single local
hospital were referred to SPS for post discharge services. Over 1,100 members were referred to
SPS between April 2013 and March 2015.

This study evaluated the effect of the ambulatory care pharmacy-based transition of care (TOC)
services provided by SPS on 30-day and 180-day readmissions compared with a control group of
matched KHS members discharged from neighboring hospitals.

2.2 Methods
2.2.1 Ambulatory Care Pharmacy-Based Transitional Care Program
The risk of readmission for individuals enrolled in KHS managed Medicaid plan was evaluated
based on their history of hospitalizations, prescription medication utilization, and social history.
Adult patients discharged from a local hospital at high risk for readmission were automatically
referred to the SPS transitional care program. Those who met the following criteria were then
excluded by the SPS team: discharged to skilled nursing facility, rehabilitation facility, or hospice;
expired in hospital; left hospital against medical advice; or hospitalized for elective procedure,
obstetrical complications, substance abuse, urinary tract infection, or suicide attempt.

16

2.2.2 Intervention
Once the qualified patient agreed to participate in the SPS transitional care program, medications
were reconciled and any discrepancies between the patient’s self-reported medication use and the
hospital discharge orders were noted. Over the 30 days following discharge, the pharmacists
worked with the outpatient providers to resolve any medication-related problems, such as
inappropriate therapy, therapeutic duplications, and potential drug interactions. In addition, the
pharmacists counseled patients to improve medication adherence. The pharmacy staff reinforced
the discharge care plan, including post-discharge appointments, facilitating authorizations for
specialist care, arranging transportation for appointments, and working with the patient’s
dispensing pharmacy to resolve insurance-related issues. Patients requiring additional assistance
were invited for a face-to-face visit, which included more intensive counseling and assistance with
organizing medications.

Medication management services were documented directly into the existing electronic health
record (EHR) system of the ambulatory care pharmacy, which included customized reporting
capabilities. Daily reports were generated by the pharmacy team, and follow-up tasks were
assigned accordingly. The clinical pharmacy team acted as a liaison to bridge the communication
gaps between the patient, their prescribers, and their dispensing pharmacy, thereby facilitating
improvements in the continuity of care between the inpatient and outpatient settings.

2.2.3. Data Collection
The primary source of data for this study was the KHS paid claims database, which covered all
enrolled beneficiaries’ inpatient records, outpatient services, emergency room visits, and
17

prescription claims. ICD-9 codes were used to identify diagnoses. Medications were identified
using specific therapeutic class codes. The data related to TOC services were collected from the
EHR system at SPS.

2.2.4 Study Population
The intervention and control patient populations were selected from the pool of adult Medicaid
managed care members of KHS health plan who were discharged from either the study hospital
or control hospital (other neighboring hospitals contracted with KHS in Kern County, California)
and met the following inclusion criteria at the time of discharge:
 Active members of KHS with an inpatient stay at participating hospitals, AND
 “High risk” as determined by KHS’s algorithm including prior healthcare utilization and
social history, OR
 Discharged with prescription claims for >=5 medications, OR
 Recently admitted to any local hospital within the last 45 days

For the intervention patients referred to SPS, an index date was defined as the discharge date of
their hospitalization immediately preceding referral. All referred patients were screened for 6
months of continuous health plan enrollment prior to the index hospitalization and 30 days of post-
discharge data. A second screen requiring 180 days of post-discharge data was applied for hospital
admissions used in the analysis of readmission at 6 months.
The control group was identified ex post by applying the KHS risk screening algorithm to its
members discharged from neighboring hospitals during the period October 2012 to March 2015.
In order to identify an index hospitalization for each control group patient, all hospitalizations for
18

each patient were converted into episodes of hospital care and matched to the index episode in an
intervention group member based on the number of prior hospitalizations and length of stay (LOS)
[±1 day]. In order to maximize the power of the analysis, we included all the patients from the
control group that were matched to at least one patient in the intervention group. Thus, intervention
patients could have more than one matched control hospitalization in the final study population.

A total of 1,123 patients were referred to SPS, of which 830 met the enrollment criteria for the 30-
day analysis and were matched to 1,005 patients receiving usual care. A total of 558 SPS patients
met the enrollment criteria for the 180-day analysis and were matched to 669 usual care patients
(Figure 2-1). In the intervention group, all patients referred to SPS were included in this intent-
to-treat (ITT) analysis, including those who did not qualify for services, who could not be
contacted, and who declined services.

2.2.5 Outcome Measures
The primary outcome measures were all-cause 30-day and 180-day hospital readmissions, which
were defined as an inpatient stay within 30 or 180 days after the index hospitalization discharge
date. The 180-day window was evaluated in order to address any concerns about the potential
overlap between the 30-day intervention period and the 30-day readmission window. Time to
readmission was also calculated and used as a secondary outcome measure.

2.2.6 Statistical Analysis
Descriptive statistics were applied to test for differences in demographics and clinical
characteristics between the intervention and control groups. Chi Square test was used to test for
19

baseline differences between intervention and control patients in the distribution of gender, race,
age levels, and indicator of prior hospitalizations. Student’s t-test was utilized to compare the mean
index hospitalization LOS and number of medications.

Logistic regression was used to estimate the impact of the transitional care intervention on the
likelihood of a 30-day or 180-day readmission. These models controlled for age, gender, race, prior
hospitalizations (yes/no), length of stay of the index hospitalization, inpatient diagnoses prior to
and including the index hospitalization, and the mix of medication classes used by the patient over
the six months prior to admission. Time to readmission was analyzed using a Cox proportional
hazards model, controlling for the same covariates used in the logistic analyses. The study
population for the Cox analysis was the same population used in the 30-day readmission analysis.
Data analyses were performed using SAS 9.4 (SAS Institute Inc, Cary, North Carolina).

2.3 Results
A total of 1,123 patients were referred to SPS TOC program during the study period, of whom 841
were continuously enrolled in the KHS Medicaid plan 6 months prior to and 30 days after the index
hospitalization. After matching, 830 intervention patients and 1,005 control observations were
included in the logistic regression model for 30-day readmission. For the analysis of 180-day
readmission, 564 referred patients were continuously enrolled in the KHS plan 6 months prior to
and 30 days after the index hospitalization, of which 558 from the intervention group and 669 from
the control group were matched.

20

Table 2-1 compares the baseline characteristics of the intervention and control groups. In the 30-
day readmission population, the age, gender, and number of medications from the intervention and
control groups were not statistically different, while the intervention group tended to have a higher
proportion of prior inpatient admissions and longer index hospitalization LOS. In the 180-day
matched population, the age, race, and gender of the two groups were similar. Similar to the 30-
day population, the intervention group had a higher proportion of patients with prior inpatient stays
and longer LOS. In order to control for the differences between groups, these demographics and
clinical characteristics were used as independent variables in both the logistic and Cox models.

After controlling for confounders, the multivariate logistic regression analysis on 30-day
readmissions showed that the SPS TOC intervention was associated with a statistically significant
28.0% reduction in 30-day readmissions [OR = 0.720 (95% CI 0.526 – 0.985)] (Table 2-2). Other
factors associated with an increased risk of 30-day readmissions were prior inpatient stays
[OR=1.930 (95% CI 1.255 – 2.969)] and longer LOS [OR=1.054 (95% CI 1.018 – 1.091)]. Patients
who were hospitalized for AMI, COPD, digestive diseases, infections and parasitic diseases, and
neoplasms had a higher likelihood of 30-day readmissions. In addition, patients with prescription
claims for anti-epileptic drugs, dialysis solutions, and dietary supplements (including intravenous
nutrition) also had higher 30-day readmission rates.

The 180 analysis indicated that the TOC intervention also significantly reduced readmissions at 6
months by 31.9% [OR 0.681(95% CI 0.507 – 0.914)] (Table 2-3). Patients with a prior
hospitalization, COPD, and infectious and parasitic diseases were more likely to be readmitted
within 180 days. Patients hospitalized for blood disorders and diabetes were more likely to be
21

readmitted within 180days but not within 30days. Use of dialysis solutions and dietary
supplements, including IV nutrition, were associated with higher risk for 180-day readmissions,
consistent with the 30-day outcomes.

Time to readmission was analyzed by Cox proportional hazards model on the 30-day ITT
population. After adjusting for all demographics and clinical characteristics, the model illustrated
that patients receiving TOC services had a 25% lower hazard of readmission, compared with
patients receiving usual care [HR=0.749 (95% CI 0.566-0.992)] (Table 2-4.).

2.4 Discussion
This study estimated the impact of community pharmacist-based transitional care on 30-day and
180-day readmission rates in a managed Medicaid population. Both the logistic regression and
Cox proportional hazard models found that the TOC services at SPS were associated with
significantly lower all-cause readmissions, at 30 days and 180 days, compared with usual discharge
care. Consequently, this study adds to the body of literature on the impacts of TOC and the role of
the pharmacist in TOC. Many transitional care interventions have shown the benefits of close post
discharge care coordination on readmission rates and healthcare utilization [12-24]. However,
these studies either focused only on specific disease conditions or only evaluated the effects over
a short period of time. O’Dell et al reported that clinical pharmacist services for cardiac patients
with unstable angina were associated with lower readmission rates compared with usual care.
However, the results were not significant in the larger pool of all cardiac patients [23]. Koehler
and colleagues designed a randomized clinical trial and showed that pharmacist-led interventions
reduced 30-day readmissions while not affecting 60-day readmissions [24]. Kirkman et al also
22

found that telephonic follow-up by pharmacists reduced 30-day readmissions. Their regression
analysis demonstrated that the 30-day readmission odds ratio for patients who received usual care
was 1.53. When the reference category was reversed, the odds ratio was comparable to our result
of 0.72. To our knowledge, our 180-day analysis exceeds the follow-up period of existing studies.

In addition, whereas previous studies have largely focused on transitional care services within
academic centers or closed systems such as the Veterans’ Administration, this study evaluates the
impact of a standalone ambulatory care pharmacy-based TOC service. In addition, prior research
on the impact of the pharmacist has focused on medication reconciliation prior to discharge. The
SPS TOC program has demonstrated that medication related problems often persist after discharge,
which requires further interventions by the pharmacist for the 30 day period after discharge. More
details on these interventions will be described in future publications.

This study analyzed all patients referred to the TOC services [the ITT population], including those
who did not qualify for services [6.7%], patients who could not be contacted [4.7%], and patients
who declined services [2.4%]. Thus, the results from this study estimate the effects of TOC
services on the entire referred population. The corresponding percentages in the 180-day
population were 3.4%, 2.2% and 5.0%, respectively. There were many possible reasons for patients
refusing services or being unable to contact, including the “cold call” nature of the phone call from
SPS, the perception that the TOC services were unnecessary, disconnected phone numbers,
homelessness, or refusal to discuss their healthcare with a professional other than their own
physician. Despite including all of these patients in the analysis, the results demonstrated a
23

significant reduction in readmissions associated with the intervention – an effect that may have
had an even larger magnitude had we excluded them from the analysis.

In addition to evaluating the impact of the SPS TOC program on readmission rates, this study also
explored the potential factors associated with higher readmission rates. In both the 30-day and
180-day analyses, prior hospitalizations within 6 months, chronic obstructive pulmonary diseases
and infectious and parasitic diseases were significantly associated with increased risk of
readmission. Similarly, using medications for electrolyte imbalance and dietary supplementation
were shown to be related with higher readmission rates. It is possible that use of these prescriptions
may be indicators for chronic illnesses, such as renal failure and/or gastrointestinal disorders.
Although the current results described many factors that could potentially affect readmissions,
future stand-alone studies focusing on causal relationships with readmissions would be warranted.
Once those factors are successfully identified, then future transitional care services may be
designed to target the appropriate patient groups for whom they would be most effective.

2.4.1 Limitations
Our study was limited by several factors. First, this study used a nonrandomized design, selecting
patients discharged from intervention and control hospitals. While we matched the intervention
and control population using number of prior hospitalizations and the LOS of the index
hospitalization, the healthcare utilization in the intervention group was still higher than that of the
control group, possibly indicating that the intervention group’s health status was worse than that
of the control group [Table 2-1]. After controlling for these imbalances between the groups in the
multivariate regression models,, the intervention was still shown to have a significant impact on
24

readmission rates. Second, the generalizability of our results may be limited, because this study
focused on the rural population of Bakersfield, CA. Finally, an observational study cannot
establish the causality of factors impacting the risk of readmission investigated in the current study.
Future studies would be necessary to answer this research question.

2.5 Conclusions
Compared with usual discharge care, the ambulatory care pharmacy-based transition of care
program significantly reduced readmission rates by 28.0% at 30 days and 31.9% at 180 days. These
are likely conservative estimates of the treatment effect, as all referred patients were included in
the intent-to-treat analysis.

