Outcomes by HCBS Participant Characteristics
Outcomes as measured by the HCBS outcome indicators vary
considerably by subpopulation and participant attributes, but the variation
takes on different forms for different measures. Table 7 presents the outcome
indicator rates by subpopulation and Table 8 by key demographic attributes,
also broken down by subpopulation.
Outcomes by Subpopulation
As shown in Table 7, for the vast majority of measures,
admission rates are substantially higher among the dual eligibles than the
Medicaid-only population. The differences in rates are dramatic, with dual
eligibles experiencing rates that are often 50 to 100 percent higher than the
Medicaid-only population. The exceptions are Short-Term Complications of
Diabetes and Infection Due to Device or Implant, where rates are higher among
Medicaid-only participants.
Not surprisingly, roughly the same pattern, with the same
exceptions, is found when the 65+ subpopulation is compared with the other
subpopulations. However, many measures also exhibit high rates among the 18-64
subpopulation. The contrast between the relatively high rates in the 65+
population and the other subpopulations is starkest for the I/DD subpopulation,
which exhibits a pattern of strikingly lower rates than all the other groups on
almost all measures.
In addition to variation by subgroup, tremendous State variation
in outcome indicator rates is evident from the minimum and maximum rates shown.
Indicators with the largest State range in percentage terms are Short-Term
Complications of Diabetes, Injurious Falls, and Asthma or Chronic Obstructive
Pulmonary Disease (COPD). Indicators with the smallest State range include
Bacterial Pneumonia, Potentially Preventable Infections Composite, and ACSC
Acute Conditions Composite. Total indicator rates for Short-Term Complications
of Diabetes, for example, range from a low of 68 to a high of 1,188, a 17-fold
difference. Rates for Bacterial Pneumonia range from 3,368 to 8,903, less than
a threefold difference.
Outcomes by HCBS Individual Participant Attributes
The HCBS population as a whole exhibits differences in
outcome indicator rates by race, gender, age, and urban/nonurban status as
measured by location in a metropolitan statistical area (MSA) or non-MSA (Table
8).
- Rates for a slight majority of outcome indicators are higher for
African Americans than for other races. The exceptions are Asthma or COPD,
Potentially Preventable Infections Composite, Bacterial Pneumonia, ACSC Acute
Conditions Composite, and Injurious Falls, where non-Hispanic whites have
higher rates. Hispanics tend to have lower rates than non-Hispanic whites for
almost all measures.
- Women tend to have worse outcomes than men, with higher rates on
all measures other than Short-Term Complications of Diabetes, Bacterial
Pneumonia, and Pressure Ulcer.
- Most of the outcome indicators exhibit unsurprising findings
related to age, with lowest rates in the 18-64 age category, higher rates in
the 65-84 category, and highest rates in the 85+ category. Two indicators,
Short-Term Complications of Diabetes and Infection Due to Device or Implant,
exhibit the reverse pattern, with rates decreasing with age. Another two indicators,
Asthma or COPD and the ACSC Chronic Conditions Composite, exhibit a peak in
rates in the 65-84 category, with lower rates among HCBS participants 85 and
older.
- Finally, rates are almost all lower in urban areas (MSAs) than in
nonurban areas. The exceptions are Pressure Ulcer, which is effectively equal
in urban and nonurban areas, and Infection Due to Device or Implant, where the
difference is trivial.
The overall rates of outcome indicators by participant
attributes vary somewhat by dual status and by subpopulation. As dual eligibles
constituted two-thirds of the HCBS population, not surprisingly, the rates for
dual eligibles look very similar to the rates overall. Among the Medicaid-only
population, however, differences by race and gender are generally more
pronounced:
- Rates for African Americans in the Medicaid-only population are
always higher than for whites and Hispanics except in Injurious Falls and
Bacterial Pneumonia. In contrast to the dually eligible, Hispanics in this
population do not exhibit lower rates, falling rather in the middle of the
distribution and experiencing rates comparable to whites.
- Rates for women in the Medicaid-only population are always higher
than for men except in Pressure Ulcer.
- Similarly, in the I/DD subpopulation, differences between African
Americans and whites are more pronounced than in the overall population, and in
both the I/DD and the SMI subpopulations, Hispanics tend to fall in the middle.
- The distribution of rates by race in the 65+ subpopulation and
the 18-64 subpopulation without I/DD or SMI look similar to the rates for the
overall HCBS population.
