Trends in Funding for Health Data
Systems
Importance of Alternatives for Improving Policy
Usefullness of Data
Conclusions
Notes
ACKNOWLEDGMENTS
Jim Knickman, Beth Stevens, and Joel Cantor of the Robert
Wood
Johnson Foundation were instrumental in developing the support
for
this survey. Ira Kaufman encouraged us to include a survey
component with factual reports. Randy Desonia, Steve Long, Mark
Epstein, and Larry Patton reviewed and commented on a draft of
the instrument.
Many MPR staff contributed to the research upon which this
article is based. Lauren Burnbauer was survey director and
played
a key role in developing methods to enhance the quality and
completeness of the data. Joy Gianolio was responsible for
survey
development during the design stage. Richard Strouse provided
advice on survey structure and design. Linda Mendenko was
responsible for sample development. Portia DeFilippes provided
programming support. Karyen Chu and Barbara Foot provided
research support. Ann Cherlow and John Mamer provided advice on
Medicaid systems and vital statistics data respectively. Daryl
Hall edited the manuscript, and Sharon Clark was responsible for
production. Harold Beebout reviewed and provided suggestions on
an earlier draft.
EXECUTIVE SUMMARY
OBJECTIVE AND METHODS
States play an important role in addressing an increasingly
complex array of health policy issues and challenges.
Appropriate
tools, including data and analytical capability, are an important
part of the infrastructure through which policymakers can address
these challenges.
In this paper, we provide a current profile of how state
officials view the adequacy of the information they have for
developing health policy and some features of the data they have
available to them. The article is based on results from a Robert
Wood Johnson Foundation-funded telephone survey of key senior
analytical staff in all the 50 states conducted by Mathematica
Policy Research in early 1994. Included are findings on:
RESULTS
Policymaker Confidence in Health Data Systems
Of 12 policy issues queried, five were clearly dominant in
state concerns in early 1994: Medicaid, cost containment, access
to care and the uninsured, maternal and child health, and managed
care. State officials were most confident in data on Medicaid,
followed by maternal and child health. Less than 10 percent of
respondents were very confident with data available on the other
three policy issues.
State data systems and the confidence policymakers have in
them reflect the history of state health programs and sources of
support. States appear relatively confident in data the supports
ongoing operations of stable programs: monitoring Medicaid,
operating individual public health programs, tracking vital
statistics, and licensing health personnel. However, state data
systems are not perceived as being well-suited to supporting
assessments of program needs or to guiding decisions about
restructuring health care systems in a changing environment.
This
includes information on care in managed care systems and in
ambulatory settings, cost containment, access, prevention and
health promotion, and quality of care.
The body of the paper includes greater detail on the
confidence with data on Medicaid policy issues, public health
issues, health status and health resources, along with some
detail
on the kinds of data of each type respondent say they do or do
not
have available and the analysis it can support. This includes
information on the perceived confidence in the ability to use
Medicaid data to develop programmatic options in the case of
budgetary shortfalls, the ability to link public health data
across programs and jurisdictions, the sophistication and
timeliness of vital statistics systems in each state, and the
information available on health resources including provider and
patient characteristics, use of services, and finances.
Trends in State Data Systems and Funding
Most respondents (68%) perceive that health data are now
better able to meet policy needs than they could three years ago.
More than two-thirds of respondents reported recent budget cuts
in
their state. The survey suggests that state data systems fared
better or at least no worse than other programs, generally
speaking. However, funding for data systems, like other
functions, was cut back, which contributed to more pressure on
staff and reduced scope of and less timely analysis. While the
actual elimination of data systems was rare (9 percent), cutbacks
in the scope of data systems were more common (26%).
States perceive a variety of actions as important in making
data more policy useful. These include both technical
improvements and efforts to better translate data to the policy
audience and apply it to policy issues. The three steps most
likely to be cited as very important were: linking databases
across programs and services (85%), increasing policymakers'
knowledge about using data for policy (83%), and improving
training for and ability of state to apply data to policy issues
(75%).