2.6 References
1. Christy, S., B. Sin, and S. Gim, Impact of an Integrated Pharmacy Transitions of Care
Pilot Program in an Urban Hospital. J Pharm Pract, 2015.
2. Readmission Reduction Program. Centers for Medicare and Medicaid Services. 2014
[cited 2015 Octorber 09].
3. O'Sullivan, D., et al., The impact of a structured pharmacist intervention on the
appropriateness of prescribing in older hospitalized patients. Drugs Aging, 2014. 31(6):
p. 471-81.
4. Forster, A.J., et al., Adverse drug events occurring following hospital discharge. J Gen
Intern Med, 2005. 20(4): p. 317-23.
5. Schnipper, J.L., et al., Role of pharmacist counseling in preventing adverse drug events
after hospitalization. Arch Intern Med, 2006. 166(5): p. 565-71.
25

6. Kucukarslan, S.N., et al., Pharmacists on rounding teams reduce preventable adverse
drug events in hospital general medicine units. Arch Intern Med, 2003. 163(17): p. 2014-
8.
7. Al-Rashed, S.A., et al., The value of inpatient pharmaceutical counselling to elderly
patients prior to discharge. Br J Clin Pharmacol, 2002. 54(6): p. 657-64.
8. Dudas, V., et al., The impact of follow-up telephone calls to patients after hospitalization.
Am J Med, 2001. 111(9B): p. 26S-30S.
9. Szkiladz, A., et al., Impact of pharmacy student and resident-led discharge counseling on
heart failure patients. J Pharm Pract, 2013. 26(6): p. 574-9.
10. Walker, P.C., et al., Impact of a pharmacist-facilitated hospital discharge program: a
quasi-experimental study. Arch Intern Med, 2009. 169(21): p. 2003-10.
11. Warden, B.A., et al., Pharmacy-managed program for providing education and
discharge instructions for patients with heart failure. Am J Health Syst Pharm, 2014.
71(2): p. 134-9.
12. Naylor, M.D., et al., Comprehensive discharge planning and home follow-up of
hospitalized elders: a randomized clinical trial. JAMA, 1999. 281(7): p. 613-20.
13. Naylor, M.D., et al., Transitional care of older adults hospitalized with heart failure: a
randomized, controlled trial. J Am Geriatr Soc, 2004. 52(5): p. 675-84.
14. Coleman, E.A., et al., Preparing patients and caregivers to participate in care delivered
across settings: the Care Transitions Intervention. J Am Geriatr Soc, 2004. 52(11): p.
1817-25.
26

15. Kirkham, H.S., et al., The effect of a collaborative pharmacist-hospital care transition
program on the likelihood of 30-day readmission. Am J Health Syst Pharm, 2014. 71(9):
p. 739-45.
16. Townsend, J., et al., Reduction in hospital readmission stay of elderly patients by a
community based hospital discharge scheme: a randomised controlled trial. BMJ, 1988.
297(6647): p. 544-7.
17. Rich, M.W., et al., A multidisciplinary intervention to prevent the readmission of elderly
patients with congestive heart failure. N Engl J Med, 1995. 333(18): p. 1190-5.
18. Stauffer, B.D., et al., Effectiveness and cost of a transitional care program for heart
failure: a prospective study with concurrent controls. Arch Intern Med, 2011. 171(14): p.
1238-43.
19. Nyweide, D.J., et al., Continuity of care and the risk of preventable hospitalization in
older adults. JAMA Intern Med, 2013. 173(20): p. 1879-85.
20. Anderson, S.L., et al., Implementation of a clinical pharmacy specialist-managed
telephonic hospital discharge follow-up program in a patient-centered medical home.
Popul Health Manag, 2013. 16(4): p. 235-41.
21. Misky, G.J., H.L. Wald, and E.A. Coleman, Post-hospitalization transitions: Examining
the effects of timing of primary care provider follow-up. J Hosp Med, 2010. 5(7): p. 392-
7.
22. Gil, M., et al., Impact of a combined pharmacist and social worker program to reduce
hospital readmissions. J Manag Care Pharm, 2013. 19(7): p. 558-63.
23. O'Dell, K.M. and S.N. Kucukarslan, Impact of the clinical pharmacist on readmission in
patients with acute coronary syndrome. Ann Pharmacother, 2005. 39(9): p. 1423-7.
27

24. Koehler, B.E., et al., Reduction of 30-day postdischarge hospital readmission or
emergency department (ED) visit rates in high-risk elderly medical patients through
delivery of a targeted care bundle. J Hosp Med, 2009. 4(4): p. 211-8.
28

2.7 Tables
Table 2-1. Descriptive Statistics of ITT Population

30-Day Population
180-Day Population

Intervention
(N=830)
Control
(N=1005)
P-
value
Intervention
(N=558)
Control
(N=669)
P-value
Male, (N & %)
269 32.41 335 33.33 0.6751 179 32.08 216 32.29 0.9381
Race, (N & %)
White
325 39.16 354 35.22
0.0309
222 39.78 241 36.02
0.3014
Hispanic
292 35.18 414 41.19 194 34.77 259 38.71
Black
110 13.25 107 10.65 77 13.80 81 12.11
Other
103 12.41 130 12.94 65 11.65 88 13.15
Age groups, (N & %)
24 and Below
66 7.95 65 6.47
0.0642
44 7.89 52 7.77
0.4255
25-34
104 12.53 139 13.83 76 13.62 103 15.40
35-44
117 14.10 167 16.62 86 15.41 110 16.44
45-54
217 26.14 297 29.55 148 26.52 198 29.60
55-64
275 33.13 291 28.96 175 31.36 176 26.31
65 and Above
51 6.14 46 4.58 29 5.20 30 4.48
Prior hospitalization (N & %)

269 32.4 158 15.72 <.0001 195 34.95 73 10.91 <.0001
Length of stay, Days (Mean & SD)

4.4964 3.8976 3.791 3.5485 <.0001 4.45 3.91 3.56 3.31 <.0001
Number of medications, (Mean &
SD)

6.9711 3.2136 6.7811 2.7855 0.1752 6.95 3.11 6.85 2.81 0.547

29

Table 2-2. Logistic Regression Analysis for Effect of TOC Intervention on 30-Day Readmissions, IIT Population
Parameter Odds Ratio 95% Confidence Limits P-Value
Intervention 0.720 0.526 0.985 0.04
Gender
Female 0.857 0.607 1.211 0.3822
Male Reference Group - - -
Race
Black 1.191 0.739 1.921 0.4732
Hispanic 0.789 0.547 1.139 0.2056
Other 0.823 0.499 1.356 0.4441
White Reference Group - - -
Age Groups
24 and Below Reference Group - - -
25-34 0.858 0.424 1.736 0.6704
35-44 0.766 0.369 1.589 0.4738
45-54 0.812 0.4 1.651 0.5658
55-64 0.660 0.318 1.37 0.2649
65 and Above 0.950 0.378 2.384 0.9126
Length of Stay, Days 1.054 1.018 1.091 0.0028
Indicator of Prior Hospitalization 1.930 1.255 2.969 0.0027
Hospital Diagnoses

Acute Myocardial Infarction 2.439 1.089 5.462 0.0302
Arrhythmia, Heart Conduction Disorders 0.507 0.105 2.46 0.3994
Blood Disease 1.380 0.491 3.883 0.5416
Heart Failure 1.225 0.568 2.643 0.6054
30

Other Disorders of The Central Nervous System 1.207 0.556 2.62 0.6342
Chronic Obstructive Pulmonary Disease and Allied Conditions 2.158 1.021 4.562 0.0441
Diabetes 1.378 0.675 2.813 0.3791
Digestive Diseases 1.658 1.04 2.643 0.0334
Endocrine Disorders, Non-Diabetes 1.031 0.441 2.414 0.943
Genitourinary Diseases 1.174 0.666 2.069 0.5793
Infectious and Parasitic Diseases 2.202 1.299 3.732 0.0034
Injury and Poisoning 1.495 0.917 2.438 0.1072
Diseases of Musculoskeletal System And Connective Tissue 1.230 0.628 2.409 0.5459
Neoplasms 1.961 1.019 3.772 0.0436
Other Circulatory Disease 1.628 0.931 2.846 0.0871
Complications of Pregnancy, Childbirth 1.483 0.654 3.361 0.3453
Respiratory Disorders 0.611 0.298 1.253 0.179
Skin and Subcutaneous Tissue Diseases 1.504 0.736 3.073 0.2634
Symptoms, Signs, and Ill-Defined Conditions 1.134 0.565 2.277 0.7234
Medication Classes

Cardiovascular 0.704 0.473 1.048 0.0836
Pulmonary 1.200 0.771 1.868 0.4195
Diabetes 1.186 0.836 1.684 0.3385
Psychotropic 0.882 0.623 1.249 0.4806
Pain 0.985 0.639 1.519 0.9463
Anti-epileptic drugs 1.591 1.138 2.225 0.0066
Anti-Parkinson treatments 0.921 0.668 1.271 0.6177
Gastrointestinal agents 0.943 0.638 1.394 0.7688
Anti-infectives 1.051 0.727 1.518 0.7914
Hormone replacement 0.593 0.287 1.224 0.1575
31

Contraception 0.396 0.117 1.342 0.1371
Male sexual dysfunction, benign prostatic hyperplasia treatment 1.188 0.524 2.696 0.6799
Bladder/urinary treatment 1.242 0.867 1.78 0.2378
Steroids (various uses) 0.983 0.63 1.532 0.9392
Cough/cold, seasonal allergy medication 0.812 0.557 1.185 0.2807
Cancer treatment 1.519 0.804 2.868 0.1975
Dialysis solutions 1.734 1.181 2.545 0.005
Dietary supplementation, incl IV nutrition 1.531 1.085 2.162 0.0153
Thyroid medication 1.346 0.876 2.069 0.1749
Osteoporosis treatment 0.680 0.262 1.766 0.4284
Dermatologic treatment 1.179 0.426 3.263 0.7507
Ophthalmic treatment 1.113 0.283 4.379 0.8788
Surgery preparations 0.914 0.634 1.318 0.6312
Nicotine replacement therapy 1.086 0.415 2.846 0.8662
Misc. medical supplies 1.333 0.693 2.564 0.3892

32

Table 2-3. Logistic Regression Analysis for Effect of TOC Intervention on 180-Day Readmissions, IIT Population
Parameter Odds Ratio 95% Confidence Limits P-Value
Intervention 0.681 0.507 0.914 0.0106
Gender
Female 0.981 0.715 1.344 0.903
Male Reference Group - - -
Race
Black 1.014 0.65 1.582 0.9505
Hispanic 0.873 0.629 1.211 0.4152
Other 0.867 0.552 1.362 0.5367
White Reference Group - - -
Age Groups
24 and Below Reference Group - - -
25-34 0.718 0.387 1.333 0.2942
35-44 0.775 0.402 1.492 0.4452
45-54 0.678 0.354 1.298 0.2406
55-64 0.673 0.347 1.304 0.2408
65 and Above 0.663 0.277 1.585 0.3553
Length of Stay, Days 1.036 0.999 1.074 0.0538
Indicator of Prior Hospitalization 3.728 2.434 5.709 <.0001
Hospital Diagnoses

Acute Myocardial Infarction 1.28 0.529 3.098 0.5845
Blood Disease 3.337 1.14 9.767 0.0279
Heart Failure 1.459 0.703 3.025 0.3103
33

Other Disorders of The Central Nervous System 1.345 0.649 2.79 0.4257
Chronic Obstructive Pulmonary Disease and Allied Conditions 2.865 1.436 5.716 0.0028
Diabetes 2.3 1.188 4.453 0.0135
Digestive Diseases 1.09 0.688 1.728 0.7136
Endocrine Disorders, Non-Diabetes 1.534 0.768 3.065 0.2255
Genitourinary Diseases 0.799 0.475 1.345 0.3984
Infectious and Parasitic Diseases 2.335 1.336 4.083 0.0029
Injury and Poisoning 1.334 0.828 2.148 0.2359
Diseases of Musculoskeletal System And Connective Tissue 0.898 0.479 1.683 0.7372
Neoplasms 0.968 0.466 2.011 0.9313
Other Circulatory Disease 1.099 0.639 1.89 0.7324
Complications of Pregnancy, Childbirth 1.011 0.492 2.078 0.9761
Respiratory Disorders 0.828 0.439 1.561 0.5595
Skin and Subcutaneous Tissue Diseases 1.168 0.576 2.37 0.6661
Symptoms, Signs, and Ill-Defined Conditions 0.591 0.286 1.222 0.1561
Medication Classes

