You are on page 1of 12

ORIGINAL ARTICLE

The Impact of Rurality on 30-Day Preventable Readmission,


Illness Severity, and Risk of Mortality for Heart Failure Medicare
Home Health Beneficiaries
Hsueh-Fen Chen, PhD;1 Erin Carlson, Dr Ph;1 Taiye Popoola, MBBS, MPH;2 & Sumihiro Suzuki, PhD3
1 Department of Health Management and Policy, School of Public Health, University of North Texas Health Science Center, Fort Worth, Texas
2 Department of Health Policy and Management, School of Medicine, University of Kansas Medical Center, Kansas City, Kansas
3 Department of Biostatistics and Epidemiology, School of Public Health, University of North Texas Health Science Center, Fort Worth, Texas

Abstract
Purpose: To examine the impact of rurality on 30-day preventable readmis-
Disclosures: The study was approved by the sion, and the illness severity and risk of mortality for 30-day preventable read-
IRB at the University of North Texas Health missions.
Science Center. No potential conflicts of Methods: We analyzed heart failure Medicare beneficiaries who received
interest are reported. The findings from this
home health services for postacute care after hospital discharge in 2009. The
study were presented at the annual meeting of
study was a cross-sectional design with the unit of analysis as the home health
the AcademyHealth, Minneapolis, Minnesota,
June 14, 2015. episode for postacute care. Data sources included the following: Medicare Ben-
eficiary Summary File, Medicare Provider Analysis Review, Outcome Assess-
Funding: This study was supported by the ment Information Set, Home Health Agency Research Identifiable File, and
Junior Faculty Seed Fund at the University of Area Health Resources File. The dependent variables were 30-day preventable
North Texas Health Science Center (UNTHSC). readmission, and the extreme/major level of illness severity and of risk of mor-
The authors thank the 3M Company, which tality for a 30-day preventable readmission. The key independent variable was
provided the 3M All Patient Refined Diagnosis
rurality defined as remote rural, adjacent rural, and micropolitan areas, with
Related Groups (APR-DRGs) software and
urban areas in the reference group.
technical support. The views expressed in this
publication are the views of the authors and do Findings: Home health beneficiaries in remote rural areas had 27% lower
not necessarily reflect the views of the 3M 30-day preventable readmission than those in urban areas. Home health ben-
Company or the UNTHSC. eficiaries in adjacent rural areas were 33% less likely to have high illness sever-
ity at readmission due to a preventable condition than those in urban areas.
For further information, contact: Hsueh-Fen
Chen, PhD, Department of Health Management Conclusions: Geographical location affects preventable readmission and ill-
and Policy, School of Public Health, University of ness severity of preventable readmission. Patients’ geographic location along
North Texas Health Science Center, 3500 Camp with patients’ risk factors should be taken into consideration in the risk adjust-
Bowie Blvd. EAD 601S, Fort Worth, TX 76107; ment model for the financial incentive program that penalizes home health
e-mail: Hsueh-Fen.Chen@unthsc.edu. agencies with high preventable readmissions.
doi: 10.1111/jrh.12142
Key words home health, illness severity, preventable readmission, risk of
mortality, rural.

The home health industry is uniquely positioned to pro- the Affordable Care Act (ACA) to reduce readmissions.
vide professional assistance to help patients transition Additionally, the number of home health episodes (an
from facilities, such as a hospital, to their homes. To qual- episode is defined as 60 days of home health care) pre-
ify for the Medicare home health benefit, Medicare bene- ceded by being in a health care facility increased from
ficiaries must be homebound, require intermittent skilled 1.9 million in 2001 to 2.3 million in 2011.1 It is expected
care, and receive physician referral; thus, Medicare home that home health services would help patients safely stay
health beneficiaries should be distinguished from general in the community and reduce the utilization of acute care
Medicare beneficiaries.1 The role of home health on pro- hospitals; however, 29% of Medicare beneficiaries who
viding postacute care is critical given the priority under received home health services for postacute care were

176 The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association
Chen et al. Rurality and Postacute Care for Home Health

readmitted to hospital.1 To achieve the ACA’s goal in re- Readmission Reduction Program (HRRP).15 We chose
ducing readmissions and improve quality of home health preventable readmissions rather than all-cause readmis-
care, the Medicare Payment Advisory Commission (Med- sions because, as discussed previously, HHAs with high
PAC) has recommended a home health financial incen- preventable readmission rates will face financial penalty
tive program to penalize home health agencies (HHAs) under the home health financial incentive program.1
with high preventable readmission rates.1
Evidence has shown that rural populations have a
higher risk of preventable hospitalization than urban Methods
populations because of the lack of health care infrastruc- Study Design and Data Sources
ture in rural areas.2-4 Approximately 18% of 6.1 mil-
lion home health episodes were provided in rural areas The study was a cross-sectional study using data from
in 2011.1 Given long driving distances and shortages of 2009. The unit of analysis was the home health episode
health care resources in rural areas,5-8 the reduction in for postacute care. The data sources included several files
preventable readmission for the rural elderly can be chal- that have patient- and community-level data. Patient-
lenging. Home health professionals may face difficulty in level data included the Master Beneficiary Summary File
arranging health care resources to meet patients’ posta- to identify Medicare beneficiaries who were enrolled in
cute care needs. Previous studies indicated that rural ben- the fee-for-service (FFS) program; the Medicare Provider
eficiaries are less likely to access home health services Analysis Review (MedPAR) file to provide patients’
than urban beneficiaries, and rural home health bene- demographics and clinical conditions, such as age and
ficiaries have poorer outcomes than urban home health diagnoses, during the hospitalizations; the Outcome As-
elderly.9,10 As 88% of HHAs are for-profit organizations sessment Information Set (OASIS) B1 version to provide
which thereby pursue profit maximization,1 the home patients’ social, clinical, and functional conditions at
health financial incentive program may discourage HHAs the time when patients were admitted to home health
from providing postacute care for the rural elderly if they care; and the Home Health Agency Research Identifiable
are at high risk of preventable readmission. Thus, inves- File (HHA RIF) to provide the number of home health
tigating the association between rurality and preventable visits that patients received during the home health
readmission is imperative under the home health finan- episode. The data at the community-level included the
cial incentive program proposed by MedPAC. Area Health Resources File (AHRF). It provides county
Previous studies showed that greater distance or characteristics, such as the number of PCPs, for the
delayed treatment increases the severity of illness and counties where patients resided.
mortality.11-13 If the rural elderly experience treatment
delays due to difficulty accessing primary care physicians
Study Sample
(PCPs) or referrals to specialists for follow-up after
hospital discharge, an elevated severity of illness and risk There were 2 study samples in the study. The first study
of mortality can be expected upon hospital readmission. sample was heart failure Medicare beneficiaries who re-
Despite MedPAC’s national focus to reduce preventable ceived home health services for postacute care within
readmission for Medicare home health beneficiaries, 14 days of hospital discharge, which was used to ex-
there is a paucity of research examining the impact of amine the association between rurality and 30-day pre-
rurality on quality of postacute care for Medicare home ventable readmission. These beneficiaries can be directly
health beneficiaries. referred by physicians in the hospital or referred by pa-
The purpose of this study was to address 2 research tients’ PCPs anytime within 14 days of hospital discharge.
questions: (1) Is there an association between rurality The HRRP identifies the International Classification of
and the likelihood of 30-day preventable readmission for Diseases, 9th Revision, Clinical Modification (ICD-9-CM)
heart failure Medicare home health beneficiaries, and (2) codes for heart failure admissions.16 Following the HRRP,
What is the impact of rurality on the severity of illness we used these ICD-9-CM codes from the primary diag-
and risk of mortality for 30-day preventable readmission nosis to identify heart failure admission at the index ad-
for heart failure Medicare home health beneficiaries? mission from MedPAR. The 14-day criterion to define
We focused on heart failure and 30-day preventable home health beneficiaries for postacute care is commonly
readmissions for the following reasons: heart failure has used in OASIS assessment and previous studies.1,17,18 We
the highest 30-day readmission rate among all diseases used the date of hospital discharge from MedPAR and the
for Medicare beneficiaries14 and is one of the diseases date of start of home health care from OASIS to ver-
that the Medicare program uses to calculate 30-day ify home health services after hospital discharge, sug-
readmission rates to penalize hospitals under the Hospital gested by Wolff et al.18 Beneficiaries who experienced

