0
Healthcare Utilization Among Vulnerable Populations ICare 2011
1
Vulnerability
• Vulnerability is a multidimensional construct with many definitions
• Vulnerable populations experience a convergence of many risk
factors (community and individual level) Low income, lack of insurance, age, race, illness
Physically vulnerable
groups
Psychologically
vulnerable groups
Socially vulnerable
groups
High risk mothers and
infants Mentally ill and disabled Abusing families
Chronically ill and
disabled
Alcohol or substance
abusers Homeless persons
Persons living with
HIV/AIDS Suicide or homicide prone Immigrants and refugees
Source: Aday, L. At Risk in America : The Health and Health Care Needs of Vulnerable
Populations in the United States. 2001
2
Vulnerability
• Multiple risk factors result in a higher risk for poor health outcomes.
• Patients with poorer health outcomes place an additional burden on
healthcare systems.
• Improved access and utilization of primary care services can
decrease the burden on hospital systems, reducing medical costs
for high risk individuals.
3
Study Objective
• Describe patterns of healthcare utilization among vulnerable sub-
groups of a low income, underserved population and identify
populations that could benefit from additional intervention to reduce
healthcare costs and improve quality of care.
4
Integrated Care Collaboration (ICC)
• Non-profit alliance of health care safety-net providers in Central
Texas (Austin area) Multi-hospital systems
Public and private clinics
Federally qualified health centers
City public health clinics
• Organized around the medically indigent population of individuals
who are uninsured or underinsured
• Maintains a fully operational Health Information Exchange (HIE)
system Database called ICare
5
ICare HIE Database
• 2011 Available Data Demographics
Encounter level data
‒ Admission and discharge dates
‒ Location/Facility
‒ Payor
‒ Provider
‒ Diagnoses (ICD-9 codes)
‒ Procedures
6
Inclusion Criteria
• Timeframe: Calendar Year 2011
• ICare patients aged 0-64 years 65+ excluded due to lack of Medicare data in ICare
• Encounter types Clinic/office
Emergency department (ED)
Inpatient (IP)
Outpatient
7
Vulnerable Groups
• Near Elderly Age of 60-64 years as of 12/31/2011
• Homeless Most recent address in the ICare recorded as “Homeless” or “No
Address” OR
Most recent address was associated with an agency that serves the
homeless (Salvation Army, Austin Resource Center for the Homeless,
etc.)
Past addresses are not retained by the ICare system. ‒ Patient defined as currently homeless at the time of the data pull may not
have been homeless at the time of their encounter.
8
Vulnerable Groups
• Behavioral Health Diagnosis of a behavioral health condition at any encounter during 2011
Mental illness diagnoses are defined by the Healthcare Costs and
Utilization Project (HCUP) Clinical Classifications Software (CCS) for
ICD-9-CM1
‒ Codes related to tobacco use have been excluded as mental illness
diagnoses
• Disabled Diagnosis of a condition which may qualify them as having a severe and
chronic disability during 2011
Disability diagnoses are defined by ICD-9 codes that are described and
approved by the Texas Department of Aging and Disability Services
(DADS) as a condition which may qualify an individual as having a
severe and chronic disability as described in federal and state law2
1. http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp
2. http://www.dads.state.tx.us/providers/guidelines/ICD-9-CM_Diagnostic_Codes.pdf
9
Methods
• Descriptive statistics to describe population Distribution of vulnerable populations
Demographics (race/ethnicity, age, sex)
Healthcare utilization (ED and IP visits)
• Poisson regression to quantify risk Risk of an ED visit
Risk of an IP visit
All risk models are adjusted for all four vulnerable groups as well as
age, sex, and race/ethnicity
10
Results
• Total of 279,992 patients
aged 0-64 years
• 58,624 (20.9%) of patients
were vulnerable
• 221,368 (79.1%) did not fall
into a vulnerable population
79.1%
20.9%
Distribution of Vulnerable Populations
Not Vulnerable Vulnerable
11
Vulnerable Population by Group
14.5%
4.1%
71.0%
25.1%
0
20
40
60
80
Near Elderly Homeless Behavioral Health Disabled
Perc
en
t o
f V
uln
era
ble
Pati
en
ts
*Groups are not mutually exclusive
12
Vulnerable Group Overlap
Condition Near Elderly Homeless Behavioral
Health Disabled
Near Elderly 8,512 106 1,982 961
Homeless 106 2,401 1,145 215
Behavioral
Health 1,982 1,145 41,642 4,758
Disabled 961 215 4,758 14,689
1,145 / 2,401 = 0.477
47.7% of Homeless patients also have a Behavioral Health Diagnosis
13
Demographic Distributions
30.9 28.2
41.4
10.6 10.7 10.1
52.2 54.7
42.8
6.3 6.5 5.6
0
20
40
60
All Patients Non-Vulnerable Vulnerable
Perc
en
t o
f P
ati
en
ts
Race/Ethnicity Distribution by Vulnerable Status
White Black Hispanic Other
14
Demographic Distributions
45.2 49.6
43.3
36.3
9.0
15.4
10.2 8.8
36.9 33.0
41.6
49.1
8.8
2.1 4.9 5.8
0
20
40
60
Near Elderly Homeless Behavioral Health Disabled
Perc
en
t o
f P
ati
en
ts
Race/Ethnicity Distribution by Vulnerable Group
White Black Hispanic Other
15
Demographic Distributions
39.0
44.0
20.4
43.6 43.8 42.9
17.4
12.3
36.7
0
10
20
30
40
50
All Patients Non-Vulnerable Patients Vulnerable Patients
Perc
en
t o
f P
ati
en
ts
Age Distribution by Vulnerable Status
Age <18 Age 18-44 Age 45-64
16
Demographic Distributions
4.7
19.5
32.6
49.3 50.2
37.7
46.0
30.2 29.7
0
15
30
45
60
Homeless Behavioral Health Disabled
Perc
en
t o
f P
ati
en
ts
Age Distribution by Vulnerable Group
Age <18 Age 18-44 Age 45-64
*Near elderly has been omitted from chart, 100% fall into “Age 45-64”.