Respondents believe that the single most important thing the
federal government can do to improve states' ability to
generate
useful population-based health data is to provide funding to
support state efforts (50%). A more coordinated and uniform
approach to data collection by federal agencies, including
standard definitions and reporting requirements; and technical
assistance were also considered important.
CONCLUSIONS
State policy officials are more confident in the ability of
current health data systems to address some health policy issues
than others. In general, confidence levels are higher for policy
issues related to long-standing responsibilities in state
programs
than for newer policy concerns, many of which involve the
relationship between public-sector interests and private sector
delivery and financing of services.
The findings are encouraging in that they show health
statistics to be faring relatively well in the recent climate of
fiscal austerity. However, they also show that the statistical
infrastructure has been somewhat eroded--particularly in terms of
the scope of analysis, timeliness of reporting, and burden on
staff. The challenge facing states in generating more
policy-useful data is complicated by the factors that determine
utility. However, addressing these challenges will be
particularly valuable in helping states respond both to a
changing
economic and social climate and to the pressures for programmatic
and systemwide monitoring that these create.
ABSTRACT
States play an important role in addressing an increasingly
complex array of health policy issues and challenges. In this
paper, we provide a current profile of how state officials view
the adequacy of the information they have for developing health
policy and some features of the data they have available to them.
The article is based on results from a Robert Wood Johnson
Foundation-funded telephone survey of key senior analytical staff
in all the 50 states conducted in early 1994.
Of 12 policy issues queried, five were clearly dominant in
state concerns in early 1994: Medicaid, cost containment, access
to care and the uninsured, maternal and child health, and managed
care. State officials were most confident in data on Medicaid,
followed by maternal and child health. Less than 10 percent of
respondents were very confident with data available on the other
three policy issues. State data systems and the confidence
policymakers have in them reflect the history of state health
programs and sources of support.
Most respondents perceive that health data are now better
able
to meet state policy needs than they could three years ago. Yet,
most states faced budget cutbacks. Health statistics fared
better
or no worse than other programs. However, funding for data
systems, like other functions, was cutback, resulting in some
erosion of the statistical infrastructure. This was most likely
to be manifested in more pressure on staff and reduced scope of
and less timely analysis. Actual elimination of data systems was
rare but cutbacks in the scope of data systems were more common.
The three steps officials are most likely to cite as very
important in making data more policy useful were: linking
databases across programs and services, increasing policymakers'
knowledge about using data for policy, and improving training for
and ability of states to apply data to policy issues.
Respondents
felt that the single most important thing the federal government
could do was to provide funding to support state efforts,
followed
by a more coordinated and uniform approach to data collection by
federal agencies and by technical assistance.
State data systems are less well suited to addressing policy
needs or guiding program restructuring in a changing environment
than to operating stable programs. Yet today's environment
emphasizes change with states in a central position. Thus,
addressing the weaknesses identified in our study will be
particularly valuable and important in the current climate.
INTRODUCTION
States play an important role in addressing an increasingly
complex array of health policy issues and challenges.1 These
include tensions created by efforts to use the Medicaid program
to
promote access in an era of escalating costs; the resurgence of
traditional public health issues associated with infectious
disease and poverty; long-standing concerns with prevention,
health promotion, and the development and distribution of health
care personnel, facilities, and services; and newer concerns
related to integrating public and private efforts to handle these
issues. It is likely that issues associated with integration and
overall health care system performance will become more
important,
especially if states take on new roles in the absence of federal
health reform.
Appropriate tools, including data and analytical capability,
are an important part of the infrastructure through which
policymakers can address these challenges. Yet, the quality and
sophistication of these tools vary considerably from state to
state, and critical weaknesses are apparent in even the most
advanced states.
This article complements our previous analysis of the
strategic issues faced by states in enhancing information for
state health policy.2 It
also compliments our earlier more
focused analysis of survey results. In that analysis, we
concentrated on findings about the adequacy of data to support
state health reform activity.3 This article provides a current
profile of how state officials view the adequacy of the
information they have for developing health policy more broadly,
and it presents some features of the data they have available to
them. The article is based on results from a Robert Wood Johnson
Foundation-funded telephone survey of key senior analytical staff
in all the 50 states. Conducted by Mathematica Policy Research
in
winter 1994, the survey provides timely information on state
policymaker perceptions of the quality of their data.