Cardiovascular 1.085 0.758 1.552 0.6558
Pulmonary 0.892 0.59 1.348 0.5866
Diabetes 1.199 0.853 1.687 0.2967
Psychotropic 1.055 0.775 1.435 0.7352
Pain 1.009 0.711 1.431 0.9604
Anti-epileptic drugs 1.042 0.768 1.413 0.7933
Anti-Parkinson treatments 0.98 0.73 1.317 0.8948
Gastrointestinal 1.088 0.767 1.543 0.6367
Anti-infectives 0.844 0.608 1.173 0.3131
34

Hormone replacement 0.89 0.494 1.603 0.6987
Contraception 0.847 0.41 1.747 0.652
Male sexual dysfunction, benign prostatic hyperplasia treatment 0.818 0.404 1.656 0.5771
Bladder/urinary 1.359 0.565 3.266 0.4931
Steroids (various uses) 0.684 0.269 1.741 0.4255
Cough/cold, seasonal allergies 1.006 0.725 1.395 0.9728
Dialysis solutions 1.611 1.127 2.302 0.0088
Dietary supplementation, incl IV nutrition 1.653 1.198 2.28 0.0022
Osteoporosis treatment 1.157 0.765 1.749 0.4896
Dermatologic treatment 0.558 0.165 1.881 0.3465
Surgery preparations 0.572 0.297 1.101 0.0944
Nicotine replacement therapy 0.9 0.339 2.391 0.8331
Misc. medical supplies 0.883 0.663 1.177 0.3973

35

Table 2-4. Cox Proportional Hazards Analysis on Time-to-Readmission
Parameter Hazard Ratio 95% Confidence Limits P-Value
Intervention 0.749 0.566 0.992 0.0435
Gender
Female 0.873 0.639 1.191 0.3914
Male Reference Group - - -
Race
Black 1.162 0.759 1.778 0.4896
Hispanic 0.808 0.581 1.124 0.2056
Other 0.873 0.555 1.373 0.5573
White Reference Group - - -
Age Groups
24 and Below Reference Group - - -
25-34 0.842 0.447 1.585 0.5941
35-44 0.744 0.387 1.43 0.3755
45-54 0.795 0.424 1.493 0.4762
55-64 0.671 0.35 1.283 0.2275
65 and Above 0.922 0.412 2.065 0.844
Length of Stay, Days 1.047 1.017 1.078 0.0018
Indicator of Prior Hospitalization 1.8 1.227 2.639 0.0026
Hospital Diagnoses

Acute Myocardial Infarction 1.882 0.97 3.654 0.0616
Arrhythmia, Heart Conduction Disorders 0.556 0.132 2.338 0.4235
Blood Disease 1.276 0.525 3.098 0.5904
Heart Failure 1.184 0.604 2.319 0.6231
36

Other Disorders of The Central Nervous System 1.255 0.635 2.481 0.5132
Chronic Obstructive Pulmonary Disease and Allied Conditions 1.825 0.952 3.497 0.0701
Diabetes 1.182 0.632 2.209 0.6007
Digestive Diseases 1.51 1.008 2.263 0.0458
Endocrine Disorders, Non-Diabetes 1.012 0.474 2.164 0.9746
Genitourinary Diseases 1.148 0.7 1.881 0.5845
Infectious and Parasitic Diseases 1.955 1.246 3.068 0.0035
Injury and Poisoning 1.392 0.917 2.111 0.1203
Diseases of Musculoskeletal System And Connective Tissue 1.133 0.625 2.057 0.6803
Neoplasms 1.65 0.939 2.9 0.0816
Other Circulatory Disease 1.483 0.913 2.409 0.1117
Complications of Pregnancy, Childbirth 1.348 0.642 2.83 0.4306
Respiratory Disorders 0.603 0.315 1.154 0.1268
Skin and Subcutaneous Tissue Diseases 1.381 0.74 2.575 0.3106
Symptoms, Signs, and Ill-Defined Conditions 1.074 0.576 2.003 0.8212
Medication Classes

Cardiovascular 0.703 0.49 1.009 0.0557
Pulmonary 1.225 0.833 1.802 0.3031
Diabetes 1.182 0.866 1.615 0.2921
Psychotropic 0.911 0.665 1.248 0.5615
Pain 0.998 0.673 1.479 0.9916
Anti-epileptic drugs 1.516 1.122 2.047 0.0067
Anti-Parkinson treatments 0.918 0.689 1.225 0.5629
Gastrointestinal agents 0.965 0.673 1.383 0.8449
Anti-infectives 1.062 0.761 1.483 0.7216
Hormone replacement 0.604 0.304 1.198 0.149
37

Contraception 0.418 0.13 1.345 0.1435
Male sexual dysfunction, benign prostatic hyperplasia
Treatment
1.176 0.588 2.351 0.6467
Bladder/urinary treatment 1.187 0.862 1.635 0.2926
Steroids (various uses) 0.957 0.643 1.424 0.8283
Cough/cold, seasonal allergy medication 0.86 0.614 1.204 0.3801
Cancer treatment 1.351 0.783 2.328 0.2794
Dialysis solutions 1.665 1.192 2.327 0.0028
Dietary supplementation, incl IV nutrition 1.453 1.071 1.972 0.0165
Thyroid medication 1.33 0.915 1.935 0.135
Osteoporosis treatment 0.761 0.327 1.771 0.5264
Dermatologic treatment 1.27 0.534 3.018 0.5885
Ophthalmic treatment 1.22 0.363 4.102 0.7485
Surgery preparations 0.901 0.649 1.249 0.5306
Nicotine replacement therapy 1.098 0.482 2.504 0.8239
Misc. medical supplies 1.25 0.707 2.21 0.4425
38

2.8 Figures

Figure 2-1. Study Population Selection Intervention group

±
180 days of claims data prior to first hospitalization AND 30 days of claims data after first hospitalization
¥
180 days of claims data prior to first hospitalization AND 180 days of claims data after first hospitalization
*Matching by +/- 1 day of length of stay and prior hospitalization counts

Insufficient claims
data
¥
(n=559)
180-day analysis
population
(n=564)
Health plan member met
inclusion criteria,
referred to pharmacy
team (n=1123)

Insufficient claims
data
±
(n=282)
30-day analysis
population
(n= 841)
Matched
hospitalizations*
Did not match
(n= 11)
Matched
hospitalizations*
Did not match
(n= 6)
ITT population, 180-day
analysis (n=558)
ITT population, 30-day
analysis, Cox analysis
(n=830)
39

CHAPTER 3
Reducing Healthcare Costs through a Community Pharmacy-Based
Transitional Care Program

Abstract
Objectives
Patients discharged from the hospital are at risk for adverse events due to poor care coordination.
An estimated 2/3 of adverse events after discharge are due to medication-related problems. As a
result, post-discharge interventions by pharmacists could potentially improve patients’ outcomes.
A growing body of evidence suggests that transition of care programs involving pharmacists can
prevent unplanned readmissions.

Synergy Pharmacy Solutions (SPS) initiated a community pharmacy-based transitional care
service for qualifying members of Kern Health Systems (KHS) managed Medicaid plan in
2013. Qualifying members were classified as high risk based on healthcare utilization and social
history, or they had prescription claims for over 5 medications at the time of hospitalization, or
they were recently hospitalized within the last 45 days. Over 1100 qualifying KHS members were
discharged from Bakersfield Memorial Hospital and referred to SPS between April 2013 and
March 2015. This study evaluates the impact of this pharmacy-based post-discharge program on
healthcare utilization and costs.

40

Methods
For each of these experimental patients, an index date was defined as the discharge date of their
hospitalization immediately preceding referral to SPS. All referred patients were screened for 6
months of continuous health plan enrollment prior to the index hospitalization and 30 days of post-
discharge data. A second screen required 180 days of post-discharge data for an analysis of
readmission at 6 months.

The control group was identified by applying the KHS risk screening algorithm to its members
discharged from neighboring hospitals during the period October 2012 to March 2015. In order
to identify an index hospitalization for each control group patient, all hospitalizations for each
patient were converted into hospital episodes and matched to an episode in the experimental group
based on the number of prior hospitalizations and length of stay [±1 day]. Experimental patients
could have more than one matched control hospitalization.

In the experimental group, all matched patients referred to SPS were included in this intent-to-treat
(ITT) analysis, including those who did not qualify for services, who could not be contacted, and
who declined services. Thirty-day and 180-day inpatient, outpatient, prescription drug, emergency
room, and total costs were analyzed by OLS and GLM regressions. The effects on inpatient,
outpatient, and emergency room costs were additionally analyzed by two-part model.
Demographics, clinical characteristics, and comorbidity profiles were used as independent
variables in all models.

41

Results
A total of 1123 patients were referred to SPS, of whom 830 met the enrollment criteria for the 30-
day analysis and were matched to 1005 patients receiving usual care. A total of 558 SPS patients
met the enrollment criteria for the 180-day analysis and were matched to 669 usual care patients.
Both OLS and GLM model showed a cost reduction in 180-day total healthcare costs. The two-
part model outputs demonstrated that TOC services were associated with a reduction in inpatient
costs.

Key Words: Pharmacy-Based Transition of Care, Healthcare Costs, Ambulatory Care

42

3.1 Introduction
The transitional period following hospital discharge can be a time of confusion and medical
vulnerability for many patients, especially those who are taking multiple medications. As a result,
inappropriate use of medications may be a contributing factor to increased hospital readmissions
(1) (2) (3). It is reported that 11% of patients with complex medical histories and medication
regimens who use inpatient services are at risk for adverse drug events after discharge (4).
Pharmacist interventions including discharge counseling, medication reconciliation, and telephone
follow-up have been shown to be associated with lower incidence of adverse drug events (5) (6)
(7) (8). Consequently, it is reasonable to hypothesize that pharmacist-led transition of care (TOC)
services may decrease readmissions and related healthcare costs after a hospital discharge.
Previous studies have demonstrated that transitional care programs provided by medical centers,
community care groups, and insurers for patients with heart failure or other chronic conditions
were associated with lower healthcare utilization and inpatient costs after discharge (9) (10) (11).

An earlier analysis of the impact of TOC services on hospital readmissions was conducted using
data from the TOC intervention evaluated here. This study documented that the TOC services
were associated with significantly reduced 30-day and 180-day readmissions(12). However, the
earlier evaluation did not investigate the impact of the TOC program on healthcare costs after
discharge. The objective this study is to evaluate the impact of the TOC services on healthcare
costs, including inpatient, outpatient, medication, emergency room (ER), and total costs at 30 days
and 180 days after the index discharge date.
43

3.2 Research Methods
3.2.1 Study Design
This study employs a non-randomized, observational cohort design to compare high risk patients
receiving a community pharmacy-based TOC service to a matched cohort of patients receiving
usual discharge care. Both cohorts of patients were members of Kern Health System [KHS], a
Medicaid managed care health plan in Kern County, California.

3.2.2 Data
Data for this study were retrieved from the KHS paid claims database. The KHS claims included
data for patient demographics (age, gender, and race), enrollment status, claims for inpatient
records, outpatient services, emergency room visits, prescription medications, and related
healthcare costs. Comorbidities were recorded through ICD-9 codes, and medications were
identified using specific therapeutic class codes.

3.2.3 Pharmacist-Provided Transitional Care Program
Synergy Pharmacy Solutions (SPS) in Bakersfield, California initiated a pharmacy-based
transitional care program to provide medication monitoring services to eligible high-risk members
of KHS. The risk of readmission was evaluated using the Johns Hopkins ACG® predictive model
(13) which is based on patients’ history of hospitalizations, prescription medication utilization,
and social history. Adult patients discharged from a local hospital at high risk for readmission
were automatically referred to the SPS TOC program. For the patients who agreed to participate,
medications were reconciled between the patient’s actual medication use and the hospital records.
44

Over the 30 days following discharge, the pharmacists at SPS collaborated with the outpatient
providers to resolve medication-related problems.

3.2.4 Study Populations
The intervention population was selected from the pool of adult Medicaid managed care members
of KHS health plan who were discharged from the study hospital during the period April 2013 to
March 2015 and who met the criteria for high readmission risk at the time of discharge:
 Active members of KHS with an inpatient stay at participating hospitals, AND
 “High risk” as determined Johns Hopkins ACG Predictive model (13), OR
 Discharged with prescription claims for ≥ 5 medications, OR
 Recently admitted to any hospital within the last 45 days

For the intervention patients referred to SPS, an index date was defined as the discharge date of
the hospitalization that resulted in a referral to SPS. One hundred and eighty days of continuous
health plan enrollment prior to the index hospitalization and 30 days of post-discharge data were
used as the first screen criterion for all referred patients. One hundred and eighty days of
continuous health plan enrollment after discharge was applied as the second screen for the analysis
of healthcare costs at 6 months. All patients referred to SPS who met minimum data requirements
were included in this intent-to-treat (ITT) analysis. This included those patients who could not be
contacted by SPS and patients who declined services.