The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association 177
Rurality and Postacute Care for Home Health Chen et al.

30-day readmission without a preventable condition because patients with this condition significantly lose
were excluded. In addition to the criteria discussed above, functioning of their gallbladder; however, the level of
additional inclusion criteria were being 65 years or older, risk of mortality for these patients is low because they
enrolling in the FFS program, and being discharged from are less likely to die. Although all study samples had
hospitals before December 1, 2009. Patients whose dates heart failure, their risk of mortality varied. For example,
of admission and discharge were the same were excluded. the risk of mortality for patients with heart failure is
The second study sample was extracted from the first lower than that for those with heart failure and having
study sample but restricted to only those who experi- respiratory failure as the secondary diagnosis. Because
enced a 30-day preventable readmission. This study sam- the percentage of patients whose severity of illness
ple was used to examine the impact of rurality on the and risk of mortality at the extreme or minor level is
severity of illness and risk of mortality for a 30-day pre- relatively small, we created a dichotomous variable to
ventable readmission. represent patients with extreme or major illness severity,
and categorized patients with moderate or minor illness
severity as the reference group. A similar procedure was
Dependent Variables
applied to create a dichotomous variable to represent
There were 3 outcome variables of interest: (1) a 30-day patients with the extreme or major level of risk of
preventable readmission; (2) the extreme/major level of mortality, with patients whose risk of mortality was at
illness severity for a 30-day preventable readmission; and the moderate or minor level as the reference group.
(3) the extreme/major level of risk of mortality for a 30-
day preventable readmission. Preventable readmissions
Key Independent Variables
within 30 days of hospital discharge were extracted from
MedPAR by applying the Prevention Quality Indicator The key independent variable of interest was the degrees
Software Version 4.5 from the Agency for Healthcare of rurality. The US Department of Agriculture and Eco-
Research and Quality.19 These conditions were diabetes nomic Research Service developed the urban influence
with short-term complications, diabetes with long-term codes (UIC) distinguishing metropolitan counties by pop-
complications, chronic obstructive pulmonary disease or ulation size (codes 1 and 2) and nonmetropolitan coun-
asthma in older adults, hypertension, heart failure, de- ties by population size of their town or city and proximity
hydration, bacterial pneumonia, urinary tract infection, to macro and micropolitan areas (codes ranging from 3 to
angina without procedure, uncontrolled diabetes, lower- 12).21 Following a previous study,22 we categorized ru-
extremity amputation among patients with diabetes, rality into 3 categories to represent the various degrees
and perforated appendix. After identifying the pre- of rurality based on the counties where patients resided:
ventable conditions, we coded and analyzed the 30-day (1) micropolitan areas defined as a micropolitan area ad-
preventable readmission as a dichotomous variable. jacent or not adjacent to a large or small metropolitan
The severity of illness and the risk of mortality for area (UIC codes 3, 5, and 8); (2) adjacent rural areas de-
30-day preventable readmissions were extracted from fined as noncore counties adjacent to a large or small
MedPAR by applying 3M All Patient Refined Diagnosis metropolitan area (UIC codes 4, 6, and 7); and (3) remote
Related Groups (APR-DRGs) software. The 3M APR- rural areas defined as noncore counties not adjacent to
DRGs defined the severity of illness as “the extent of metropolitan/micropolitan areas or noncore counties ad-
physiological decompensation or organ system loss of jacent to a micropolitan area that does not contain a town
function” and the risk of mortality as “the likelihood of of at least 2,500 or more residents (UIC codes 9, 10, 11,
dying.”20 The severity of illness and risk of mortality were and 12), with urban counties defined as a large or small
categorized into 4 subclasses—extreme, major, moderate, metropolitan areas (UIC codes 1 and 2) in the reference
and minor levels—based on patients’ clinical condi- group.
tions that take several factors into account, including
patients’ age, principle diagnosis, secondary diagnoses,
Control Variables
the presence of procedures, and the interaction of these
factors.20 Given the definition and the categories for the In addition to rurality, we controlled for several covari-
severity of illness and risk of mortality, a disease may ates at the patient and community level in the analytical
have extreme/major level of illness severity but have models.17,23-25 At the patient level, patient demographics
moderate/minor level of risk of mortality. The example included age, race/ethnicity, and gender. Beneficiaries’
provided by Averill et al20 is that patients with acute social conditions included 2 variables: whether patients
cholecystitis but without any comorbid conditions are were dual-eligible and whether patients lived alone.
categorized into the extreme level of severity of illness Patients’ clinical conditions included: (1) beneficiaries’