Thus, while the overall HCBS indicators exhibit a general
pattern of higher rates of adverse outcomes for African Americans relative to
whites and lower rates among Hispanics, the I/DD and SMI subpopulations exhibit
a starker contrast between African Americans and whites and the Hispanic
advantage disappears. Across all subpopulations, rates of outcome indicators
are generally higher for women, older adults, and nonurban areas, but the
number of exceptions varies slightly by subpopulation.
Outcomes by HCBS Availability, HCBS Use, and HCBS Policy Environment
Much attention has been paid in the last few decades to
policies that promote the availability and use of home- and community-based
long-term care services, as these services may address unmet need in the
community and avoid costly institutional care. We therefore display in Tables
9-13 the overall outcome indicator rates by the potential availability and use
of specific services.
Table 9 displays outcome indicators by whether a State offers selected services through its State plan, regardless of whether individuals
receive the service. Similarly, Table 10 displays outcome indicators by whether
a State offers selected services through at least one waiver, regardless
of whether the waiver service is universally available or whether an individual
receives the service. Tables 11 and 12 display the outcome indicators broken
down by receipt of the State plan or waiver service, respectively. Finally,
Table 13 exhibits the outcome indicators by key attributes of State Medicaid
and long-term care policy.
The numbers in all of these tables are subject to caveats
and have several plausible interpretations. First, the services listed are
generally offered or used in conjunction with other services, so differences in
rates are unlikely to be due to differences in that one service alone. Second,
differences in rates may be attributable to effectiveness of care—HCBS
participants who receive more services have fewer adverse events—or to
differences in case mix of people from group to group. For example, some services are more appropriate for
individuals at a lower level of need, while other services are directed at a
higher level of need, and outcome indicator rates are likely to be higher in
groups with a higher level of need. Similarly, States that offer more services
or invest more in HCBS may include more individuals with lower levels of need,
so their outcome indicator rates should also be lower. The numbers in these
tables do not adjust for differences in case mix.
Outcomes by HCBS Availability
Outcomes by Availability of State Plan Services
Table 9 depicts outcome indicators by coverage or
availability of services provided under State plans. While State plan services
must be offered to all Medicaid beneficiaries who qualify, actual supply and
quality of services may vary within a State. Our data reflect inclusion in the State
plan as a measure of potential availability.
- Generally, Table 9 shows that outcome indicator rates across most
indicators are lower in States with State plan services that include home
health therapies, hospice care, and transportation (and almost all States offer
these services, as shown in Table 1).
- Rates are also lower in States that offer personal care services,
adult day care and, to a less consistent degree, residential care, services
that are not offered by all States.
- In contrast, higher rates of most outcome indicators are found in
States that offer private duty nursing (offered by more than half of States),
and no systematic results are found for targeted case management (offered by
almost all States).
Outcomes by Availability of Waiver Services
Table 10 presents parallel data to Table 9 but for 1915(c)
waiver services only. In contrast to State plan service coverage, the potential
availability of a service through a 1915(c) waiver is more often associated
with higher hospital admission rates.
- Table 10 shows that outcome indicator rates across most
indicators are higher in States that offer personal care, case management,
adult day/health care, and durable medical equipment and supplies through at
least one 1915(c) waiver in the State.
- Rates are lower only in the case of transportation services being
offered, and results for residential care are mixed.
The generally higher rates in States that offer most of
these services may reflect differences in case mix or may reflect waiver
services being offered in response to high need. In addition, waiver services may
be limited in geography or capacity such that the waiver is not really
available to large numbers of people in a State's HCBS population.
Outcomes by HCBS Use
Outcomes by Use of State Plan Services
Table 11 shows outcome indicators by use of State
plan services, which are roughly consistent with outcomes by availability of State
plan services.
- Admission rates are lower among people who use home health
therapies, personal care services, targeted case management, and adult day
care.
- Rates are higher for people using hospice care, transportation services,
and private duty nursing through the State plan than for people who do not use
these State plan services.
- Results for residential care are mixed.
Outcomes by Use of Waiver Services
Table 12 displays outcome indicator rates by use of specific
waiver services. Whereas outcomes by availability and use are similar for State
plan services, the results for waiver service use presented in Table 12 provide
more of a contrast to results for waiver service availability.
- Indicator rates are dramatically lower for individuals who use
adult day care, residential care, or transportation through a 1915(c) waiver.
- Those using personal care or durable medical equipment and
supplies still exhibit higher rates.
- Results for case management are mixed.
Summary of Outcomes by Use of State Plan Services and Waiver Services
Together, Tables 9 and 11 represent a tendency for lower
outcome indicator rates to be associated with potential availability and use of
many but not all State plan services relevant to HCBS participants.