A control group of patients was identified ex post by KHS by applying the study’s risk screening
algorithm to its members discharged from neighboring hospitals during the period April 2013 to
45

March 2015. An index hospitalization was identified for each control group patient by converting
the patient’s hospitalization history into episodes of hospital care. Then, each episode was screened
for 180 days of continuous enrollment before the discharge date. All episodes identified for the
control patients that met minimum data requirements were matched to intervention episodes based
on the number of prior hospitalizations within 180 days before the index date and the length of
stay (LOS) of the index hospitalization [±1 day]. To increase the power of the analysis, we included
all control patients who had discharge episode that was matched to at least one discharge in the
intervention group. Thus, intervention patients could have been matched with more than one
patient from the control group.

3.2.5 Outcome Measures
The primary outcome measures were costs broken down by type of service: inpatient costs,
outpatient costs, medication costs, ER costs, and total healthcare costs at 30 and 180 days after the
index hospitalization discharge date. The costs of the SPS TOC services were included in
outpatient costs. All costs were adjusted to 2013 US dollars using the US medical consumer price
index (CPI) (14). Thirty-day readmissions were measured as secondary outcomes.

3.2.6 Statistical Methods
Chi-square and t-test statistics were utilized to test for differences in demographics, baseline
healthcare utilization, comorbidities, and medication usage between the intervention and control
group. Multivariate regression analyses were conducted to evaluate the adjusted effects of TOC
on healthcare costs at 30 days and 180 days. Ordinary least squares (OLS) regressions were used
as base case models for all types of costs, including inpatient costs, outpatient costs, medication
46

costs, ER costs and total costs. The models controlled for age, gender, race, prior hospitalization
within 6 months, LOS of the index hospitalization, comorbidities, and the mix of medication
classes used over the six months prior to the index hospitalization.

Sensitivity analyses were carried out to examine the robustness of the results from base-case
analyses using methods that account for skewed or zero mass distributions of some outcomes. We
performed generalized linear model (GLM) regressions defined by a logarithmic link function with
a Gamma distribution for all healthcare costs. This specification of GLM regression is widely used
for modeling healthcare costs with skewed distributions (15-22). To validate the distribution used
in the GLM model, the modified park test was applied.

Next, logistic-GLM two-part models (TPMs) were used for analyses of inpatient costs outpatient
costs, and ER costs, because 90.35% of the patients in the sample had no inpatient costs, 87.19%
of the patients had no outpatient costs, and 74.55% of the patients had no ER costs within 30 days
after index date. In the 180-day population, a total of 72.86% of patients had no inpatient costs,
68.13% had no outpatient costs, and 39.69% had no ER costs within 180 days after index. There
are two processes in TPM regressions (23, 24). First, a logistic regression is performed to estimate
the probability of having non-zero costs. Then, an OLS or GLM regression models the costs in the
samples of patients with positive healthcare costs. The probability of having any costs and the
expected costs conditional on having any costs are then multiplied to calculate the expected costs,
and the marginal effects are calculated accordingly (24).
47

3.3 Results
A total of 1,123 patients discharged from the participating study hospital were referred to the SPS
transitional care program during the study period, of whom 841 were continuously enrolled in the
KHS health plan for 6 months prior to and 30 days after the index hospitalization. After removing
unmatched hospitalizations by length of stay and number of prior hospitalizations, 830 intervention
patients and 1,005 control observations were included in the analysis for 30-day healthcare costs.
To analyze the impacts on 180-day healthcare costs, an additional screening was performed to
exclude patients that were not continuously enrolled in the health plan for 6 months after the index
hospitalization. A total of 558 patients from the intervention group and 669 from the control group
were included in the 180-day analysis (Figure 3-1).

The baseline characteristics of the intervention and control groups were displayed in Table 3-1
and Table 3-2. In the 30-day population, no significant differences were observed on the age,
gender, and number of medications between the intervention and control groups, while a larger
proportion of patients in the intervention group tended to have at least one prior inpatient admission
and a longer LOS of the index hospitalization compared to the control group. The distribution of
age, race, and gender of the two groups in the 180-day population were also balanced, while the
intervention group had a higher proportion of prior inpatient stays and longer LOS of the index
hospitalization.

The observed differences between the intervention and control patients in terms of rate of prior
hospitalizations and LOS were largely due to matching multiple control patients to the available
intervention patients. Specifically, some of the control episodes that were matched to intervention
48

episodes were likely to have had no prior hospitalizations and/or shorter LOS of the index
hospitalization. The comparison of baseline demographics and clinical characteristics were re-run
using a more restrictive 1:1 match which matched 681 control hospital episodes with 681
intervention episodes. The prior healthcare utilizations and comorbidities between those matched
groups were balanced (data not shown). The impact of these differences in baseline characteristics
on study results is minimized, as these variables are included as explanatory variables in all
regression models to control their impacts on healthcare costs.

OLS regressions were performed as the base-case analysis on all type of costs at 30 days and 180
days after index date, controlling for other factors that potentially affect the healthcare costs. Table
3-3 illustrates that TOC services were associated with lower costs over the first 30 days following
discharge, but these differences were not statistically significant. The average 6-month post-
discharge total healthcare costs and inpatient costs of the control group were $8,383 and $5,650.
The marginal effects of TOC program on 180-day total costs, inpatient costs, and outpatient costs
were statistically significant: -$5,353 (95% CI: -$9,871, -$835), -$4,735 (95% CI: -$9,144, -$326),
and -$175 (95% CI: -$341, -$9), respectively.

Healthcare costs are frequently skewed. In response, we applied GLM regression models to
estimate the impacts of the TOC service on all types of healthcare costs at 30 days and 180 days.
Modified park test outposts demonstrated that all the healthcare costs were in Gamma distribution
(Table 3-4). Overall, the GLM estimates were consistent with the direction of the OLS results.
Table 3-3 shows that the TOC service was associated with a statistically significant savings of
$2,139 on 180-Day total healthcare costs. A TPM was performed on hospitalization, outpatient
49

and ER costs, because a large proportion of the patients had no costs associated with these
healthcare services during the study period (Table 3-5). The TPM results were consistent with the
OLS estimates, especially in the direction of estimates. The TPM using GLM as the second
regression showed that TOC was associated with a statistically significant savings of $1,723 on
180-day inpatient costs, while the TPM results using OLS as the second regression showed a non-
statistically significant reduction in 180-day inpatient costs of $3,607.

3.4 Discussion
This study evaluated the impact of a pharmacy-based transitional care program on 30-day and 180-
day healthcare costs in a managed Medicaid population. Both the OLS and GLM regressions
illustrated that the TOC services at Synergy Pharmacy Solution were associated with significantly
lower 180-day total healthcare costs after index hospitalization, compared with usual discharge
care. In addition, the TPM model with the second step of GLM regression showed that the SPS
TOC program was associated with significantly lower 180-day inpatient costs. As all the types of
healthcare costs were verified with Gamma distribution, the estimates from GLM and TPM with
GLM models were expected to accurately reflect the influence from the TOC program.

To our knowledge, this study adds to the body of literature on the impact of TOC on healthcare
costs. Whereas previous studies have largely focused on transitional care services within integrated
medical centers (25), this study is one of the first evaluating the impact of a standalone ambulatory
care pharmacy-based TOC service on overall healthcare costs using claims data. The SPS TOC
program has demonstrated that medication related problems often persist after discharge, which
50

requires further interventions by the pharmacist for the 30-day period after discharge. More details
on these interventions will be described in future publications.

A recently published study on an insurer-initiated, pharmacist-led care transition program has
demonstrated that post-discharge medication monitoring reduced the 30-day readmission rate by
50% and saved $1,347 per member from the reduced readmissions (11). The analysis excluded all
referred patients who could not be contacted or refused services. We replicated these selection
criteria by dropping our ITT approach and comparing only those intervention patients actually
receiving TOC services to controls [Appendix Exhibit 1]. The multivariate logistic regression for
30-day readmission rates estimated that the SPS TOC program was associated with a 49%
reduction in 30-day readmission [Appendix Exhibit 2], nearly doubling the reduction of 28%
estimated using an ITT population (12).

Our analyses have several strengths. First, our costs were estimated using the actual amounts paid
by KHS retrieved directly from claims data, as opposed to national averages from HCUP or other
publicly available databases. Second, the intervention and control patients were discharged from
hospitals within the same county (Kern, California), which minimizes the potential for regional
bias.

Finally, this study applied ITT principles in selecting its intervention population, which is a
conservative method for analyzing the impact of the TOC services. Within this ITT population,
3.9% patients refused the TOC services and 8.3% patients could not be contacted [Appendix
Exhibit 1]. The sensitivity analysis, which excluded patients who refused services or were unable
51

to be contacted, was conducted to evaluate the effect of the ITT approach on 30-day readmissions.
The results from this sensitivity analysis found larger TOC effects, which suggest that patients
who refuse services or who cannot be contacted are at higher risk for readmission than average. If
this is true, then future TOC programs might be improved by focusing on interventions designed
to prevent loss to follow-up or patient non-participation.

This study has several limitations. First, patient selection was nonrandomized. While the number
of prior hospitalizations and the LOS of the index hospitalization were matched, the prior
healthcare utilization in the intervention group was still higher than that of the control group,
possibly indicating poorer health status in the intervention population. These imbalances would
likely impact the estimated results in a conservative direction. We controlled for these imbalances
between the groups by including these covariates in the multivariate regression models. Still, the
observational nature of this study precluded us from controlling for unobservable confounders;
therefore, we cannot establish the causality of factors impacting the healthcare related costs
investigated in the current study.

Second, since we did not have information on patients in the control group who would have
declined to participate or who would not have been reached by phone, we could not exclude those
patients from the analysis. In order to obtain conservative results, we included the entire ITT
population from the intervention group in the regression models.

52

Third, the generalizability of our conclusions may be limited, because the study was restrictively
performed on the rural population of Kern County, California. Although the internal validity of
the analysis was strong, these results would need to be replicated in other populations.

3.5 Conclusion
Compared with usual discharge care, the ambulatory care pharmacy-based transition of care
program was associated with significantly reductions in 180-day total healthcare costs. These are
likely conservative estimates of the treatment effect, as all referred patients were included in the
intent-to-treat analysis.

3.6 References
1. O'Sullivan, D., et al., The impact of a structured pharmacist intervention on the
appropriateness of prescribing in older hospitalized patients. Drugs Aging, 2014. 31(6):
p. 471-81.
2. O'Connor, M.N., P. Gallagher, and D. O'Mahony, Inappropriate prescribing: criteria,
detection and prevention. Drugs Aging, 2012. 29(6): p. 437-52.
3. Gallagher, P.F., M.N. O'Connor, and D. O'Mahony, Prevention of potentially
inappropriate prescribing for elderly patients: a randomized controlled trial using
STOPP/START criteria. Clin Pharmacol Ther, 2011. 89(6): p. 845-54.
4. Forster, A.J., et al., Adverse drug events occurring following hospital discharge. J Gen
Intern Med, 2005. 20(4): p. 317-23.
5. Al-Rashed, S.A., et al., The value of inpatient pharmaceutical counselling to elderly
patients prior to discharge. Br J Clin Pharmacol, 2002. 54(6): p. 657-64.
53

6. Dudas, V., et al., The impact of follow-up telephone calls to patients after hospitalization.
Am J Med, 2001. 111(9B): p. 26S-30S.
7. Szkiladz, A., et al., Impact of pharmacy student and resident-led discharge counseling on
heart failure patients. J Pharm Pract, 2013. 26(6): p. 574-9.
8. Walker, P.C., et al., Impact of a pharmacist-facilitated hospital discharge program: a
quasi-experimental study. Arch Intern Med, 2009. 169(21): p. 2003-10.
9. Jackson, C., et al., Incremental Benefit of a Home Visit Following Discharge for Patients
with Multiple Chronic Conditions Receiving Transitional Care. Popul Health Manag,
2015.
10. Stauffer, B.D., et al., Effectiveness and cost of a transitional care program for heart
failure: a prospective study with concurrent controls. Arch Intern Med, 2011. 171(14): p.
1238-43.
11. Polinski, J.M., et al., An Insurer's Care Transition Program Emphasizes Medication
Reconciliation, Reduces Readmissions And Costs. Health Aff (Millwood), 2016. 35(7): p.
1222-9.
12. Ni, W., et al., Impact of a Pharmacy-Based Transitional Care Program on Hospital
Readmissions. American Journal of Managed Care, Forthcoming.
13. http://acg.jhsph.org/index.php/the-acg-system-advantage/predictive-models.
08/15/2016].
14. http://www.bls.gov/cpi/#data. 08/18/16].
15. Karve, S.J., et al., Health Care Utilization and Costs among Medicaid-enrolled Patients
with Schizophrenia Experiencing Multiple Psychiatric Relapses. Health Outcomes
Research in Medicine, 2012. 3(4): p. e183-e194.
54