178 The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association
Chen et al. Rurality and Postacute Care for Home Health

severity of illness or risk of mortality during the index ad- heart failure Medicare beneficiaries who received home
mission. The methods used to create an indicator variable health services for postacute care were Caucasian. More
for patients’ severity of illness and risk of mortality for than 90% of our study sample required assistance in
preventable 30-day readmissions, discussed previously, medication management or ADL. Finally, approximately
were applied to create the indicator variables for patients’ 93% of the study sample had shortness of breath.
severity of illness and risk of mortality during the in- For the second study sample, 79% were readmitted to
dex admission; (2) whether beneficiaries felt depressed the same hospital. Approximately 64% of the preventable
or hopeless; (3) the frequency of anxiety; (4) patients’ readmissions were at the extreme/major level of illness
cognitive functions; (5) conditions when beneficiaries felt severity, while 55% of the index admission presented at
dyspnea or shortness of breath; (6) beneficiaries’ activities the extreme/major level of illness severity. Similar con-
of daily living (ADL) functioning; (7) whether beneficia- ditions were observed for the risk of mortality as 54% of
ries had pressure or stasis ulcer; (8) whether beneficiaries the index admissions were at the extreme/major level of
required assistance in medication management; and (9) risk of mortality, while 58% of readmissions were at the
the intensity of home health visits per week. extreme/major level of risk of mortality.
At the community level, we included the number of Table 3 presents the OR and 95% CI from the mul-
PCPs per 1,000 population, the number of acute care hos- tivariable logistic regression for a 30-day preventable
pital beds per 1,000 population, and per capita income at readmission. Regardless of including the extreme/major
the county level. The operational definition of variables level of illness severity or the extreme/major level of risk
and the data sources are presented in Table 1. of mortality in the regression, the likelihood of 30-day
preventable readmission for patients in remote rural
areas was 27% lower than that in urban areas (OR:
Statistical Analysis 0.73; 95CI: 0.57-0.94). However, we did not observe a
statistically significant difference among adjacent rural
We applied multivariable logistic regression to examine
areas, micropolitan areas, and urban areas. Among all
the impact of rurality on each of 3 outcome variables: (1)
covariates, several were significantly associated with 30-
30-day preventable readmissions, (2) the extreme/major
day preventable readmission (P< .05), including African
level of illness severity for 30-day preventable readmis-
American race; the extreme/major level of severity and
sions, and (3) and the extreme/major level of risk of
the extreme/major level of mortality during the index
mortality for 30-day preventable readmissions. For the
admission; patients who experienced anxiety daily or all
model estimating 30-day preventable readmission, we in-
of the time; patients who felt shortness of breath when
cluded the extreme/major level of illness severity during
they walked less than 20 feet, with minor effort, or at
the index admission. In a sensitivity analysis, we replaced
rest; patients who reported depression or hopelessness;
the extreme/major level of illness severity with the ex-
and patients who had a pressure or stasis ulcer. Few
treme/major level of risk of mortality during the index
variables were negatively associated with 30-day pre-
admission. Estimated odds ratio (OR) and 95% confi-
ventable readmission: gender (being a female), requiring
dence intervals (CI) are reported.
assistance in 1 to 3 ADLs (compared to those who did
not have any ADL assistance), requiring assistance in
medication management (compared to those who did
Results
not have any assistance in medication management),
We identified 51,667 postacute home health care and increases in intensity of home health visits per
episodes as our first study sample, and 4,862 posta- week.
cute home health care episodes that had 30-day pre- Table 4 presents the results for the impact of rural-
ventable readmission were our second study sample. ity on the likelihood of being at the extreme/major
Table 2 presents descriptive statistics for the study vari- level of illness severity or the extreme/major level of
ables for the 2 study samples. For our first study sample, risk of mortality at readmission due to a preventable
approximately 83% was from urban areas, with about condition. We did not observe significant difference
2%, 4%, and 11% of heart failure home health benefi- in the extreme/major level of illness severity and the
ciaries from remote rural areas, adjacent rural areas, and extreme/major level of risk of mortality for a preventable
micropolitan areas, respectively. Approximately 51% and readmission across various degrees of rurality, with one
49% of the index admissions were at the extreme/major exception indicating that beneficiaries living in adjacent
level of illness severity and extreme/major level of risk of rural areas were 33% less likely to have extreme/major
mortality, respectively. Approximately 43% of the study level of illness severity than those in urban areas
sample was aged 85 and older. The majority (83%) of (OR: 0.67; 95% CI: 0.46-0.98). It is expected that the

The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association 179
Rurality and Postacute Care for Home Health Chen et al.

Table 1 Operational Definition of Study Variables

Variables Operational Definition Data Source(s)