Tables 10 and 12 represent a more mixed picture: the
existence of a 1915(c) waiver offering a particular service (in at least one of
the States waivers) is not associated with lower rates of outcome indicators,
but use of selected services is associated with dramatically lower rates. These
dramatically lower rates may reflect effectiveness of the services or
differences in the underlying health of individuals who use and do not use
these services.
Outcomes by HCBS Policy Environment
Of interest as commonly used summary measures are outcome
rates with respect to the State policy variables captured in Table 13. First,
we consider the restrictiveness of State eligibility policies with respect to programs
for medically needy people. We classified States into one of three categories
for this variable: (1) the State has no program for medically needy people; (2)
the State has a program but it is quite restrictive in terms of income and
asset requirements; or (3) the State has a program that is less restrictive in
terms of income and asset requirements.
Second, we consider the nursing home level of care (LOC)
necessary for individuals to qualify for many State HCBS waiver programs, the
definition of which varies by State: high (most restrictive), middle, and low
(least restrictive), based on a previously published categorization (Mollica &
Reinhard, 2005). Finally, we consider the percentage of State Medicaid long-term
care funds that are spent on HCBS as opposed to institutional care. This is a
commonly used metric to indicate the extent of State support for HCBS. We
compare rates of outcome indicators for each of these State policy attributes.
For the overall HCBS population, the results in terms of
outcome indicators are strikingly consistent:
- Across all measures, the outcome indicator rates monotonically
increase with the restrictiveness of the program for medically needy people. States
without a program have the highest rates of admissions and those with the least
restrictive policies have the lowest rates.
- Similarly, States with high requirements for nursing home LOC
have the highest admission rates and States with low requirements have the
lowest rates.
- Finally, States that spend more on HCBS as a percentage of total
long-term care spending (Burwell, et al., 2006) have lower rates of adverse
outcomes as reflected in the outcome indicators.
Despite the consistent results, interpretation is difficult.
It is impossible to distinguish with these descriptive results whether greater
access to services reduces rates of adverse outcomes or whether States that are
more supportive of HCBS extend services to healthier individuals, resulting in
fewer adverse outcomes among those served.
The relationship of the State policy variables considered in
Table 13 to the outcome indicators is roughly consistent across subpopulations
and by dually eligible status. For each subgroup, the results exhibit the same
basic pattern but are slightly less uniform than the pattern found in the
overall HCBS population. The main deviations from the overall pattern are in
the nursing home LOC results:
- In the SMI subpopulation, the reverse pattern for nursing home
LOC is found for Asthma or Chronic Obstructive Pulmonary Disease, Congestive
Heart Failure, and Infection Due to Device or Implant. More restrictive
policies are associated with lower rates of admission.
- Among the 65+ subpopulation and the 18-64 without I/DD or SMI
subpopulation, the high LOC groups sometimes have lower rates than the middle
LOC groups.
These deviations from the overall pattern may be related to
specific provisions of the LOC criteria that do not translate well to
restrictiveness for each population for each outcome indicator.
Outcomes by Area Characteristics
Whereas Table 8 displayed HCBS outcome indicator rates by
attributes of the individuals in the HCBS population, Tables 14-19
display outcome indicator rates by aggregated area characteristics
(county and/or State). These characteristics correspond to important individual
attributes such as sociodemographics and health status. The motivation behind
these comparisons is twofold:
- First, it is useful to identify the types of areas and populations in which high rates of adverse outcomes as represented by the HCBS outcome indicators are likely to be present. These areas can potentially be targeted with policies aimed at reducing hospital admissions among HCBS participants.
- Second, differences by individual and area-level attributes can be used to identify predictors of risk that should be included in risk adjustments if the indicators are eventually to be used to compare policies.
Outcomes by Supply of Health Care Providers
In Table 14, we consider the relationship between the
outcome indicators and the supply of health care providers of various types per
capita: acute care hospital beds, nursing home beds, home health agencies,
intermediate care facilities for the mentally retarded (ICFs-MR), and inpatient
psychiatric beds. These rates are calculated first at the county level and then
at the State level.
We also consider whether the county is "underserved" by mental health providers or by primary care providers, two variables defined by the Health Resources and Services Administration in the Area Resource File. These variables are based on number of providers per capita only (not adjusted for need). These "underserved" variables are only available at the county level, not the State level.