17
Demographic Distributions
44.3 43.8 46.0
55.8 56.2 54.0
0
15
30
45
60
All Patients Non-Vulnerable Patients Vulnerable Patients
Perc
en
t o
f P
ati
en
ts
Sex Distribution by Vulnerable Status
Male Female
18
Demographic Distributions
43.7
77.7
45.6 43.8
56.3
22.3
54.4 56.2
0
20
40
60
80
100
Near Elderly Homeless Behavioral Health Disabled
Perc
en
t o
f P
ati
en
ts
Sex Distribution by Vulnerable Group
Male Female
19
Demographics Summary
• Race/Ethnicity A higher proportion of vulnerable patients are White and a lower
proportion are Hispanic compared to non-vulnerable patients
Homeless patients have a higher proportion of White and Black
Disabled patients have a higher proportion of Hispanic
• Age Homeless had the lowest proportion of children <18
• Sex Overall, vulnerable patients are distributed similarly to non-vulnerable
patients
Homeless is the only group where men are more predominant than
women
20
Emergency Department Utilization
53.9% 54.0% 53.6%
1.8 1.6
2.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
20
30
40
50
60
Total Patients Not Vulnerable Vulnerable
Avera
ge N
um
ber
of
ED
Vis
its
Perc
en
t w
ith
ED
Vis
it
ED Utilization by Vulnerable Status
% With at Least One ED Visit Avg Number of ED Visits
21
Emergency Department Utilization
33.5%
82.5%
54.3% 60.4%
1.8
3.3
2.8 2.6
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0
20
40
60
80
100
Near Elderly Homeless Behavioral Health
Disabled
Avera
ge N
um
ber
of
ED
Vis
its
Perc
en
t w
ith
ED
Vis
it
ED Utilization by Vulnerable Group
% With at Least One ED Visit Avg Number of ED Visits
22
Frequent Emergency Department Users
0.4%
1.4%
9.0%
4.5%
3.3% 9.3
13.6 13.2 13.9
15.5
0.0
3.0
6.0
9.0
12.0
15.0
18.0
0.0
2.0
4.0
6.0
8.0
10.0
Not Vulnerable
Near Elderly Homeless Behavioral Health
Disabled
Avera
ge N
um
ber
of
ED
Vis
its
Perc
en
t o
f E
D U
sers
% Frequent User Avg Number of ED Visits
23
Inpatient Utilization and Length of Stay
11.7%
15.5% 15.2%
17.7% 18.1%
3.6
4.9 4.6
4.9
5.7
0
2
4
6
8
0
4
8
12
16
20
Not Vulnerable
Near Elderly Homeless Behavioral Health
Disabled
Avera
ge L
en
gth
of
Sta
y (
days)
Perc
en
t w
ith
IP
Vis
it
% With IP Visit Average LOS (days)
24
Utilization Summary
• A similar proportion of vulnerable and non-vulnerable patients had
at least one ED visit; however, vulnerable ED users had a higher
average number of ED visits.
• Homeless patients had the highest rate of ED use and the greatest
average number of ED visits among ED users. They also had the
highest rate of frequent ED utilization among ED users.
• Near elderly patients had a lower rate of ED utilization than the
overall and non-vulnerable populations.
• Disabled patients had the highest rate of IP utilization and the
longest average length of stay among IP users. Disabled frequent
users had the greatest average number of ED visits.