Outcome Variables
30-day preventable readmission 1 if patients readmitted to hospital due to a preventable condition within 30 days of MedPAR
hospital discharge; 0 otherwise
The extreme/major level of 1 if severity of illness is at the extreme or major level at the preventable readmission; 0 if
illness severity severity of illness is at the moderate or minor level at the preventable readmission
The extreme/major level of risk 1 if risk of mortality is at the extreme or major level at the preventable readmission; 0 if
of mortality risk of mortality is at the moderate or minor level at the preventable readmission
Key Independent Variables
Remote rural areas 1 if the county’s UIC is from 9 to 12; 0 otherwise AHRF
Adjacent rural areas 1 if the county’s UIC is 4, 6, and 7; 0 otherwise
Micropolitan areas 1 if the county’s UIC is 3, 5, and 8; 0 otherwise
Urban areas 1 if the county’s UIC is 1 and 2; 0 otherwise (reference group)
Covariates
The extreme/major level of 1 if severity of illness is at the extreme or major level at the index admission; 0 if severity MedPAR
illness severity of illness is at the moderate or minor level at the index admission
The extreme/major level of risk 1 if risk of mortality is at the extreme or major level at the index admission; 0 if risk of
of mortality mortality is at the moderate or minor level at the index admission
Age Two indicator variables identifying patients 75-84 years and 85+ years, with patients
65-74 years as the reference group
Race/ethnicity Two indicator variables identifying patients as African American, and Hispanic and other
races, with white as the reference group
Female 1 if gender is female, with male in the reference group.
Live alone 1 if patients live alone; 0 otherwise. OASIS
Dual eligible 1 if patients are Medicare/Medicaid dual eligible; 0 otherwise
Depression and/or hopeless 1 if patients feel depressed and/or hopeless; 0 otherwise
Anxiety Two indicator variables identifying patients with anxiety less often than daily, and with
anxiety daily or all of the time. Patients without anxiety all of the time are in the
reference group
Cognitive function Two indicator variables identifying patients with: (1) requiring assistance under
stress/unfamiliar conditions or in specific conditions and (2) requiring substantial help
in routine situations or total dependence on others. Patients being alert or oriented
were in the reference group.
Dyspnea Four indicator variables identifying patients with dyspnea at different levels of activities.
These activities are: (1) when walking more than 20 feet or climbing stairs; (2) when
walking less than 20 feet or dressing; (3) when under minor efforts, such as eating or
talking; and (4) when at rest during day or night. Patients who never felt dyspnea or
shortness of breath are in the reference group.
Pressure or stasis ulcer 1 if patients have pressure and/or stasis ulcer; 0 otherwise.
Requirement of assistance in 1 if patients required assistance in managing their oral, inhalant, or injectable
medication management medications; 0 otherwise
ADL ADL functions include grooming, toileting, bathing, dressing upper body, dressing lower
body, transferring, ambulating, and eating.
Three dummy variables represent patients with different levels of ADL functions: (1)
patients required assistance in 1 to 3 ADL functions; (2) patients required assistance in
4 to 6 ADL functions; and (3) patients required assistance in at least 7 ADL functions or
were completely dependent on someone to perform ADL functions. The reference
group is the patients who could independently perform all ADL functions.
Intensity of home health visits (Number of visits divided by the length of home health care) multiplied by 7. The number HHA RIF
per week of visits include the visits from 6 home health care professionals: nurses, physical
therapists, occupational therapists, speech therapists, home health aides, and
medical social workers.
Primary care resources Number of PCPs/1,000 population AHRF
Acute care hospital bed Number of acute care hospital beds/1,000 population
Income Per capita income

MedPAR, Medicare Provider Analysis Review; AHRF, Area Health Resources File; OASIS, Outcome Assessment Information Set; HHA RIF, Home Health
Agency Research Identifiable Research File.

180 The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association
Chen et al. Rurality and Postacute Care for Home Health

Table 2 Descriptive Analysis for Study Variables

Percentage or Meana Percentage or Meanb


Variables (SD; N = 51,667) (SD; N = 4,862)

Dependent Variables
Preventable readmission 9.41 (29.20) NA
The extreme/major level of illness severity at the preventable readmission NA 63.70 (48.09)
The extreme/major level of risk of mortality at the preventable readmission NA 57.90 (49.38)
Key Independent Variables
Remote rural areas 2.26 (14.87) 2.08 (14.29)
Adjacent rural areas 4.02 (19.63) 4.02 (19.66)
Micropolitan areas 10.78 (31.01) 10.86 (31.11)
Covariates
Patients’ Clinical and Functional Conditions
The extreme/major level of illness severity at the index readmission 51.48 (49.98) 55.35 (49.72)
The extreme/major level of risk of mortality at the index readmission 48.88 (49.99) 53.60 (49.88)
Patients aged 65 and 74 20.02 (40.74) 22.83 (41.98)
Patients aged 75 and 84 37.49 (48.41) 37.39 (48.39)
African American 12.36 (32.91) 14.38 (35.09)
Other race 4.49 (20.72) 4.50 (20.73)
Female 58.26 (49.31) 55.02 (49.75)
Live alone 30.05 (45.85) 28.30 (45.05)
Dual eligible for the Medicare and Medicaid 3.97 (19.52) 4.20 (20.05)
Depression or hopeless 17.77 (38.22) 20.51 (40.38)
Levels of anxiety:
Not anxious all of the time Reference
Less often than daily 25.81 (43.76) 25.89 (43.81)
Daily or all of the time 17.17 (37.71) 20.05 (40.04)
Level of cognitive functions:
Being alert or oriented and able to focus and shift attention independently Reference
Required assistance under stress/unfamiliar conditions or in specific conditions 38.41 (48.64) 40.43 (49.08)
Required substantial help in routine situations or totally dependent 2.21 (14.70) 2.49 (15.58)
Level of dyspnea or shortness of breath:
Never felt dyspnea or shortness of breath Reference
When walking more than 20 feet or climbing stairs 26.31 (44.03) 20.71 (40.53)
When walking less than 20 feet or dressing 38.43 (48.64) 36.46 (48.14)
When under minor efforts, such as eating or talking 22.98 (42.07) 28.94 (45.35)
When at rest during day or night 5.84 (23.46) 9.40 (29.18)
Pressure or stasis ulcer 7.36 (26.12) 9.26 (28.99)
Required assistance in medication management 94.66 (22.48) 92.34 (26.60)
Level of ADL assistance:
Without any assistance in ADL Reference
Required assistance in 1 to 3 ADL 28.78 (45.27) 24.68 (43.12)
Required assistance in 4 to 6 ADL 28.41 (45.10) 28.75 (45.27)
Required assistance in at least 7 ADL or complete dependence on ADL 34.31 (47.47) 37.80 (48.49)
Number of home health visits per week (log-transformed) 2.83 (1.94) 2.19 (1.77)
Community Factors
Number of PCPs per 1,000 population 0.75 (0.30) 0.75 (0.30)
Number of acute hospital beds per 1,000 population 3.10 (2.23) 3.14 (2.18)
Income per capita ($1,000) 40.07 (11.65) 39.72 (11.58)

a
The percentage or mean was based on the study sample from heart failure Medicare beneficiaries who received home health services for postacute
care within 14 days of hospital discharge.
b
The percentage or mean was based on the study sample who both appeared in the index admission and experienced 30-day preventable readmission.

extreme/major level of illness severity and of risk of mor- 2.26; 95% CI: 1.95-2.63 and OR: 2.51; 95% CI: 2.17-2.90
tality at the index admission is strongly associated with for the extreme/major level of illness severity and the
the extreme/major level of illness severity and of risk of extreme/major level of risk of mortality, respectively).
mortality at readmission for a preventable condition (OR: Additionally, beneficiaries who felt depressed or hopeless

The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association 181
Rurality and Postacute Care for Home Health Chen et al.