We review the survey design and methods, followed by the
results and conclusions. Included are findings about the
following:
SURVEY DESIGN AND METHODS
The results in this article are based on a telephone survey
of
key senior state officials in the legislative and executive
branches of all 50 states and the District of Columbia. The
interviews were conducted between January and March 1994 by
executive interviewers in the Survey Center at Mathematica Policy
Research.
The survey was designed to obtain information from eight to
nine individuals in each state. The sample frame was structured
such that we could identify and survey the most appropriate
senior
person responsible for both health policy analysis and for
providing advice on each of eight functions. The following
individuals and/or functions were identified: the governor's
health aide, health analysis for legislative committees; central
budget department staff dealing with health programs; a health
reform entity, that is, the lead agency or task force charged
with
responsibility for health reform; Medicaid agency policy
analysis;
public health agency policy analysis; vital and population-based
health statistics; and the database commission or other agency
responsible for health resource and utilization analysis. We
attempted to include two legislative staffers where possible so
that both houses of the legislative body could be represented.
This strategy resulted in a total possible sample of 452
interviewees, given differences in the structure of state
legislative bodies and staff support in 51 political entities.
We devoted considerable attention to defining and locating
appropriate respondents.4
While some lists of state officials
exist (e.g., state Medicaid directors), they are spotty and not
sufficiently targeted for this study. We therefore identified
respondents through an interactive process using one or more key
informants in each state.5
Definitions of appropriate respondents
were designed to prompt the selection of the most senior person
who is both policy advisor and analyst in each area.
These efforts paid off. While our sample included some
respondents who substituted for others, we were able to interview
senior officials or their staff who met our selection criteria.
The sample consisted of governors' aides from all states except
Idaho, 91 legislative analysts (at least one from all states
except Hawaii), budget department analysts from all states except
Louisiana and New Jersey, 47 staffers in health reform entities
(excluding Arizona, Idaho, Kentucky, and New Jersey), and
individuals from each of the other four functional areas
(Medicaid, public health, vital statistics, health
resources/database). Of 452 potential respondents, 442
interviews
were completed;6 item
response rates were also high. At least 7
respondents answered for each state.
The instrument was structured so that all respondents were
generally asked a core set of questions. Governor's and
legislative aides, as well as budget department staffers (termed
"central policy officials") were asked to respond in
general,
while other individuals were asked to respond in terms of data
relevant to their area of responsibility. Questions in any given
policy area were asked only of the central policy officials and
the other respondents responsible for this area. For a more
general understanding of how well state data matches policy
needs,
we also asked central policy officials to rate the importance of
various policy issues and assess their confidence with data
available on these issues. Items in the instrument generally
were
in this order moving from general to more specific policy areas.
The unit of analysis, for the most part, is the respondent.
Because the number of respondents is generally proportional to
the
number of states, the information we obtained reflects
perspectives across all states. We decided not to construct
individual measures here for each state because it would have
been
more difficult to present results and because we perceived that
it
might make it appear as if perspectives were consistent across
respondents in a given state when this was not always true. We
deliberately focused on policymaker perceptions of the
availability and adequacy of data rather than to actually audit
the data available. Perceptions obviously have an element of
subjectivity. However, the extent to which data are useful
depends very much on policymakers' perceptions, knowledge, and
confidence about what is available. In addition, data systems
are
sufficiently complex that we did not regard it feasible to
realistically describe them in detail using a survey approach.
Given the structure of the survey and the purpose of the
study, this article emphasizes patterns across states rather than
performance of individual states. However, we also believed it
was important to place policymakers' perspectives in some context
in order to interpret or validate their perceptions. We
therefore
asked respondents who are responsible for certain programs to
answer certain factual questions about the data actually
collected. (These respondents are referred to as "Medicaid
program respondents," "public health respondents,"
"vital and
health statistics respondents," and "health
resources/database
respondents.") These factual items were obtained last to
avoid
influencing responses on perceptions.