16. Rascati, K.L., et al., Adherence, persistence of use, and costs associated with second-
generation antipsychotics for bipolar disorder. Psychiatric Services, 2011. 62(9): p.
1032-1040.
17. Nadkarni, A., et al., Medical costs and utilization in patients with depression treated with
adjunctive atypical antipsychotic therapy. Clinicoecon Outcomes Res, 2013. 5: p. 49-57.
18. Tarride, J.-E., et al., Health status, hospitalizations, day procedures, and physician costs
associated with body mass index (BMI) levels in Ontario, Canada. ClinicoEconomics and
outcomes research: CEOR, 2012. 4: p. 21.
19. Blough, D.K., C.W. Madden, and M.C. Hornbrook, Modeling risk using generalized
linear models. Journal of Health Economics, 1999. 18(2): p. 153-171.
20. Manning, W.G., A. Basu, and J. Mullahy, Generalized modeling approaches to risk
adjustment of skewed outcomes data. Journal of Health Economics, 2005. 24(3): p. 465-
488.
21. Manning, W.G. and J. Mullahy, Estimating log models: to transform or not to transform?
Journal of Health Economics, 2001. 20(4): p. 461-494.
22. Garrido, M.M., et al., Choosing models for health care cost analyses: issues of
nonlinearity and endogeneity. Health Services Research, 2012. 47(6): p. 2377-97.
23. Greene, W.H., Econometric Analysis. 7 ed. 2012, Upper Saddle River, NJ, USA: Pearson
Education, Inc.
24. Deb, P., W. Manning, and E. Norton. Preconference Course: Modeling Health Care
Costs and Counts. in ASHE-Cornell University Conference. 2010.
25. Christy, S., B. Sin, and S. Gim, Impact of an Integrated Pharmacy Transitions of Care
Pilot Program in an Urban Hospital. J Pharm Pract, 2015.
55

3.7 Tables
Table 3-1. Descriptive Statistics of 30-Day ITT Population
30-Day Population
Intervention (N=830) Control (N=1005) p-value
Male, (N & %) 269 32.41 335 33.33 0.6751
Race, (N & %)
White 325 39.16 354 35.22
0.0545
Hispanic 292 35.18 414 41.19
Black 110 13.25 107 10.65
Asian 82 9.88 99 9.85
Other 21 2.53 31 3.09
Age groups, (N & %)
24 and Below 66 7.95 65 6.47
0.0642
25-34 104 12.53 139 13.83
35-44 117 14.1 167 16.62
45-54 217 26.14 297 29.55
55-64 275 33.13 291 28.96
65 and Above 51 6.14 46 4.58
Prior hospitalization (N
& %)
269 32.4 158 15.72 <0.0001
Length of stay, Days (Mean & SD) 4.4964 3.8976 3.791 3.5485 <0.0001
Number of medications, (Mean & SD) 6.9711 3.2136 6.7811 2.7855 0.1752
Comorbidities (N & %)

Acute Myocardial Infarction 26 3.13 24 2.39 0.3299
Arrhythmia, Heart Conduction Disorders 18 2.17 13 1.29 0.1478
Blood Disease 18 2.17 12 1.19 0.1014
Heart Failure 51 6.14 19 1.89 <.0001
Other Disorders of The Central Nervous
System
26 3.13 40 3.98 0.3321
Chronic Obstructive Pulmonary Disease 58 6.99 53 5.27 0.1254
Diabetes 54 6.51 38 3.78 0.0077
Digestive Diseases 134 16.14 164 16.32 0.92
Endocrine Disorders, Non-Diabetes 27 3.25 34 3.38 0.8771
Genitourinary Diseases 80 9.64 92 9.15 0.7233
Infectious and Parasitic Diseases 64 7.71 90 8.96 0.3389
Injury and Poisoning 88 10.6 124 12.34 0.2472
Diseases of Musculoskeletal System 54 6.51 71 7.06 0.6366
56

Neoplasms 56 6.75 62 6.17 0.6157
Other Circulatory Disease 85 10.24 81 8.06 0.1051
Respiratory Disorders 56 6.75 54 5.37 0.2174
Skin and Subcutaneous Tissue Diseases 45 5.42 39 3.88 0.116
Symptoms, Signs, and Ill-Defined Conditions 60 7.23 45 4.48 0.0115

Table 3-2. Descriptive Statistics of 180-Day ITT Population
180-Day Population
Intervention (N=558) Control (N=669) p-value
Male, (N & %) 179 32.08 216 32.29 0.9381
Race, (N & %)
White 222 39.78 241 36.02
0.4345
Hispanic 194 34.77 259 38.71
Black 77 13.8 81 12.11
Asian 54 9.8 71 10.64
Other 11 2 17 2.55
Age groups, (N & %)
24 and Below 44 7.89 52 7.77
0.4255
25-34 76 13.62 103 15.4
35-44 86 15.41 110 16.44
45-54 148 26.52 198 29.6
55-64 175 31.36 176 26.31
65 and Above 29 5.2 30 4.48
Prior hospitalization (N
& %)
195 34.95 73 10.91 <0.0001
Length of stay, Days (Mean & SD) 4.45 3.91 3.56 3.31 <0.0001
Number of medications, (Mean & SD) 6.95 3.11 6.85 2.81 0.547
Comorbidities (N & %)
Acute Myocardial Infarction 18 3.23 11 1.64 0.0695
Arrhythmia, Heart Conduction Disorders 8 1.43 7 1.05 0.539
Blood Disease 12 2.15 6 0.9 0.069
Heart Failure 34 6.09 14 2.09 0.0003
Other Disorders of The Central Nervous
System
17 3.05 28 4.19 0.291
Chronic Obstructive Pulmonary Disease 35 6.27 40 5.98 0.831
Diabetes 38 6.81 25 3.74 0.0151
57

Digestive Diseases 99 17.74 99 14.8 0.163
Endocrine Disorders, Non-Diabetes 20 3.58 29 4.33 0.5041
Genitourinary Diseases 62 11.11 75 11.21 0.956
Infectious and Parasitic Diseases 41 7.35 49 7.32 0.9876
Injury and Poisoning 64 11.47 76 11.36 0.9522
Diseases of Musculoskeletal System 32 5.73 47 7.03 0.3594
Neoplasms 32 5.73 30 4.48 0.3198
Other Circulatory Disease 61 10.93 50 7.47 0.0355
Respiratory Disorders 36 6.45 37 5.53 0.4975
Skin and Subcutaneous Tissue Diseases 32 5.73 22 3.29 0.0375
Symptoms, Signs, and Ill-Defined Conditions 39 6.99 23 3.44 0.0047

Table 3-3. Effects of TOC Program on Healthcare Costs, OLS and GLM results
OLS Regressions
30-Day Healthcare Costs 180-Day Healthcare Costs

Marginal
Effects
Standard
Error
P-Value
Marginal
Effects
Standard
Error
P-Value
Inpatient Costs -$961.65 568.28 0.091 -$4,735.74 2247.31 0.035
Medication Costs -$63.15 65.86 0.338 -$531.07 371.35 0.153
Outpatient Costs -$19.90 10.75 0.064 -$175.38 84.6 0.038
ER Costs $3.50 8.47 0.679 $89.09 47.62 0.062
Total Costs -$1,041.21 571.73 0.069 -$5,353.10 2302.77 0.02
GLM Regressions
30-Day Healthcare Costs 180-Day Healthcare Costs

Marginal
Effects
Standard
Error
P-Value
Marginal
Effects
Standard
Error
P-Value
Inpatient Costs -$3,947.54 5,022.04 0.43 -$1,644.15 1,348.39 0.22
Medication Costs -$32.55 52.29 0.53 -$491.94 332.71 0.14
Outpatient Costs -$236.75 398.01 0.55 -$188.91 105.04 0.07
ER Costs -$8.69 9.48 0.36 $35.56 34.78 0.31
Total Costs -$498.26 307.76 0.11 -$2,139.35 1,067.88 0.045

58

Table 3-4. Modified Park Test Results
Costs Estimates Standard Error
30-Day Total Cost 2.139 0.046
30-Day Rx Cost 2.071 0.046
30-Day Inpatient Cost 1.935 0.579
30-Day Outpatient Cost 2.202 0.13
30-Day ER cost 2.299 0.234
180-Day Total Cost 1.974 0.066
180-Day Rx Cost 1.961 0.05
180-Day Inpatient Cost 2.483 0.159
180-Day Outpatient Cost 2.391 0.083
180-Day ER cost 2.248 0.164

Table 3-5. Effects of TOC Program on Healthcare Costs, TPM results
30-Day Healthcare Costs

Two Part Model Two Part Model
(First = Logistic Second=GLM) (First = Logistic Second=OLS)

Marginal
Effects
Standard
Error
P>z
Marginal
Effects
Standard
Error
P>z
Inpatient Costs -$190.69 $319.44 $0.55 -$228.43 $682.25 $0.74
Outpatient Costs -$18.72 $12.19 $0.13 -$25.65 $12.67 $0.04
ER Costs $0.09 $7.30 $0.99 $1.84 $8.65 $0.83
180-Day Healthcare Costs

Two Part Model Two Part Model
(First = Logistic Second=GLM) (First = Logistic Second=OLS)

Marginal
Effects
Standard
Error
P>z
Marginal
Effects
Standard
Error
P>z
Inpatient Costs -$1,723.95 $854.88 $0.04 -$3,607.36 $2,400.29 $0.13
Outpatient Costs -$79.12 $53.28 $0.14 -$199.47 $91.53 $0.03
ER Costs $41.14 $32.69 $0.21 $81.25 $47.15 $0.09

59

3.8 Figures
Figure 3-1. Flow Chart

*Matching by +/- 1 day length of stay and +/- 1 prior hospitalization
Did not match
(n= 11)*
Intervention Group
(n = 1,123)
Matched by LOS and
prior hospitalizations*
(n= 859, 76.5%)
Did not meet 30-day post-index
continuous enrollment (n=83)
Control Group
(n = 1,463)
Matched by LOS and
prior hospitalizations*
(n= 1088, 74.4%)
30-Day ITT
population (n=830,
73.9%)
30-Day control
population
(n=1005, 68.7%)
Did not meet 30-day post-index
continuous enrollment (n=29)
Did not meet 180-day post-index
continuous enrollment (n=336)
180-Day ITT
population (n=558,
49.7%)
180-Day control
population (n=669,
45.7%)
Did not meet 180-day post-index
continuous enrollment (n=272)
Did not match
(n= 0)*
Did not meet 180-day pre-index
continuous enrollment (n=253)
Did not meet 180-day pre-index
continuous enrollment (n=375)
n= 870(77.5%)
n= 1088 (74.4%)
60

Figure 3-2. Patient Disposition

ITT population
(n=830)
Refused (n=33, 3.9%) Population (n=797)
Unable to be contacted
(n=69, 8.3%)
Population receiving
TOC services (n=728)
61

CHAPTER 4
Budget Impact Analysis on Pharmacist-Provided Transitional Care
Program

Abstract
Objectives
To estimate the budget impact of adding an outpatient, pharmacy-based transition of care (TOC)
program to the benefit of a US managed Medicaid health plan.

Methods
A dynamic budget impact analysis was conducted using a decision-tree model developed in
Microsoft Excel. The effect on inpatient and total healthcare costs from the payer perspective was
estimated in the 2-year period following initial hospital discharge. Inputs were based on a total
plan population of 240,000 lives, with a high-risk population of 7.5%. The TOC program was
assumed to initially cover 30% of the high-risk group, with expansion to 60% over the 2 years.
We previously reported that this program reduced readmission within 6 months by 32% and saved
the health plan$2,139 per patient referred to the program, inclusive of program cost, compared to
patients receiving usual discharge care. Sensitivity analyses were performed to test the impact of
uncertainty of model inputs on the results, with the cost of TOC services ranging from $99 to
$2,000 per patient referred.

62

Results
The model showed that the TOC program was cost-saving at over $3 per member per month,
translating into over $25 million in total healthcare cost savings over 2 years. These results were
primarily driven by the estimated reduction in inpatient costs associated with the program, which
were estimated at $20 million over the 2 years. Sensitivity analyses illustrated that within all the
reasonable ranges of model input parameters, including the upper limit of TOC services set to
$2,000 per patient referred, the TOC program resulted in cost savings to the health plan.

Conclusions
The TOC program resulted in cost savings of over $25 million to the health plan over a period of
2 years.