25
Risk of ED Utilization by Vulnerable Group
Group Risk Ratio (RR) % Increase p-Value
Near Elderly 0.67 -50% <0.001
Homeless 1.48 48% <0.001
Behavioral Health 1.00 0% 0.994
Disabled 1.12 12% <0.001
• Near Elderly patients had a 50% lower risk of having an ED visit, compared
to patients who were not near elderly.
• Homeless patients had a 48% greater risk and Disabled patients had a
12% greater risk of having an ED visit, compared to patients who were not
homeless or disabled, respectively.
• Behavioral Health patients had no change in risk compared to non-
Behavioral Health patients.
26
Behavioral Health Results
• Result that BH patients had no increased risk of an ED visit was
unexpected Further analysis necessary for explanation
• Looked at ED utilization by Behavioral Health sub-group
BH Subgroup
Adjustment disorders Mood disorders
Anxiety disorders Personality disorders
ADHD/Disruptive behavior Psychotic disorders
Dementia Alcohol-related
Developmental disorders Substance-related
Disorders usually diagnosed in
childhood
Suicide/intentional injury
Impulse control Miscellaneous mental disorders
27
BH Exploratory Analysis
BH Subgroup % w ED Visit BH Subgroup % w ED Visit
Adjustment disorders 47.9 Mood disorders 59.3
Anxiety disorders 57.7 Personality disorders 59.3
ADHD/Disruptive
behavior 43.6
Psychotic disorders 68.6
Dementia 61.8 Alcohol-related 74.2
Developmental disorders 63.6 Substance-related 72.1
Disorders usually
diagnosed in childhood 39.6
Suicide/intentional
injury 90.0
Impulse control 50.6
Miscellaneous mental
disorders 66.4
• Utilization quite different by Behavioral Health sub-group
28
BH Exploratory Analysis
• Looked at correlation between rates of ED utilization and clinic
utilization among BH sub-groups
• Groups with high ED utilization had low clinic utilization
R² = 0.7938
0
20
40
60
80
100
0 20 40 60 80 100
% w
ith
ED
Vis
it
% with Clinic Visit
Correlation Between ED and Clinic Utilization
29
BH Exploratory Analysis
• Re-defined Behavioral Health group to include only the following
sub-groups Anxiety disorders
Delirium, dementia, amnesic and other cognitive disorders
Personality disorders
Schizophrenia and other psychotic disorders
Alcohol-related disorders
Substance-related disorders
Suicide and intentional self-inflicted injury
• Re-ran risk analysis using new BH group
30
Risk of ED Utilization by Vulnerable Group
BH Group 1 BH Group 2
Group RR
%
Increase p-Value RR
%
Increase p-Value
Near Elderly 0.67 -50% <0.001 0.66 -52% <0.001
Homeless 1.48 48% <0.001 1.40 40% <0.001
Behavioral
Health 1.00 0% 0.994 1.25 25% <0.001
Disabled 1.12 12% <0.001 1.09 9% <0.001
• Risk of an ED visit remained similar for Near Elderly, Homeless, and
Disabled groups
• New group of BH patients have a 25% greater risk of having an ED visit
than patients not in group
31
Risk of Inpatient Utilization by Vulnerable Group
Group Risk Ratio (RR) % Increase p-Value
Near Elderly 1.10 10% 0.002
Homeless 1.26 26% <0.001
Behavioral Health 1.89 89% <0.001
Disabled 1.52 52% <0.001
• All vulnerable groups had an increased risk of having an inpatient visit
32
Risk of IP Utilization by Vulnerable Group
BH Group 1 BH Group 2
Group RR
%
Increase p-Value RR
%
Increase p-Value
Near Elderly 1.10 10% 0.002 1.12 12% <0.001
Homeless 1.26 26% <0.001 1.18 18% <0.001
Behavioral
Health 1.89 89% <0.001 2.21 121% <0.001
Disabled 1.52 52% <0.001 1.56 56% <0.001
• Risk of an IP visit remained similar for Near Elderly, Homeless, and
Disabled groups
• Limited group of BH patients have a 121% greater risk of having an IP visit
than patients not in group
33
Risk Summary
• Homeless and disabled patients had an increased risk of ED
utilization.
• Near elderly patients had a decreased risk of ED utilization.
• Behavioral health patients overall had neither an increased or
decreased risk of ED utilization. A limited group of more severe
behavioral health patients had an increased risk of ED utilization.
• All vulnerable groups had an increased risk of inpatient utilization.
34
Limitations
• Homeless status is only accurate as of last encounter captured in
ICare
• Results can only be generalized to patients under age 65
• Study population is limited to patients who are publically insured or
uninsured Entire study population may be considered “Vulnerable”.
Top Related