RESULTS
Confidence in Health Data Systems to Address
Priority Health
Issues
Central policy officials are more confident in the ability
of
current health data systems to address some health policy issues
than others. However, only a minority are "very
confident" about
the ability of the data systems to address most issues, and
confidence levels are low for several high- priority issues. In
general, confidence levels are higher for policy issues related
to
long-standing responsibilities in state programs than for newer
policy concerns, many of which involve the relationship between
public-sector interests and private sector delivery and financing
of health services.7
However, even in these areas, the level of
confidence suggests that there is considerable room for
improvement in the data.
Table 1
shows how central policy officials rated 12 policy
issues of potential concern to states. Each was rated as at
least
"medium" priority by three-quarters of all respondents
and most
issues were rated this way by considerably more respondents. It
is clear that there were five dominant issues in early 1994:
The level of
confidence is a bit lower for data on maternal and child health
(40 percent very confident, 46 percent somewhat confident). In
contrast, only a little more than 10 percent of respondents are
very confident with data available on cost containment, access
and
the uninsured, and managed care. Confidence levels are similarly
low for data on prevention/health promotion and quality of
care--issues that one-third or more of respondents viewed as high
priority.
Confidence in Data on Medicaid Policy
Issues
Medicaid is the second largest expenditure in most states,
just behind combined state costs of elementary and secondary
education.8 Medicaid
spending has been growing rapidly for a
variety of reasons.9 With
a program structure based on
entitlement, Medicaid spending is less easily predicted or
controlled than spending for the discretionary, grant-funded
programs. Additionally, federal legislation has mandated
significant changes to the Medicaid program (e.g., OBRA
expansions
in eligibility, EPSDT expansions) with little lead time and
limited attention to the effect on program resources at the state
level. Since Medicaid is such a large part of most state
budgets,
trends in spending need to be monitored very closely so that
cost-containment initiatives can be developed on a timely basis
to
keep spending within the budget, a legal requirement in most
states. In the Medicaid portion of the survey, we therefore
focused on asking respondents to assess how well their state
Medicaid information systems can help with analysis relevant to
this issue, that is, whether the systems can guide a
policy-related a response when spending is projected to
considerably exceed the budget for the current year and action is
necessary. We focused particularly on the ability of systems to
identify the reasons for higher-than-expected expenditures, since
this could help to identify potential solutions to this problem,
and to analyze which solution it would be most useful to pursue.
Our results suggest that state officials are reasonably
confident in the ability of their state Medicaid information
systems to address the issue of Medicaid spending, although there
appears to be room for improvement (Table 2
). Almost all
policymakers are either very confident or somewhat confident in
the ability of their systems to identify why expenditures may be
higher than expected; central policy officials and Medicaid
policy
respondents are similarly confident. Overall, 41 percent of the
former and 39 percent of the latter are very confident in their
system's ability to support analysis of this issue; only about 10
percent are not very confident or not at all confident. However,
policymakers are less confident about the ability of state
Medicaid information systems to support analysis of options for
reducing expenditures under these circumstances: only 31 percent
of central policy officials and 21 percent of Medicaid
respondents
are very confident about data system capabilities in this area.
To independently measure the capacity of state Medicaid
systems to support decision making, we asked Medicaid policy
respondents about the ability of their systems to isolate
specified kinds of Medicaid expenditures (
Figure 1
). Virtually all said they can isolate expenditures for services
provided to
special enrollee groups, such as pregnant women or children, and
for newly covered or expanded services, such as case management
programs or services provided under waivers. Medicaid policy
respondents also said they can track the impact of various
reimbursement policies by identifying expenditures such as cost
settlements (90 percent) or revenues such as provider-specific
taxes or donations (80 percent). Roughly three-quarters can
isolate spending for newly covered enrollees or services that are
provided by other departments and reimbursed through fund
transfers. Of the areas covered in the survey, states are least
able to isolate expenditures associated with care by capitated
plans (69 percent). Note that this figure does not refer to the
ability to count services for managed care enrollees, a major
need
of states, but instead refers to access to data on current
expenditures for capitated enrollees through premium payments.