In general, one could generate multiple hypotheses about the
expected direction of the association between the outcome indicators and
supply. A greater supply of acute care hospital beds may enable or encourage
more inpatient admissions, so admission rates would be expected to be higher. A
greater supply of nursing home beds and ICFs-MR may mean that individuals who
remain in the community are healthier on average, resulting in lower rates of
admissions (the selection hypothesis). A greater number of home health agencies
may be associated with greater access to home-based care, also leading to lower
admissions.
On the other hand, a greater supply of all types of health
care providers, including inpatient psychiatric beds, might indicate a sicker
population such that outcome indicator rates would be expected to be higher. Higher
rates of hospital admissions in areas with more health care providers might
also reflect greater numbers of medical personnel identifying problems and
encouraging care. In our data, we can observe the direction of the relationship
but cannot distinguish among the multiple hypotheses that may underlie the
observed association. The State-level results are largely consistent with the
latter hypothesis:
- Higher rates of admissions as represented by the outcome
indicators are associated with more acute care hospital beds, more nursing home
beds, more home health agencies, and more inpatient psychiatric beds. This
pattern is highly consistent across all types of providers.
- On the other hand, higher rates of the outcome indicators are
associated with fewer ICFs-MR per capita for the overall HCBS population, more
consistent with the selection hypothesis. However, this result is driven by the
non-I/DD subpopulation for whom the number of ICFs-MR should be less relevant. Among
the I/DD subpopulation, more ICFs-MR per capita is associated with higher rates
of the outcome indicators.
Thus, overall, the State-level results support the
hypothesis that a greater number of health care providers is indicative of a
sicker HCBS population that is more likely to have higher rates of potentially
avoidable hospital admissions. The results also support the hypothesis that
greater attention, identification of problems, and care seeking result from a
greater number of medical personnel in the area.
The two county-level variables representing "underserved"
status provide results that are counterintuitive at first glance:
- Counties that are "underserved" by mental health providers or by primary care providers are consistently associated with lower rates across the outcome indicators.
This phenomenon may also be explained by the sickness of the
population, because areas with more need have more providers, consistent with
the State-level results for supply described above. The other county-level
results in Table 14, however, are less consistent than the State-level results
and sometimes at odds with them.
- There is a general pattern of higher indicator rates in areas
with a greater supply of acute care hospital beds, but the pattern is not as
consistent overall and seems to be driven by the Medicaid-only and I/DD groups.
It is not true of the SMI, 18-64 without I/DD or SMI, or 65+ subpopulations.
- Oddly, the county-level results for supply of psychiatric beds
are in direct opposition to the State-level results, exhibiting higher rates of
the outcome indicators in areas with fewer psychiatric beds.
Although the county-level variables should contain more
specific and relevant information, they also contain more statistical noise,
because individuals do not always use an array of health care providers based
in the same county. In addition, county lines are somewhat arbitrary, so
aggregating to the State level may provide a cleaner picture.
Outcomes by Area-Level Age and Race
Table 15 displays outcome indicators by the area-level
distribution of age and race, first by county and then by State:
- Overall, counties with more non-Hispanic whites tend to have
higher rates of the outcome indicators while areas with more Hispanics tend to
have lower rates, although the latter result is less consistent when looking
only at dual eligibles.
- The pattern for African Americans is inconsistent across measures
overall, but a higher percentage of African Americans is associated with higher
rates of most outcome measures in the I/DD subpopulation.
- The age pattern at the county level shows that a higher
percentage of people age 65+ is associated with higher rates of outcome
measures. The same pattern holds for the percentage of people age 85+. The main
exception to the county-level age pattern is the Pressure Ulcer measure, for
which rates are higher when the percentage of people age 65+ or age 85+ is
lower.
The State-level results are very similar to the county-level
results except that the overall pattern of higher rates in areas with a higher
proportion African American is more pronounced and consistent, especially among
older adults.
Outcomes by Area-Level Socioeconomic Status
Table 16 displays outcome indicators by several area-level
indicators of socioeconomic status: median income, percentage of the population
in poverty, and number of Federally Qualified Health Centers (FQHCs) per
capita. The FQHC variable could be interpreted in two ways. First, it could be
a sign of access to care in low-income groups, which would lead one to expect
lower outcome indicator rates in areas with more FQHCs. Second, it could be a
proxy for poverty and health status in that more FQHCs indicates more
low-income individuals in need of health care.
- At the county level overall, the results exhibit a striking
inverse relationship between socioeconomic status and outcome indicator rates,
with lower income and higher poverty consistently associated with higher rates
of adverse outcomes.