Table 3 Multivariate Regression Models for the Likelihood of 30-Day Preventable Readmission

Variables Odds Ratio (95% CI) Odds Ratio (95% CI)

Remote rural areas 0.73∗ (0.57-0.94) 0.73∗ (0.56-0.94)


Adjacent rural areas 0.87 (0.72-1.06) 0.87 (0.72-1.05)
Micropolitan areas 0.98 (0.87-1.10) 0.98 (0.87-1.10)
Covariates
Patients’ Clinical and Functional Conditions
The extreme/major level of illness severity at the index admission 1.19∗∗∗ (1.10-1.28) NA
The extreme/major level of risk of mortality at the index admission NA 1.21∗∗∗ (1.13-1.31)
Patients aged 65 and 74 0.92 (0.83-1.02) 0.94 (0.85-1.03)
Patients aged 75 and 84 0.95 (0.87-1.03) 0.95 (0.88-1.04)
African American 1.26∗∗∗ (1.13-1.41) 1.26∗∗∗ (0.12-1.41)
Other race 1.04 (0.85-1.27) 1.04 (0.85-1.26)
Female 0.90∗∗ (0.84-0.97) 0.91∗∗ (0.84-0.98)
Live alone 1.00 (0.92-1.09) 1.01 (0.93-1.09)
Dual eligible for the Medicare and Medicaid 0.97 (0.79-1.19) 0.97 (0.79-1.19)
Depression or hopeless 1.12∗ (1.02-1.23) 1.12∗ (1.02-1.23)
Levels of anxiety:
Not anxious all of the time Reference
Less often than daily 1.03 (0.95-1.13) 1.03 (0.94-1.13)
Daily or All of the time 1.15∗∗ (1.04-1.27) 1.15∗∗ (1.04-1.28)
Level of cognitive functions:
Being alert or oriented and able to focus and shift attention independently Reference
Required assistance under stress/unfamiliar conditions or in specific conditions 0.97 (0.90-1.05) 0.97 (0.90-1.05)
Required substantial help in routine situations or totally dependent 0.99 (0.77-1.26) 0.98 (0.77-1.25)
Level of dyspnea or shortness of breath:
Never felt dyspnea or shortness of breath Reference
When walking more than 20 feet or climbing stairs 1.14 (0.94-1.38) 1.14 (0.94-1.38)
When walking less than 20 feet or dressing 1.40∗∗∗ (1.17-1.68) 1.40∗∗∗ (1.16-1.68)
When under minor efforts, such as eating or talking 1.83∗∗∗ (1.51-2.20) 1.82∗∗∗ (1.51-2.20)
When at rest during day or night 2.30∗∗∗ (1.86-2.85) 2.30∗∗∗ (1.86-2.84)
Pressure or stasis ulcer 1.38∗∗∗ (1.21-1.57) 1.37∗∗∗ (1.20-1.55)
Required assistance in medication management 0.76∗∗∗ (0.66-0.88) 0.76∗∗∗ (0.66-0.88)
Level of ADL assistance:
Without any assistance in ADL Reference
Required assistance in 1 to 3 ADL 0.70∗∗∗ (0.61-0.80) 0.70∗∗∗ (0.61-0.80)
Required assistance in 4 to 6 ADL 0.95 (0.83-1.09) 0.95 (0.84-1.09)
Required assistance in at least 7 ADL or complete dependence in ADL 1.03 (0.91-1.17) 1.03 (0.91-1.17)
Intensity of home health visits per week (log-transformed) 0.49∗∗∗ (0.47-0.52) 0.49∗∗∗ (0.47-0.52)
Community Factors
Number of PCPs per 1,000 population 0.92 (0.78-1.09) 0.92 (0.78-1.09)
Number of acute hospital beds per 1,000 population 1.00 (0.98-1.02) 1.00 (0.98-1.02)
Income per capita ($1,000) 1.00 (0.99-1.00) 1.00 (0.99-1.00)


P< .05; ∗∗ P< .01; ∗∗∗ P< .001.

(OR: 1.24; 95% CI: 1.02-1.51) or felt shortness of breath pendence in ADL) were likely to be at the extreme/major
with minor effort or at rest (OR: 1.52; 95% CI: 1.04-2.21 level of risk of mortality for a preventable readmission.
and OR: 1.82; 95% CI: 1.17-2.81 for minor effort and
at rest, respectively) or who had pressure/stasis ulcer
(OR: 1.63; 95% CI: 1.23-2.15) were likely to be at the
Discussion
extreme/major level of illness severity for a preventable Our study found that heart failure Medicare home health
readmission. Finally, we also observed that beneficiaries beneficiaries living in remote rural areas were less likely
who had pressure/stasis ulcer (OR: 1.60; 95% CI: 1.23- to have 30-day preventable readmission than those in
2.09) or required 4 or more ADL assistance (OR: 1.38; urban areas. Our finding is contrary to what one may
95% CI: 1.06-1.78 for 4-6 ADL assistance; OR: 1.38; 95% expect in rural populations, who, in general, have poorer
CI: 1.08-1.77 for at least 7 ADL assistance or complete de- health outcomes and less access to care. Nevertheless, our

182 The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association
Chen et al. Rurality and Postacute Care for Home Health

Table 4 Multivariate Regression Models for the Likelihood of Having Severity of Illness or Risk of Mortality at the Extreme/Major Level for a 30-Day
Preventable Readmission

Severity of Illness Risk of Mortality


at the Extreme at the Extreme
or Major Level or Major Level
Variables Odds Ratio (95% CI) Odds Ratio (95% CI)

Remote rural areas 0.64 (0.39-1.06) 0.75 (0.46-1.23)