These survey findings may at first seem counter-intuitive
because of recognized problems with Medicaid data. We speculate
that our findings illustrate two issues. First, states find it
easier to work with Medicaid data than the federal government
does, as the federal government often struggles to collect
comparable and timely data across all states. Second, state
expectations of the scope of analysis that data might need to
support may be different from those of federal government
officials or researchers. Other work, generally indicates that
many states do not have well-developed modeling or statistical
capabilities and have difficulties tracking certain kinds of
expenditures.10 On the
other hand, because they can draw on
bills, Medicaid data may be better than data for other state
programs, a factor influencing state perceptions.
Confidence in Data to Address Public Health
Issues
To gain insight into the ability of data to support
decisions
related to public health, we asked respondents if they believed
that their state data might help them decide if funding for
public
health clinics should be reduced as part of state budget cutting.
Responses to this common example of a public health policy
initiative can tell us something about how well state information
systems help to track service use, spending, and outcomes so that
this information can be used in the policy process. The
information is important because it provides empirical evidence
of
the need for a program (e.g., by showing the absence of
alternatives and coverage of those served), what is to be gained
from a program (e.g., by showing individuals reached and
outcomes), and program cost-efficiency (e.g., by showing cost per
unit of service). This information is particularly important
today in that those responsible for public health programs are
now
considering whether to affiliate with managed care plans, and
policymakers are assessing the extent to which the traditional
public health infrastructure needs to be maintained.
State officials are generally not very confident in the
ability of their systems to provide all the data relevant to
this
issue (Table 3
). They are most confident about the ability to
identify the number of clinic visits in the past year--one-third
of the central policy officials and half of the public health
respondents are very confident about the data available for this
kind of analysis. Respondents are least confident about being
able to identify alternative sources of care for clinic users and
about the effects of the clinics on health outcomes. More than
half the respondents are not very or not at all confident about
the data available for these kinds of analyses, and less than 10
percent are very confident. Respondents also have little
confidence in the ability of data to support analyses of payer
source for visits, the implications of a 20 percent cut on
service
capacity, cost per unit of service, and the unduplicated count of
all individuals served by county health services. Although
central policy officials and public health respondents may see
this issue differently, there is no pattern to these differences,
and they could reflect random variation.
We also asked public health officials about how they handle
the analysis of encounter data (
Figure 2
). Certain analyses,
such as those involved in identifying the number of unique
individuals when some individuals participate in more than one
program, are difficult to perform unless program data at the
individual level are centrally available. Seventy percent
of
public health respondents said they are required to report data
at
this level--either alone or in conjunction with aggregate
reporting. However, of this 70 percent, only 54 percent said
they
are able to link encounter data within programs by patient, 40
percent can link encounter data to cost of services, 37 percent
can link these data across programs for patients in the same
jurisdiction, and 14 percent can link the data across both
programs and jurisdictions. Clearly, the small number of states
with such capabilities to link data weakens the analytical
capability states have to address the issues of service delivery
across public health programs. In an environment in which
managed
care is growing and public health
departments are trying to
position themselves to function in this environment, the absence
of data on service delivery could jeopardize their success and
make it difficult to respond to any policy concerns about the
continued need for public health programs.
All public health departments do not have the same
structure,
a factor that could affect how difficult it is to generate data.
Among our respondents, 36 percent of local service sites are
politically autonomous, 25 percent are agents of states, and 40
percent are a mixture. Public health departments also deliver
services differently. Half use a single integrated facility in
each jurisdiction to provide care, and half use multiple
facilities. However, we did not detect a clear relationship
between the form of organization in terms of the degree of
autonomy and the ability to provide the kinds of data listed in
Table 3. The one exception is that totally autonomous sites tend
to be less likely to provide data of the type covered in the
survey. It is most likely that the form of organization
reflects,
at least in part, the complexity of the state and its level of
resources as well as other less measurable factors. However,
because of the complex relationships between form and other
variables its not possible to identify any specifically whether
structure of delivery affects data availability.