Keywords: Budget Impact Analysis, Pharmacy-Based Transition of Care, Readmissions,
Ambulatory Care

63

4.1 Introduction
Inappropriate use of medications during the transitional care period following inpatient discharge
could possible cause readmission and increasing healthcare costs [1]. Previous studies have
demonstrated that almost 60% of all medications errors occur during the transitional period and
around 11% of patients with complex medication regimens would be experienced an adverse drug
event after hospital discharge [2, 3]. Pharmacist-provided services after discharge, such as
telephone follow-up, discharge counseling, and medication reconciliation have been shown
associated with lower incidence of adverse drug events [4-7]. Thus, increasing medication
monitoring services have been provided to control unavoidable inpatient readmission and
unnecessary healthcare costs [8-11].

Synergy Pharmacy Solutions (SPS) in Bakersfield, California initiated a pilot pharmacist-provided
transitional care program for discharged managed Medicaid members of Kern Health Systems
(KHS) classified as high risk based on their medical history in 2013. Those qualified patients
admitted to a single local hospital were referred to SPS transitional care program. Once a referred
patient agreed to participate in the SPS transitional care program, medications were reconciled and
any discrepancies between the patient’s self-reported medication use and the hospital discharge
orders were noted in SPS electronic health record system. Over the 30 days following discharge,
SPS pharmacy staffs cooperate with the outpatient providers to resolve any medication-related
problems.

Previous companion studies evaluated the effects of the pharmacist-provided services at SPS on
inpatient readmission and healthcare costs at 30 days and 180 days after the discharge [10] [11].
64

These studies demonstrated that the SPS transitional services were associated with reduced 180-
day inpatient readmission, inpatient costs, and total healthcare costs. Hence, it is meaningful to
provide this transitional care model to a larger scope of population from the payer perspective. The
objective of this study is to create a budget impact model to simulate the economics effects of
pharmacy-based transitional care program on the KHS managed Medicaid population if the
transitional care model would be gradually utilized to patients discharged from more participated
hospitals contracted with KHS in the future as planned.

4.2 Methods
4.2.1 Interventions
The population transferred to the transitional care program was selected from the pool of adult
Medicaid managed care members of KHS health plan who were discharged from inpatient stays
and met the following inclusion criteria at the time of discharge:
 Active members of KHS with an inpatient stay at participating hospitals, AND
 “High risk” as determined by KHS’s algorithm including prior healthcare utilization and
social history, OR
 Discharged with prescription claims for >=5 medications, OR
 Recently admitted to any local hospital within the last 45 days
The SPS pharmacy staff attempted to contact the patients or their caregivers within 2 to 4 days
after patient’s inpatient discharge. Once a patient agreed to use the SPS transitional care services,
pharmacists at SPS would review the current clinical status, apply a medication reconciliation
process including an evaluation of discharge and home mediations, remind the follow-up physician
appointments, and solve issues on medication use. Over the 30 days following discharge, the
65

pharmacists also communicated with the outpatient providers to resolve any medication-related
problems, such as inappropriate therapy, therapeutic duplications, and medication interactions.
Patients requiring additional assistance were invited for a face-to-face visit at SPS, in which more
intensive counseling and assistance with organizing medications would be provided. In addition,
the pharmacy staff reinforced the discharge care plan, including facilitating authorizations for
specialist care, arranging transportation for appointments, and working with the patient’s
dispensing pharmacy to resolve insurance-related issues.

All transitional services provided by SPS team were documented directly into the existing
electronic health record (EHR) system at SPS, which included customized reporting capabilities.
Daily reports were generated by the pharmacy team, and follow-up tasks were assigned
accordingly. The clinical pharmacy team acted as a liaison to bridge the communication gaps
between the patient, their prescribers, and their dispensing pharmacy, thereby facilitating
improvements in the continuity of care between the inpatient and outpatient settings.
Patients’ healthcare utilizations were assessable from administrative claims databases at KHS,
which covered inpatient claims, outpatient services, emergency room visits, and medication claims
for services rendered within the United States.

4.2.2 Budget Impact Model
Recently, budget impact analysis becomes an essential part of a comprehensive economic
assessment of the healthcare intervention and is increasing required by decision makers. A budget
impact model could possibly address the expected changes in the expenditure of a healthcare
66

system after adoption of any new interventions and medications [12] [13, 14] [15] [16] [17, 18]
[19].

A dynamic budget impact analysis was conducted using a decision-tree model developed in
Microsoft Excel (Microsoft Corporation, Redmond, Washington, USA). This budget impact model
estimates the economic effects of adopting SPS transitional care program on the KHS managed
Medicaid high risk population from the payer perspective and modeled the economic impacts on
inpatient costs and total healthcare costs, including inpatient, outpatient, medication and ER costs,
of discharged high risk patient in every six months from the beginning of year 1 and calculated the
total budget impacts within a 2-year period (Figure 4-1).

Qualified patients discharged from hospitals were simulated into two scenarios. At the presence of
the SPS pharmacy-based transitional program, patients admitted to participating hospitals will be
referred to the SPS transitional services, while patients from unparticipating hospitals would only
be offered standard of care post discharge. At the absence of pharmacy-based transitional program,
all patients would be merely accessible to standard post-discharge services. The simulated
accumulative total healthcare costs and inpatient costs in two years were compared between the
two scenarios.

4.2.3 Data Inputs
4.2.3.1 KHS Healthcare Plan
Kern Health System healthcare plan covers around 240,000 members and the covered population
increased about 8.7% from December 2015 to December 2016 [20] [21]. According to the criteria
67

of high risk population, about 7.5% of the KHS Medicaid recipients could meet the qualification
of SPS transitional care program and 30% of those patients were exposed to the post-discharge
services. As the expansion of SPS transitional program, the program coverage rate was expected
to be doubled within 2 years (Table 4-1).

4.2.3.2 Healthcare Utilization and Costs
Healthcare utilization and costs of the studied population were estimated using KHS administrative
claims data in the companion studies [10] [11]. The average total healthcare costs and inpatient
costs of the high risk population using standard post-discharge services were $8,383 and $5,650,
respectively. The average cost of the transitional care services per referral patient was $98.88. The
cost saving on the total healthcare costs covering the transitional care program costs within 6
months after discharge with the SPS transitional program was $2,139, inclusive of program, and
the cost saving on inpatient costs within 6 months was $1,723.

4.2.4 Sensitivity Analyses
In order to test the impact of uncertainty with model inputs on the results, two types of sensitivity
analysis were applied to the total and inpatient healthcare costs, respectively.

4.2.4.1 One-Way Sensitivity Analysis for Total Healthcare Costs
In the setting of total healthcare costs, one-way sensitivity analysis was performed to show the
sensitivity range of budget impact according to the variations of key input parameters. As the
indication of the budget impact model, several factors were associated with budget impact. Thus,
in the one-way sensitivity analysis, the inputs of those factors were changed within plausible
68

ranges as calculated from KHS administrative claims database, as seen in the literature and as
advised by clinical experts [10] [11] [20]. The range of KHS population increase rate was set as
±100% of the base value and the range of high risk population and the services coverage increase
rate as set ± 33%. The healthcare cost saving associated with SPS transitional care services ranged
from $1,000 to $6,000 based on one companion study evaluating the effects of transitional care
services on healthcare costs [11]. More, the range of tranistional care cost was set from $50 to
$2,000 in case these costs could change when the services were provided on a larger population
setting.

4.2.4.2 Sensitivity Analysis for Inpatient Costs
The main aim of pharmacy-based transitional care services was to reduce unnecessary
readmissions among high risk population as readmission could be a heavy burden to the healthcare
system. Data analysis using administrative claims provided the estimated avoidable
hospitalizations with SPS transitional care services within 180 days after hospitalization
(Appendix). Meanwhile, the national average inpatient costs were obtained from the Healthcare
Costs and Utilization Project (HCUP) [22]. Applying those numbers in the budget impact model
on inpatient costs instead of the baseline values gave the sense of the robustness of the budget
analytic model if the services provided to a different population.

4.3 Results
The budget impact model on total healthcare showed that in the first half of year one, the SPS
transitional program could result in a cost saving of $4.3 million to the KHS healthcare plan. The
healthcare cost saving per member per month (PMPM) was $3. This total cost saving could reach
69

$26 million if the transitional program could be expanded to more participant hospitals as planed
to the end of year two (Table 4-2). The potential cost saving on inpatient costs could reach $3.5
million in the first half of year one and the 2-year inpatient cost saving would increase to $21
million (Table 4-3).

The one-way sensitivity analysis of budget impact on total healthcare cost was utilized to test the
robustness of the model in the presence of uncertainty of input parameters. Within the ranges of
input parameters, the 2-year total healthcare cost savings from SPS transitional program ranged
from $12 million to $72 million and were most sensitive to the impact of SPS transitional care
service on healthcare costs per referral patients. Moreover, the tornado graph showed that the
budget impact varied associated with the cost of the transitional care program. However, even
though the cost per referral patients increased to $2,000, this transitional care programs could still
lead to a cost saving of over $2.8 million to the healthcare plan in 2 years. The budget impact was
also sensitive to the proportion of high risk patients. Higher proportion of the high risk patient in
the SPS transitional care program covered population, the more cost saving would be expected.
Meanwhile, the predicted savings were also positively associated with hospitalization rate of high
risk patients, the KHS population expansion rate and the increase rate of transitional care coverage
in the 2 years (Figure 4-2).

The sensitivity analysis on inpatient costs was performed using the national average inpatient costs
from HCUP. Comparing to the average inpatient costs calculated from the KHS administrative
data, the national average inpatient cost is higher ($13,655 vs. $8,383). Moreover, a Poisson
regression was performed to estimate the effects of the transitional care program on
70

hospitalizations. According to the regression outputs, every 100 patients receiving the transitional
care services, 20 hospitalizations could be avoided in the following 6 months. Consequently, the
sensitivity analysis illustrated an even higher cost savings that the inpatient cost saving for 6 month
was around $1.8 million and for 2-year period was over $10.5 million (Table 4-4).

4.4 Discussion
Economic effects of the SPS pharmacy-based transitional care program providing medication
reconciliation as well as medication monitoring to discharge patients on healthcare utilization and
costs had been studied [10] [11] . The SPS transitional care services was found to be associated
with reduced readmissions and healthcare costs after inpatient discharge. Consequently, it is
valuable to expand this transitional care model to a larger scope of population and deemed to
evaluate the potential costing savings from the transitional care program in the future setting. Thus,
the current study created a decision tree model to simulate the budget impacts within a horizon of
2 years. More, the current budget impact model was built dynamically to reflect the potential
changes of patient population and the increase of participated hospitals.

As stated in the Method section, the budget impact analysis was modeled from the payer’s
perspective. The main source of input parameters was collected from companion studies in which
impacts of SPS transitional care program on healthcare utilization and costs within 6 months after
discharge were estimated using KHS administrative claims data [10, 11]. In those studies, inpatient
costs, outpatient, medication costs and ER costs were collected from the SPS claims database.
Moreover, the costs of providing SPS transitional care had also been included in the total
healthcare costs in those analyses as the SPS transitional care services were billed as outpatient
71

service in KHS billing system. Thus, the impact on healthcare costs estimated from those studies
was convincible to be applied in the budget impact model to predict the cost savings for expanding
SPS transitional care program.

One-way sensitivity analysis was performed to verify the robustness of the budget impact model
in the presence of uncertainty from input variables. Within all the reasonable ranges of model input
parameters, the outputs showed that providing SPS transitional care services was associated with
positive healthcare cost savings. This sensitivity analysis also demonstrated that cost savings were
sensitive to the amount of patients who qualified for the SPS transitional care program as well as
the patient’s accessibility to the transitional care program.

Besides the sensitivity analysis on total healthcare costs, a sensitivity analysis on the inpatient
costs was performed. As the expansion of the transitional care program, the service recipients
could be from other population rather than Medicaid beneficiaries. Thus, the inpatient costs could
be slightly changed. In this analysis, the impacts of transitional care on 6 months hospitalizations
were estimated using multivariate regression analysis and the national average inpatient costs were
referenced from the Healthcare Costs and Utilization Project [22]. As the national average
inpatient costs were around $5,000 higher than the average from the KHS Medicaid plan, the
sensitivity analysis showed a larger inpatient cost savings if the transitional care program would
be provided to a more diversified patient population.

Our study still contained a number of limitations. First, as a dynamic analysis was required, the
change of KHS healthcare plan coverage, the variation of beneficiary population and the increasing
72

service accessibility need to be applied in the model. However, those inputs had to be either
estimated from data analysis in previous years or use the numbers from KHS transitional care
development plans. More, as time changes, the fee schedule of healthcare services and medication
costs could also change. Consequently, the estimated effects of the SPS transitional on healthcare
costs using data from 2013-2015 could be not accurate for the future settings. Although current
budget impact model cannot avoid those uncertainties, the sensitivity analysis provided evidence
that variations of those variables within wide ranges didn’t change the cost savings significantly.