Adjacent rural areas 0.67∗ (0.46-0.98) 0.96 (0.66-1.40)
Micropolitan areas 0.84 (0.66-1.07) 0.81 (0.64-1.02)
Covariates
Patients’ Clinical and Functional Conditions
The extreme/major level of illness severity at the index admission 2.26∗∗∗ (1.95-2.63) NA
The extreme/major level of risk of mortality at the index admission NA 2.51∗∗∗ (2.17-2.90)
Patients aged 65 and 74 1.16 (0.95-1.43) 1.01 (0.83-1.24)
Patients aged 75 and 84 1.08 (0.91-1.28) 1.11 (0.94-1.31)
African American 0.95 (0.76-1.20) 0.88 (0.70-1.10)
Other race 1.24 (0.82-1.88) 0.87 (0.59-1.29)
Female 0.88 (0.75-1.02) 0.82∗∗ (0.71-0.95)
Live alone 0.93 (0.79-1.10) 0.88 (0.75-1.04)
Dual eligible for the Medicare and Medicaid 1.40 (0.90-2.16) 0.94 (0.63-1.42)
Depression or hopeless 1.24∗ (1.02-1.51) 1.17 (0.97-1.41)
Levels of anxiety:
Not anxious all of the time Reference
Less often than daily 1.04 (0.87-1.25) 0.99 (0.83-1.18)
Daily or all of the time 0.71∗∗∗ (0.58-0.87) 0.76∗∗ (0.62-0.93)
Level of cognitive functions:
Being alert or oriented and able to focus and shift attention independently Reference
Required assistance under stress/unfamiliar conditions or in specific conditions 0.99 (0.85-1.16) 1.04 (0.89-1.22)
Required substantial help in routine situations or totally dependent 1.26 (0.76-2.10) 0.95 (0.59-1.54)
Level of dyspnea or shortness of breath:
Never felt dyspnea or shortness of breath Reference
When walking more than 20 feet or climbing stairs 1.10 (0.75-1.61) 0.93 (0.64-1.36)
When walking less than 20 feet or dressing 1.28 (0.89-1.85) 1.11 (0.77-1.60)
When under minor efforts such as eating or talking 1.52∗ (1.04-2.21) 1.14 (0.78-1.66)
When at rest during day or night 1.82∗∗ (1.17-2.81) 1.35 (0.88-2.06)
Pressure or stasis ulcer 1.63∗∗∗ (1.23-2.15) 1.60∗∗∗ (1.23-2.09)
Required assistance in medication management 0.84 (0.63-1.13) 0.82 (0.62-1.09)
Level of ADL assistance
Required assistance in 1 to 3 ADL 1.05 (0.80-1.38) 1.09 (0.83-1.42)
Required assistance in 4 to 6 ADL 1.32∗ (1.01-1.71) 1.38∗ (1.06-1.78)
Required assistance in at least 7 ADL or complete dependence in ADL 1.18 (0.91-1.52) 1.38∗ (1.08-1.77)
Intensity of home health visits 0.76∗∗∗ (0.69-0.83) 0.76∗∗∗ (0.70-0.84)
Community Factors
Number of PCPs per 1,000 population 1.25 (0.89-1.76) 1.32 (0.95-1.83)
Number of acute hospital beds per 1,000 population 0.97 (0.94-1.01) 0.97 (0.94-1.01)
Income per capita ($1,000) 1.00 (0.99-1.00) 1.00 (0.99-1.00)


P< .05; ∗∗ P< .01; ∗∗∗ P< .001.

findings are consistent with a previous study based on di- commonly found that rural populations were less likely
abetic Medicare beneficiaries, indicating that diabetic el- to utilize health care than urban elderly.29 Our study
derly in remote rural areas were less likely to have 30-day population was homebound and had, on average, 4.2
readmission than those in urban areas.22 Increases in diagnosed conditions.30 About 80% of our study sample
travel distance to access care increase monetary and non- was aged 75 and older and 93% had shortness of breath.
monetary cost for patients and their families.26 Rural res- Given their seniority, homebound status, and clinical
idents, on average, travel longer and are poorer and more conditions, they were more likely to rely on their family
sensitive to traveling cost than urban residents.27,28 It is or community to arrange transportation to receive care.

The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association 183
Rurality and Postacute Care for Home Health Chen et al.

For home health beneficiaries in remote rural areas, hav- daily or all of the time, depression/hopelessness, pressure
ing to travel a long distance makes accessing facility-based or stasis ulcer, and dyspnea with specified conditions.
health care even more challenging. Thus, our findings Home health professionals can use the risk factors identi-
of low 30-day preventable readmission for heart failure fied in this study to identify high-risk patients who may
Medicare home health beneficiaries in remote rural areas benefit from early intervention in an effort to further re-
may reflect unmet postacute care needs that are currently duce 30-day preventable readmission.
not well understood, rather than reflect good quality of Along with the risk factors identified above, we found
postacute care during the period of home health care. that patients who required assistance in medication
Among those readmitted to the hospital for a pre- management or assistance with 1 to 3 ADLs are less
ventable condition, we found that beneficiaries in likely to experience 30-day preventable readmission,
adjacent rural areas had a lower likelihood of having compared to those who did not require assistance in
extreme/major level of illness severity for a preventable medication management or ADL, respectively. The
readmission, compared to beneficiaries in urban areas. conditions for preventable readmission in our study,
Although evidence related to health or health behaviors such as short-/long-term complications of diabetes,
for the elderly in adjacent rural areas and micropolitan require patients rigorously follow medication regi-
areas is lacking, studies showed that more than 50% mens. Home health beneficiaries, on average, have
of the rural population has the same physician for 8 medications.33 Medication errors within this popula-
more than 5 years, and rural physicians are satisfied tion are of immense concern.34 As discussed previously,
with their relationship with their patients and clinical the majority of our study sample was aged 75 years and
autonomy.31,32 Given our definition for remote rural, older and had multiple clinical conditions. Their health
adjacent rural, and micropolitan areas, the resources in literacy and capabilities to follow the rigorous medication
adjacent rural and micropolitan areas are better than regimens were critical to reduce preventable readmis-
resources in remote rural areas. It is likely that heart sion. Home health professionals should be encouraged
failure home health Medicare beneficiaries in adjacent to assess beneficiaries’ health literacy and medical man-
rural areas do not have the disadvantage of long driving agement capabilities. Such assessments may include
distances which affects those in remote rural areas. This repeatedly educating and monitoring home health
population has the benefit of continuity of care from elderly who independently manage their medication or
the same providers. The longer-term patient-provider do not receive the ADL assistance through the phone or
relationship helps physicians who understand patients telehealth.
well and can detect changes in patients’ clinical condition Finally, patients’ conditions, such as depression or
at the early stage. However, further investigation is re- hopelessness, dyspnea, ADL assistance, and pressure or
quired to understand the mechanisms underlying the low stasis ulcer are positively associated with the likelihood
likelihood of having an extreme/major level of severity of being at the extreme/major level of illness severity
of illness for a preventable readmission in adjacent rural or of risk of mortality for a 30-day preventable read-
areas. mission. Although these clinical conditions are assessed
Furthermore, our findings regarding risk factors for at the beginning of home health care, these conditions
preventable readmissions have critical implications for are also likely to appear during the index admission.
clinical practice. In addition to rurality, our model in- Hospitals may need to reassess patients’ clinical condi-
cluded many risk factors for readmission which are rel- tions or ensure that patients are able to receive con-
evant during the initial hospitalization and as the pa- tinuous care in the community before patients are sent
tient receives home health care. We found that patients home. Home health professionals also need to coordi-
with the extreme/major level of illness severity or of risk nate with physicians when home health professionals as-
of mortality during the index admission are at elevated sess patients with those conditions in the beginning of
risk of experiencing 30-day preventable readmission. Fac- home health care in efforts to strengthen opportunities
tors related to these 2 variables should be identified and for home health elderly affected by heart failure to safely
communicated between hospital discharge coordinators stay in the community.
and home health professionals during the care transition In addition to clinical implications, our findings
process to facilitate home health professionals’ prioritiza- hold critical implications for the home health financial
tion of patient care. Furthermore, patient demographics incentive program proposed by MedPAC. First, like other
and several clinical conditions assessed in the beginning studies which found that place matters,35,36 our findings
of a home health care episode are significantly associ- indicated that where Medicare home health beneficiaries
ated with 30-day preventable readmission. These risk fac- live affects the likelihood of preventable readmission and
tors include African American race, experiencing anxiety the level of illness severity for a preventable readmission.