- On average, median income is a stronger predictor of outcome
indicator rates than the percentage in poverty. The difference in rates by
median income is larger across most measures than the difference in rates by
percentage in poverty.
- More FQHCs is consistently associated with higher outcome
indicator rates, consistent with the interpretation that the number of FQHCs is
a proxy for poverty and poorer health status.
- The overall results for socioeconomic status at the county level
are roughly consistent across dual eligibles, Medicaid-only participants, and
the four subpopulations. The results for the percentage of the population in poverty are
considerably less consistent, and even reversed for some measures (e.g.,
infection-related measures) among the elderly subpopulation and the group under
65 without I/DD or SMI. Median income remains a strong inverse predictor of
outcome indicator rates.
- The State-level results for Table 16 are consistent with the
county-level results in terms of median income.
- Percentage in poverty is a less consistent predictor across measures
at the State level and again varies in consistency and even direction at the
subpopulation level. For example, States with a higher proportion in poverty
tend to have lower outcome indicator rates among HCBS participants who are 65
and older. It may be because the group in poverty is concentrated in younger
populations. In this sense, median income may capture a broader sense of
socioeconomic status than percentage in poverty.
The State section of the table includes two additional
variables not available at the county level, the percentage of persons 65+ who
are living alone and the percentage of female labor participation.
- The State-level results are consistent in exhibiting higher rates
of outcome indicators in States with a higher percentage of older individuals
living alone, which one would expect.
- In general, States with a higher percentage of female labor force
participation have higher rates of outcome indicators.
These results are consistent with a lower availability of
informal care, although other explanations are plausible.
Outcomes by Area-Level Health Status
Outcomes by Prevalence of Chronic Conditions
Table 17 displays outcome indicator rates by State-level
prevalence of chronic conditions (diabetes, asthma, cardiovascular disease,
high blood pressure, and serious mental illness). These were not available at a
county level.
- Unsurprisingly, the results for the overall HCBS population show
consistently that areas with higher prevalence of each chronic condition have
higher rates of adverse outcomes as represented by the outcome indicators. The
pattern is strong for all conditions except high blood pressure. Although
generally exhibiting the same pattern, high blood pressure shows a smaller
magnitude of difference between States with high and low prevalence and the
direction of the difference is not consistent for all measures.
- These results are fairly consistent across subpopulations, with
the pattern of results for high blood pressure being the least consistent
across subpopulations but others showing a consistently strong relationship
between chronic condition prevalence and outcomes.
Outcomes by Prevalence of Any Disability
As a parallel to Table 17, Table 18 presents outcome indicator
rates by State-level prevalence of disability of several types. The table shows
the percentage of the population in the following groups: ages 18-64 with any
disability, age 65+ with any disability, ages 18-64 on Social Security
Disability Insurance (SSDI), and people on SSDI with a diagnosis of mental
retardation.
- Outcome indicator rates for the overall HCBS population are
substantially higher in States with a higher prevalence of disability as
indicated by three of the four disability categories.
- Among those age 65 and older, outcome indicator patterns are the
opposite, with slightly higher rates in States with lower prevalence of
disability. This anomalous result varies by subpopulation. Among the I/DD and
SMI subpopulations, higher prevalence of disability is associated with higher
outcome indicator rates for all four measures of disability, while the other
subpopulations drive the overall, slightly less consistent result.
Outcomes by Prevalence of Specific Disabilities Among Older Adults
Table 19 presents outcome indicators by specific types of
disability (sensory, physical, self-care, or mental disability and difficulty
going outside home) among older adults (65+) only.
- Very consistently across almost all outcome indicators, somewhat
higher rates of admissions are associated with higher rates of sensory,
physical, and self-care disabilities and with difficulty going outside home.
- The pattern for mental disability is much less consistent and
tends toward the reverse direction.
- The inconsistency of the overall results for prevalence of mental
disability is driven by stark differences by dual status and subpopulation. Medicaid-only
HCBS participants and the 18-64 subpopulation exhibit the more intuitive
pattern of higher outcome indicator rates with higher prevalence of mental
disability, although prevalence is measured for people 65 and older.
- Oddly, the results for the 65+ subpopulation are generally in the
opposite direction from what would be expected for four of the five types of
disability. Only for prevalence of sensory disability among those 65 and older
does the 65+ subpopulation exhibit higher outcome indicator rates with higher
prevalence.
The results for the 65+ subpopulation are counterintuitive
and difficult to explain, but most of the results for the overall HCBS
population remain intuitive and plausible.
Return to Contents
Proceed to Next Section