Another potential issue on the modeling was that the additional costs of expanding the transitional
care services were not included. As the current setting, SPS pharmacy provided services to patients
discharged from one local hospital. If more patients would be referred to SPS for the transitional
care services, the operation costs of the transitional care services could increase. Consequently,
this change could also possibly change the predicted cost saving. As the costs of per referral
patients were utilized as an input variable in the one-way sensitivity analysis, the uncertain budget
impact because of the additional operation costs were expected to be simulated in the sensitivity
analytic outputs.

Finally, one sensitivity analysis applied the national average inpatient costs from the HCUP project
[22]. This project was conducted in 2013 and the costs could be different from currently updated
hospital settings. If more recent studies have been conducted on similar topic, these numbers
should be updated.

73

4.5 Conclusions
The budget analysis showed that gradually expanding the SPS transitional program to KHS
covered Medicaid population will lead to a cost saving of over $25 million from payer perspective
within 2 years. It is meaningful to provide those evidences to decision makers for the development
of current transitional care program.

4.6 References
1. O'Sullivan D, O'Mahony D, O'Connor MN, Gallagher P, Cullinan S, O'Sullivan R,
Gallagher J, Eustace J, Byrne S (2014) The impact of a structured pharmacist
intervention on the appropriateness of prescribing in older hospitalized patients. Drugs
Aging 31:471-481
2. American Pharmacists A, American Society of Health-System P, Steeb D, Webster L
(2012) Improving care transitions: optimizing medication reconciliation. J Am Pharm
Assoc (2003) 52:e43-52
3. Forster AJ, Murff HJ, Peterson JF, Gandhi TK, Bates DW (2005) Adverse drug events
occurring following hospital discharge. J Gen Intern Med 20:317-323
4. Al-Rashed SA, Wright DJ, Roebuck N, Sunter W, Chrystyn H (2002) The value of
inpatient pharmaceutical counselling to elderly patients prior to discharge. Br J Clin
Pharmacol 54:657-664
5. Dudas V, Bookwalter T, Kerr KM, Pantilat SZ (2001) The impact of follow-up telephone
calls to patients after hospitalization. Am J Med 111:26S-30S
74

6. Szkiladz A, Carey K, Ackerbauer K, Heelon M, Friderici J, Kopcza K (2013) Impact of
pharmacy student and resident-led discharge counseling on heart failure patients. J Pharm
Pract 26:574-579
7. Walker PC, Bernstein SJ, Jones JN, Piersma J, Kim HW, Regal RE, Kuhn L, Flanders SA
(2009) Impact of a pharmacist-facilitated hospital discharge program: a quasi-
experimental study. Arch Intern Med 169:2003-2010
8. Jackson C, Kasper EW, Williams C, DuBard CA (2015) Incremental Benefit of a Home
Visit Following Discharge for Patients with Multiple Chronic Conditions Receiving
Transitional Care. Popul Health Manag
9. Stauffer BD, Fullerton C, Fleming N, Ogola G, Herrin J, Stafford PM, Ballard DJ (2011)
Effectiveness and cost of a transitional care program for heart failure: a prospective study
with concurrent controls. Arch Intern Med 171:1238-1243
10. Ni W, Colayco DC, Hashimoto J, Komoto K, Wearda B, McCombs J (Forthcoming)
Impact of a Pharmacy-Based Transitional Care Program on Hospital Readmissions.
American Journal of Managed Care
11. Ni W, Colayco DC, Hashimoto J, Komoto K, Wearda B, McCombs J (Submitted)
Reducing Healthcare Costs through a Community Pharmacy-Based Transitional Care
Program.
12. Sullivan SD, Mauskopf JA, Augustovski F, Jaime Caro J, Lee KM, Minchin M,
Orlewska E, Penna P, Rodriguez Barrios JM, Shau WY (2014) Budget impact analysis-
principles of good practice: report of the ISPOR 2012 Budget Impact Analysis Good
Practice II Task Force. Value Health 17:5-14
75

13. Mauskopf JA, Sullivan SD, Annemans L, Caro J, Mullins CD, Nuijten M, Orlewska E,
Watkins J, Trueman P (2007) Principles of good practice for budget impact analysis:
report of the ISPOR Task Force on good research practices--budget impact analysis.
Value Health 10:336-347
14. Marshall DA, Douglas PR, Drummond MF, Torrance GW, Macleod S, Manti O, Cheruvu
L, Corvari R (2008) Guidelines for conducting pharmaceutical budget impact analyses
for submission to public drug plans in Canada. Pharmacoeconomics 26:477-495
15. Trueman P, Drummond M, Hutton J (2001) Developing guidance for budget impact
analysis. Pharmacoeconomics 19:609-621
16. Mauskopf JA, Earnshaw S, Mullins CD (2005) Budget impact analysis: review of the
state of the art. Expert Rev Pharmacoecon Outcomes Res 5:65-79
17. van de Vooren K, Duranti S, Curto A, Garattini L (2014) A critical systematic review of
budget impact analyses on drugs in the EU countries. Appl Health Econ Health Policy
12:33-40
18. Mauskopf J, Chirila C, Birt J, Boye KS, Bowman L (2013) Drug reimbursement
recommendations by the National Institute for Health and Clinical Excellence: have they
impacted the National Health Service budget? Health Policy 110:49-59
19. Mauskopf JA, Tolson JM, Simpson KN, Pham SV, Albright J (2000) Impact of
zidovudine-based triple combination therapy on an AIDS drug assistance program. J
Acquir Immune Defic Syndr 23:302-313
20. (2016) Kern Health Systems Annual Report.
http://www.kernfamilyhealthcare.com/files/KHS%20Board%20packet%20for%20Dec.%
2015,%202016.pdf Accessed 12/21/2016
76

21. Gowda C (2015) Kern Health System Transition of Care Model Development Report.
22. Agency for Healthcare Research and Quality. All patient readmissions within 30 days:
national statistics, 2013
http://hcupnet.ahrq.gov/HCUPnet.jsp?Id=D8072526039C960C&Form=DispTab&JS=Y
&Action=Accept August 17, 2016
77

4.7 Tables
Table 4-1. Budget Model Input Parameters
Variables Input Parameters References
Kern Health Plan

Total KHS population 240,000 [20] [21]
KHS population increase rate (per 6 months) 4.3% [20] [21]
High risk percentage 7.5% [20] [21]
Transitional care service coverage percentage 10.0% [20] [21]
Transitional care coverage increase rate (per 6 months) 7.5% [20] [21]
Healthcare Utilization and Costs

Hospitalization admission rate of high risk population (per 6 months)
37% [10]
Average 6-month healthcare cost with standard post-discharge care
$8,383.14 [11]
Average 6-months inpatient cost with standard post-discharge care
$5,650.20 [11]
6-month healthcare cost savings with SPS transitional care
$2,139.35 [11]
6-months inpatient healthcare cost savings
$1,723 [11]
Average TOC costs per patient
$98.88 [11]
National average inpatient costs
$13, 655 [22]

78

Table 4-2. Budget Impact Analysis on Total Healthcare Costs
Year 1 First 6 Months Year 1 Second 6 Months Year 2 First 6 Months Year 2 Second 6 Months Total
Scenario 1: With Transitional Care
Standard Care $39,519,970.71 $36,802,972.73 $33,779,240.49 $30,427,418.58 $140,529,602.51
Transitional Care $12,614,829.86 $16,446,584.44 $20,584,545.08 $25,047,960.61 $74,693,919.99
Total Costs $52,134,800.58 $53,249,557.16 $54,363,785.57 $55,475,379.18 $215,223,522.49
Scenario 2: No Transitional Care

Standard Care $56,457,101.02 $58,884,756.36 $61,416,800.89 $64,057,723.32 $240,816,381.59
Budget Impact -$4,322,300.44 -$5,635,199.20 -$7,053,015.32 -$8,582,344.14 -$25,592,859.10
PMPM -$3.00 -$3.75 -$4.50 -$5.25 -$4.17

Table 4-3. Budget Impact Analysis on Inpatient Costs
Year 1 First 6 Months Year 1 Second 6 Months Year 2 First 6 Months Year 2 Second 6 Months Total
Scenario 1: With Transitional Care
Standard Care $26,636,288.85 $24,805,043.99 $22,767,061.58 $20,507,948.15 $94,716,342.57
Transitional Care $7,932,517.87 $10,342,020.17 $12,944,072.45 $15,750,778.83 $46,969,389.32
Total Costs $34,568,806.72 $35,147,064.17 $35,711,134.03 $36,258,726.98 $141,685,731.89
Scenario 2: No Transitional Care

Standard Care $38,051,841.22 $39,688,070.39 $41,394,657.42 $43,174,627.68 $162,309,196.71
Budget Impact -$3,483,034.49 -$4,541,006.22 -$5,683,523.39 -$6,915,900.71 -$20,623,464.81
PMPM -$2.42 -$3.02 -$3.63 -$4.23 -$3.36

79

Table 4-4. Sensitivity Analysis on Budget Impact of Inpatient Costs
Year 1 First 6 Months Year 1 Second 6 Months Year 2 First 6 Months Year 2 Second 6 Months Total
Scenario 1: with Transitional Care

Standard Care $40,876,652.67 $38,066,382.80 $34,938,848.79 $31,471,962.11 $145,353,846.36
Transitional Care $12,083,671.90 $15,754,087.24 $19,717,815.59 $23,993,295.27 $71,548,870.01
Total Costs $52,960,324.57 $53,820,470.04 $54,656,664.38 $55,465,257.38 $216,902,716.38
Scenario 2: No Transitional Care

Standard Care $58,395,218.10 $60,906,212.48 $63,525,179.61 $66,256,762.34 $249,083,372.53
Budget Impact -$5,434,893.53 -$7,085,742.44 -$8,868,515.23 -$10,791,504.95 -$32,180,656.15
PMPM -$3.77 -$4.72 -$5.66 -$6.60 -$5.24

80

4.8 Figures
Figure 4-1. Budget Impact Model

81

Figure 4-2. One-way Sensitivity Analysis of Budget Impact Model

-$80,000,000.00 -$60,000,000.00 -$40,000,000.00 -$20,000,000.00 $0.00
Average TOC Costs per Patient
KHS Population Increase Rate
TOC Coverage Increase Rate
Hospitalization Admission Rate
High Risk Percentage
Estimated Cost Savings
Budget Impact ($)
82

4.9 Appendix - Data Analysis for Sensitivity Analytic Input Parameters
As the impacts of the SPS transitional care services on hospitalizations within 6 months after
discharge was needed in the sensitivity analysis, multivariate regressions were performed using
the KHS administrative claims databases.

4.9.1 Data Collection
Information of patients’ demographics (age, gender, and race), inpatient records, inpatient stay
diagnoses, and prescription medications were collected from KHS claims database. Patients’
responses to the transitional care program were recorded in SPS electronic health record system.

4.9.2 Analytic Population
The intervention and control patient populations were selected from the pool of adult Medicaid
managed care members of KHS health plan who were discharged from either the study hospital or
control hospitals. For the intervention patients referred to the SPS post-discharge care, an index
date was defined as the discharge date of their hospitalization immediately preceding referral. The
control group was identified ex post by applying the KHS risk screening algorithm to its members
discharged from neighboring hospitals during the period April 2013 to March 2015. In order to
identify an index hospitalization for each control group patient, all hospitalizations for each patient
were converted into episodes of hospital care and matched to the index episode in an intervention
group member based on the number of prior hospitalizations and length of stay (LOS) [±1 day].
To maximize the statistical power of the analysis, we included all the patients from the control
group that were matched to at least one patient in the intervention group. Consequently,
83

intervention patients could have more than one matched control hospitalization in the final study
population.

Comparison was conducted on 30 days and 180 days after discharge, respectively. Continuous
enrollment was required during the observation period. As the results, 830 patients from the
intervention group were matched to 1,005 patients receiving usual care for the 30-day analysis and
558 SPS patients were matched to 669 usual care patients for the 180-day analysis. In the
intervention group, all patients referred to SPS were included in this intent-to-treat (ITT) analysis,
including those who did not qualify for services, who could not be contacted, and who declined
services.

4.9.3 Outcome Measures
The primary outcome measures were the counts of inpatient stays with 30-day and 180-day after
index hospitalization, which were defined as the count of inpatient admissions within 30 days or
180 days after the index hospitalization discharge date.

4.9.4 Statistical Analysis
Multivariate ordinary least square and Poisson regression were used to estimate the impact of the
pharmacist-provided services on the likelihood of a 30-day or 180-day hospitalizations. All models
were controlled for age, gender, race, prior hospitalizations (yes/no), length of stay of the index
hospitalization, inpatient diagnoses prior to and including the index hospitalization, and the mix
of medication classes used by the patient over the six months prior to admission.