184 The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association
Chen et al. Rurality and Postacute Care for Home Health

Although our study found that increases in intensity ulations. In remote rural areas, the low preventable read-
of home health visits reduce preventable readmission, mission for home health elderly in our study was likely a
and the likelihood of presenting with extreme/major result of unmet need of postacute care. Although reduc-
level of illness severity or extreme/major level of risk of ing readmissions is the nation’s top priority, it is imper-
mortality upon preventable readmission, home health ative to identify the unmet postacute care needs among
professionals cannot completely overcome the external Medicare home health beneficiaries in remote rural
challenges that already exist in the communities where areas. Additionally, the factors that affect the likelihood
patients live. Second, we found that patients with the ex- of preventable readmission for home health beneficiaries
treme/major level of illness severity or of risk of mortality are complex. The rural homebound elderly and their
during the index admission are more likely to have a families may have different preferences for care from the
30-day preventable readmission than their counterparts. urban homebound elderly when they need hospital care
These beneficiaries are likely to be unwelcomed by for deteriorating clinical conditions or newly developed
HHAs, given that 88% of HHAs are for-profit.1 Although complications. Applying both qualitative and quantitative
the purpose of the home health financial incentive pro- research designs to explore the mechanisms for meeting
gram is to reduce the readmission rate while improving the needs of home health elderly living in different levels
quality of home health care, access to home health for of rurality would help policy makers and stakeholders ad-
postacute care for these high-risk patients is likely to be a dress the needs of these vulnerable elderly. Furthermore,
great concern under the home health financial incentive the impact of rurality on other quality indicators, such
program. Efforts should be made to mitigate the potential as 30-day mortality rate, emergency visits after hospital
for unintended consequences resulting from the home discharge, or patient safety for Medicare home health
health financial incentive program targeting reducing beneficiaries is largely unknown. Future studies focusing
preventable readmission. Toward these efforts, the model on other quality indicators are recommended. Moreover,
that will be used to calculate the penalty rate for HHAs Medicare home health beneficiaries directly referred by
under the financial incentive program should include the the physicians in the hospital may be different from those
risk factors during the index admission, the risk factors referred by their PCPs a few days after hospital discharge.
appearing in OASIS, as well as the characteristics of Investigating the differences in beneficiaries’ functional
communities where patients live. and clinical conditions between these 2 referral sources
There are limitations in our study. First, our study is recommended. Finally, several programs, such as the
was based on administrative data and did not include HRRP and increases in Medicare Part D prescription
other clinical conditions, such as vital signs and labora- drug coverage, have been implemented under the
tory results. Second, patients’ socioeconomic status, such ACA. The impact of the ACA on quality of postacute
as education and health literacy, and coordination be- care for Medicare home health beneficiaries warrants
tween hospital, physicians, and home health profession- examination.
als are likely to affect 30-day preventable readmission
and the level of illness severity and of risk of mortal- References
ity for a preventable readmission for our study sample.
These variables were not available and are beyond the 1. Medicare Payment Advisory Commission. Report to the
scope of our data sets. Third, our study was based on Congress: Medicare Payment Policy: Chapter 9: Home
Health Services. Washington, DC: MedPAC; 2014.
a cross-sectional study design, which did not allow us
2. Rosenthal TC, Fox C. Access to health care for the rural
to observe the change in outcome measures from time
elderly. JAMA. 2000;284(16):2034-2036.
to time and to control unobserved factors, such as so-
3. Laditka JN, Laditka SB, Probst JC. Health care access in
cial context and culture in the community, which were
rural areas: evidence that hospitalization for ambulatory
likely to affect the outcome measures in our study. Fi-
care-sensitive conditions in the United States may
nally, one should carefully consider the population of in-
increase with the level of rurality. Health Place. 2009;
terest before generalizing our findings to other popula- 15(3):731-740.
tions because health plans may apply different mecha- 4. Mobley LR, Root E, Anselin L, Lozano-Gracia N,
nisms to provide postacute care for Medicare beneficia- Koschinsky J. Spatial analysis of elderly access to primary
ries enrolled in the Medicare Advantage program or other care services. Int J Health Geogr. 2006;5:19.
programs. doi:10.1186/1476-072X-5-1.
Despite the limitations, our findings provide perspec- 5. Goodman DC, Fisher E, Stukel TA, Chang C. The distance
tives for future research. Given the multiple clinical con- to community medical care and the likelihood of
ditions and physical limitations, Medicare home health hospitalization: is closer always better?Am J Public Health.
beneficiaries are one of Medicare’s most vulnerable pop- 1997;87(7):1144-1150.

The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association 185
Rurality and Postacute Care for Home Health Chen et al.

6. Gregory PM, Malka ES, Kostis JB, Wilson AC, Arora JK, 20. Averill RF, Goldfield N, Steinbeck B, et al. Development
Rhoads GG. Impact of geographic proximity to cardiac of the All Patient Refined DRGs (APR-DRGs). Available
revascularization services on service utilization. Med Care. at:http://solutions.3m.com/3MContentRetrievalAPI/
2000;38(1):45-57. BlobServlet?lmd=1225920498000&assetId=118060
7. Reschovsky JD, Slaiti AB. Access and quality: does rural 6514429&assetType=MMM_Image&blobAttribute=
America lag behind?Health Aff. 2005;24(4):1128-1139. ImageFile. Accessed December 15, 2014.
8. Rickets TC.Workforce issues in rural areas: a focus on 21. The United States Department of Agriculture Economic
policy equity. Am J Public Health. 2005;95(1):42-48. Research Services. Urban Influence Codes. Available at:
9. Hartman L, Jarosek SL, Virnig BA, Durham S. http://www.ers.usda.gov/data-products/urban-influence-
Medicare-certified home health care: urban-rural codes/documentation.aspx. Accessed November 15,
differences in utilization. J Rural Health. 2007; 2014.
23(3):254-257. 22. Bennett KJ, Probst JC, Vyavaharkar M, Glover SH.
10. Schlenker RE, Powell MC, Goodrich GK. Rural-urban Lower rehospitalization rates among rural Medicare
home health care differences before the Balanced Budget beneficiaries with diabetes. J Rural Health.
Act of 1997. J Rural Health. 2002;18(2):359-372. 2012;28(3):227-234.
11. Etzioni DA, Fowl RJ, Wasif N, Donohue JH, Cima RR. 23. Fortinsky RH, Madigan EA, Sheehan TJ,
Distance bias and surgical outcomes. Med Care. 2013; Tullai-McGuiness S, Kleppinger A. Risk factors for
51(3):238-244. hospitalization in a national sample of Medicare home
12. Hogan C. Patterns of travel for rural individuals health care patients. J Appl Gerontol. 2014;33(4):
hospitalized in New York State: relationships between 474-493.
distance, destination, and case mix. J Rural Health. 1988; 24. Rosati RJ, Huang L, Navaie-Waliser M, Feldman PH. Risk
4(2):29-41. factors for repeated hospitalizations among home
13. Nallamothu B, Fox KA, Kennelly BM, et al. Relationship healthcare recipients. J Healthc Qual. 2003;25(2):
of treatment delays and mortality in patients undergoing 4-10.
fibrinolysis and primary percutaneous coronary 25. Herrin J, St Andre J, Kenward K, Joshi MS, Audet AM,
intervention. The Global Registry of Acute Coronary Events. Hines SC. Community factors and hospital readmission
Heart. 2007;93(12):1552-1555. rates. Health Serv Res. 2015;50(1):20-39.
14. Jencks SF, Williams MV, Coleman EA. Rehospitalizations 26. Cauley SD.The time price for medical care. Rev Econ Stat.
among patients in the Medicare fee-for-service program. 1987;69:59-66.
N Engl J Med. 2009;360(14):1418-1428. 27. Erlyana E, Damrongplasit KK, Melnick G. Expanding
15. Centers for Medicare and Medicaid Services: Readmission health insurance to increase health care utilization: will it
Reduction Program. https://www.cms.gov/Medicare/ have different effects in rural vs. urban areas?Health
Medicare-Fee-for-Service-Payment/AcuteInpatient Policy. 2011;100(2-3):273-281.
PPS/Readmissions-Reduction-Program.html. Accessed 28. Nelson RE, Hicken B, West A, Rupper R. The effect
August 5, 2015. of increased travel reimbursement rates on health
16. Yale New Haven Health Services Corporation/Center for care utilization in the VA. J Rural Health. 2012;28(2):192-
Outcomes Research and Evaluation. Fiscal Year 2013 201.
Hospital Readmissions Reduction Program: Measure 29. Weeks WB, Bott DM, Lamkin RP, Wright SM. Veterans
Methodology Report. Available at: http://www. health administration and Medicare outpatient health
henlearner.org/wp-content/uploads/2012/03/Hospital- care utilization by older rural and urban New England
Excess-Readmissions-Metric_Methodology-Report.pdf. veterans. J Rural Health. 2005;21(2):167-171.
Accessed August 28, 2014. 30. Caffrey C, Sengupta M, Moss A, Harris-Kojetin L,
17. Madigan EA, Gordon NH, Fortinsky RH, Koroukian SM, Valverde R. Home health care and discharged hospice
Piña I, Riggs JS. Rehospitalization in a national care patients: United States. 2000and 2007. Natl Health
population of home health care patients with heart Stat Report. 2011;27(38):1-27.
failure. Health Serv Res. 2012;47(6):2316-2338. 31. Pathmam DE, Williams ES, Konrad TR. Rural physician
18. Wolff JL, Meadow A, Weiss CO, Boyd CM, Leff B. satisfaction: its sources and relationship to retention.
Medicare home health patients’ transitions through acute J Rural Health. 1996;12(5):366-377.
and post-acute care settings. Med Care. 2008;46(11): 32. Donahue KE, Ashkin E, Pathmam DE. Length of
1188-1193. patient-physician relationship and patients’ satisfaction
19. AHRQ Quality Indicators— Quality Indicators Software and preventive service use in the rural south: a
Instructions, SAS Version 4.5. Available at: http://www. cross-sectional telephone study. BMC Fam Pract. 2005;
qualityindicators.ahrq.gov/Downloads/Software/SAS/ 6:40. doi:10.1186/1471-2296-6-40.
V45/Software_Instructions_SAS_V4.5.pdf. Accessed April 33. Cannon KT, Choi MM, Zuniga MA. Potentially
10, 2014. inappropriate medication use in elderly patients receiving

186 The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association
Chen et al. Rurality and Postacute Care for Home Health

home health care: a retrospective data analysis. Am J affect the treatment of Medicare beneficiaries. Health Aff
Geriatr Pharmacother. 2006;4(2):134-143. (Millwood). 2004; Suppl Variation: Var33-44.
34. Meredith S, Feldman PH, Frey D, et al. Possible 36. LaVeist T, Pollack K, Thorpe R Jr, Fesahazion R, Gaskin
medication errors in home healthcare patients. J Am D. Place, not race: disparities dissipate in southwest
Geriatr Soc. 2001;49(6):719-724. Baltimore when blacks and whites live under similar
35. Baicker K, Chandra A, Skinner JS, Wennberg JE. Who conditions. Health Aff (Millwood). 2011;30(10):1880-
you are and where you live: how race and geography 1887.

The Journal of Rural Health 32 (2016) 176–187 


c 2015 National Rural Health Association 187

You might also like