84

4.9.5 Results
Both of the two models demonstrated that the transitional care program was associated with
reduced hospitalizations. The TOC program reduced the number of readmissions by 6 per 100
patients within 30 days and by 19 per 100 patients within 180 days, compared to patients receiving
usual care (Appendix Table 4.1). These results are consistent the estimates from the multivariate
logistic regressions.

Appendix Table 4-1. Impact of TOC Program on Hospitalizations
Marginal Effects 95% Confidence Limits P-Value

OLS Regression
30-Day -0.061 -0.104 -0.018 0.006
180 Day -0.196 -0.321 -0.068 0.003

Poisson Regression
30-Day -0.056 -0.096 -0.016 0.006
180 Day -0.197 -0.298 -0.097 <0.001

85

CHAPTER 5
Summary and Future Directions

Hospital readmission can have negative consequences for patients and costly for both public and
private payers. A report from healthcare cost and utilization project demonstrate that 18% of all
inpatient admissions paid by Medicare were readmitted within 30 days, accounting for over $15
billion annually [1]. Beyond the issue of costs, repeat hospitalizations may place patients at risk
for hospital acquired infection and stress.

Although it may be necessary to readmit some patients, certain readmissions may be prevented
through providing additional interventions from either hospitals or other healthcare providers.
Previous studies have shown that suboptimal medication therapy during the transitional care period
following hospital discharge could be a major contributory factor to hospital readmissions and
consequently increase healthcare utilization [2]. Forster and colleagues estimated that 11% of
patients experienced an adverse drug event after discharge from inpatient services [3]. In another
study, 20% of readmissions were considered to be preventable if the patient had received
appropriate medication monitoring after discharge [3].

The post-discharge services that have been traditionally provided to improve hospital readmissions
include patient education, medication adherence counseling, and medication reconciliation [4-9].
Pharmacist involvement in discharge counseling, medication reconciliation, and telephone follow-
up was found with a lower incidence of adverse drug events [5-8]. Consequently, it is reasonable
86

to hypothesize that pharmacist-provided transitional care services may effectively decrease
readmission rates.

Pharmacists at Synergy Pharmacy Solutions (SPS) in Bakersfield, California initiated an outpatient
pharmacy-based transitional care program in 2013. This program provided services to members of
Kern Health Systems (KHS) managed Medicaid health plan at high risk. The risk level of
individuals for readmission was evaluated based on their history of hospitalizations, prescription
medication utilization, and social history. High-risk patients discharged from Bakersfield
Memorial Hospital in the study period were automatically referred to the SPS transitional care
program.

Over 30 days after discharge, pharmacists at SPS worked with the outpatient providers to resolve
any medication-related problems, such as inappropriate therapy, therapeutic duplications, and
potential drug interactions. Additionally, the pharmacy staff reinforced the discharge care plan,
including post-discharge appointments, facilitating authorizations for specialist care, arranging
transportation for appointments, and working with the patient’s dispensing pharmacy to resolve
insurance-related issues.

Once a patient agreed to use SPS transitional care services, medications were reconciled and any
discrepancies between the patient’s self-reported medication use and the hospital discharge orders
were noted. Over the 30 days following discharge, the SPS service team cooperated with the
outpatient providers to resolve any medication-related problems, such as inappropriate therapy,
therapeutic duplications, and potential drug interactions. In addition, SPS pharmacists counseled
87

patients to improve medication adherence. The pharmacy staff also reinforced the discharge care
plan, including post-discharge appointments, facilitating authorizations for specialist care,
arranging transportation for appointments, and working with the patient’s dispensing pharmacy to
resolve insurance-related issues. Medication management services were documented into the
existing SPS electronic health record (EHR) system, which included customized reporting
capabilities. Daily reports were generated by the pharmacy staff, and follow-up tasks were
assigned accordingly. The clinical pharmacy team acted as a liaison to bridge the communication
gaps between the patient, their prescribers, and their dispensing pharmacy, thereby facilitating
improvements in the continuity of care between the inpatient and outpatient settings.

We evaluated the effect of the transition of care (TOC) services on healthcare utilization and costs
using a retrospective cohort study designed and conducted using claims administrative data from
KHS and electronic health records from SPS. The intervention group of patients referred to SPS
for the TOC services was compared against a control group of matched KHS members discharged
from neighboring hospitals. The impact of the TOC program on hospital readmission and
healthcare costs with 30 and 180 days after discharge were estimated using multivariable
regressions and survival analysis controlling for demographics, comorbidities, and prior healthcare
utilization. Those analyses demonstrated that the TOC program was associated with a reduction of
28% on 30-day readmission. The reduced readmission within 180 days increased to 32% and saved
the health plan around $2,139 per patient referred to the program.

As this TOC model gradually expands to a larger population setting covered by KHS healthcare
plan, a dynamic budget impact analysis was conducted using a decision-tree model. The economic
88

effect on inpatient costs and total healthcare costs of high risk patient from payer perspective was
estimated. The simulated outputs showed that the TOC program was cost-saving at over $3 per
member per month, translating into over $25 million in total healthcare cost savings over 2 years.

Both administrative claims data analysis and budget impact predictions clearly showed that the
pharmacy-based transitional care model at Synergy Pharmacy Solutions resulted in reduction in
healthcare utilization and costs. It is meaningful to provide those services to a larger scope of
population. However, hurdles may exit to the implementation of the current transitional care model.
Provision of this type of services is common place in closed health care systems where all providers
are paid based on capitation. Systems like Kaiser Permanente have merged the role of physicians
with the role of the insurance provider and definitive have the incentive to adopt new services that
can improve patient outcomes and reduce costs. However, hospitals reimbursed by insurance plans
applying fee-for-services payment systems may not have any incentive to participate in such
transitional care program. It is a challenge to the transitional care program because health records
from the discharge hospital are critical for medication reconciliation at SPS. Fortunately, payers
including Medicare and Medicaid establish programs to control the unnecessary readmission and
imposes financial penalty to hospitals with excess rate of readmissions for certain diseases. Other
private payers also have regulations to deny reimbursement to preventable 30-day readmission.
Those policies impose the need of post-discharge services that improve readmission from the
hospital perspective.

Moreover, SPS transitional care services were paid per service by KHS healthcare plan. Charges
for follow-up phone calls to patients, clinical onsite interviews and communication with outpatient
89

providers could cause high bills to payers. Hence, other payers rather than KHS need to be
comfortable that the SPS transitional services provided were needed and effective. Two strategies
are available to solve this difficulty. First, SPS pharmacist team may distribute results from current
studies as supporting evidence to potential payers in order to trigger their interests. Second, a pay-
for-performance payment system should be utilized instead of fee-for-service payment. Under pay-
for-performance system, the transitional care services would be paid when the patient reached 30
days post admission or the provider could be paid a flat rate per referral and pay back a set fee if
the patient is readmitted.

This dissertation demonstrated the value of providing transitional care model in current setting. To
apply this transitional care model in a larger population setting, cooperation among healthcare
providers, payers and SPS pharmacy team is deemed to fulfil the achievement.

References
1. Hines, A.L., et al., Conditions With the Largest Number of Adult Hospital Readmissions
by Payer, 2011: Statistical Brief #172, in Healthcare Cost and Utilization Project
(HCUP) Statistical Briefs. 2006: Rockville (MD).
2. O'Sullivan, D., et al., The impact of a structured pharmacist intervention on the
appropriateness of prescribing in older hospitalized patients. Drugs Aging, 2014. 31(6):
p. 471-81.
3. Forster, A.J., et al., Adverse drug events occurring following hospital discharge. J Gen
Intern Med, 2005. 20(4): p. 317-23.
90

4. Kucukarslan, S.N., et al., Pharmacists on rounding teams reduce preventable adverse
drug events in hospital general medicine units. Arch Intern Med, 2003. 163(17): p. 2014-
8.
5. Al-Rashed, S.A., et al., The value of inpatient pharmaceutical counselling to elderly
patients prior to discharge. Br J Clin Pharmacol, 2002. 54(6): p. 657-64.
6. Dudas, V., et al., The impact of follow-up telephone calls to patients after hospitalization.
Am J Med, 2001. 111(9B): p. 26S-30S.
7. Szkiladz, A., et al., Impact of pharmacy student and resident-led discharge counseling on
heart failure patients. J Pharm Pract, 2013. 26(6): p. 574-9.
8. Walker, P.C., et al., Impact of a pharmacist-facilitated hospital discharge program: a
quasi-experimental study. Arch Intern Med, 2009. 169(21): p. 2003-10.
9. Warden, B.A., et al., Pharmacy-managed program for providing education and
discharge instructions for patients with heart failure. Am J Health Syst Pharm, 2014.
71(2): p. 134-9.

Abstract (if available)

Abstract Objectives: Pharmacist interventions may prevent adverse events after hospital discharge and reduce the risk of readmission as well as healthcare costs. This study evaluates the impact of pharmacy-based transitional care services on healthcare utilization and costs for members of a US managed Medicaid health plan. ❧ METHODS: Synergy Pharmacy Solutions (SPS) provided pharmacy-based transitional care services for high risk hospital discharges covered by Kern Health Systems (KHS) managed Medicaid plan. Over 1,100 qualifying discharged patients from Bakersfield Memorial Hospital were referred to SPS. A control group of matched KHS discharges from neighboring hospitals were then identified by matching each to a hospitalization episode in the experimental group based on the number of prior hospitalizations and length of stay [±1 day]. ❧ Thirty-day and 180-day readmissions were analyzed using logistic regression and Cox proportional hazards models. Thirty-day and 180-day inpatient, outpatient, prescription drug, emergency room, and total healthcare costs were analyzed by ordinary least squares (OLS) and generalized linear model (GLM) regressions. Demographics, clinical characteristics, and comorbidity profiles were used as independent variables in all models. ❧ A dynamic budget impact model reflecting possible changes in model parameters over the prediction period was developed. Effects of the expansion of the transitional care services on inpatient and total healthcare costs from the payer perspective were stimulated in a 2-year period following initial hospital discharge. ❧ RESULTS: We found a 28.0% reduction in 30-day readmissions and a 31.9% reduction in 180-day readmissions. Cox model on time to readmission estimated the overall hazard ratio of readmission at 0.749. The GLM model showed a $2,139 reduction in 180-day total healthcare costs. The budget impact analysis predicted that the transitional care program could be cost-saving at over $3 per member per month, translating into over $25 million in total healthcare cost savings over 2 years. ❧ CONCLUSIONS: The analyses demonstrated that patients receiving pharmacy-based transitional care services achieved reduced readmission rates and lower healthcare costs compared with those who received usual care. Consequently, the expansion of the SPS transitional care program was predicted to result in cost savings of over $25 million in a 2-year period. These estimates can be used by health plans to make budgetary decisions on cost-saving programs.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Selected papers on the evaluation of healthcare costs of prematurity and necrotizing enterocolitis using large retrospective databases

PDF

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

PDF

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

PDF

Documenting the impact of under-treatment for chronic disorders

PDF

Topics in U.S. healthcare

PDF

Predictive performance of asthma quality of care measures – a comparison under skewed outcome and cost distributions

PDF

Economic aspects of obesity

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

Value in health in the era of vertical integration

PDF

Advances and applications for economic evaluation methods in health technology assessment (HTA)

PDF

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

PDF

Crohn’s disease: health outcomes and resource utilization in the biologic era

PDF

Three essays on emerging issues in hemophilia care

PDF

The impact of adherence to guidelines on the health care expenditures of COPD patients

PDF

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

PDF

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

PDF

Developing an agent-based simulation model to evaluate competition in private health care markets with an assessment of accountable care organizations

PDF

Thesis on Medicare Part D

PDF

Predicting hospital length of stay (LOS) using the National Inpatient Sample

PDF

Value based purchasing: decision-making processes underlying hospital acquisitions of orthopedic devices

Asset Metadata

Creator Ni, Weiyi (author)

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

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Pharmaceutical Economics and Policy

Degree Conferral Date 2017-05

Publication Date 03/30/2017

Defense Date 02/14/2017

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

Tag budget impact analysis,healthcare costs,OAI-PMH Harvest,pharmacy-based transitional care,readmissions

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McCombs, Jeffrey (committee chair), Florea, Naomi (committee member), Romley, John (committee member)

Creator Email niweiyi2012@gmail.com,weiyini@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC11258203

Unique identifier UC11258203

Identifier etd-NiWeiyi-5131.pdf (filename)

Legacy Identifier etd-NiWeiyi-5131

Dmrecord 349828

Document Type Dissertation

Format theses (aat)

Rights Ni, Weiyi

Internet Media Type application/pdf

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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

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

Repository Email cisadmin@lib.usc.edu