Almost all respondents (92 percent) reported that major
initiatives are underway to integrate health data systems, for
instance, across programs or between the public and private
sector. Responses to open-ended questions suggest that states
interpreted the question on systems integration broadly. They
mentioned a range of efforts from linkage projects that were not
described in detail (21 respondents) to ones involving a range of
public health program data (12 respondents), to other linkage
projects involving vital statistics, and hospital discharge and
similar data (7 respondents), to more narrow data projects (7
respondents). Some respondents referred to projects that are
supported in whole or in part with federal or foundation funds.
Some of the more interesting integration efforts focusing on
public health program data include developing an automated family
health management system for the range of programs in this area;
developing an automated system across programs for making
appointments for patients with the ultimate goal of expanding it
to encounters; creating an automated eligibility screening
system,
which would allow prospective clients to identify at one central
location which programs are available to them in the state; and
integrating client encounter data within the same mini-computer
with the goal of cross-program analysis. Other approaches
include
creating a community health management information system as well
as a data institute involving the public and private sectors. In
many cases, these projects were still in the planning or
implementation stage at the time of the survey.
We also asked specifically about initiatives to monitor
progress against Year 2000 objectives, an important national and
public health initiative. Eighty-four percent of public health
respondents indicated that they have initiatives of this type.11
Of the 43 state efforts, 16 are state funded, 2 are federally
funded, and 25 are funded by both.
Federal funding has historically influenced the development
of
public health program data, both because the federal government
requires certain program-specific systems and reports and because
of the design of and support for more general population-based
health data (e.g., vital statistics, communicable disease
reporting). Both public health and vital statistics respondents
agree with this view, with 49 percent and 34 percent,
respectively, agreeing strongly or somewhat that "federal
funding
has played a major role in the development of public health
program data." The greatest positive effect of federal
support
for health data on the population is to improve the ability of
states to implement, maintain, and utilize data collection
systems
(21 percent of respondents), although respondents did not always
specify what the improvements were (e.g., new staff, new
equipment, new funds, or other assistance). Responses to the
categorical nature of federal funding were mixed: 17 percent of
respondents reported that it has a positive effect on the
development of particular data bases, while 22 percent considered
it a fragmentation of systems that causes numerous and
duplicative
reporting requirements. Twenty percent of respondents said that
federal support does not negatively affect their state efforts.
Eighteen percent reported that federal funds are insufficient to
meet data requirements.
Confidence in Data to Address Issues of
Health Status and Health
Resources
General Levels of Confidence
Data on the health of the U.S. population and on health
resources are useful for addressing a variety of policy issues.
For example, data on the health of the population are important
for designing and evaluating initiatives to promote health,
designing programmatic initiatives, and assessing the impact of
health reform. Data on health resources--conceived here widely
as
data on health facilities and providers, service utilization, and
spending--are useful for examining access, designing and
monitoring cost-containment programs, and exploring health reform
needs. For these reasons, we focused our policy questions in the
survey on general confidence in the ability of data to address
the
issues of the population's health or health resources. We also
posed these questions to a range of respondents broader than that
addressed by the previous questions.12
Relatively few respondents are very confident about either
kind of data, yet only a minority were not very or not at all
confident (Table 4
). That is, about two-thirds of those
surveyed,
regardless of position, tend to be somewhat confident in the
ability of data to support analyses of policy issues related to
population health status and health resources.
Vital Statistics and Other Data on Health
Status
Vital statistics, such as birth and death records, are the
most developed and long-standing statistics on the population's
health. Data on reported communicable diseases also have a long
history, and the comparability and quality of these data has been
enhanced by extensive federal-state cooperation.
For simplicity, we primarily used birth certificate data as
the lens through which we examined traditional sources of
information on the population's health. Perceptions on the
quality of these data vary widely both by state and by data
element of interest (
Figure 3
). Virtually all respondents
perceive
legal data such as name, address, and date of birth to be either
excellent (67 percent) or good (31 percent). Demographic data
are
similarly perceived, though when compared with legal data, they
are more likely to be viewed as good (45 percent versus 31
percent
for legal data) rather than excellent (45 percent versus 67
percent for legal data). In contrast, most respondents view data
on pregnancy history, labor and delivery, and the condition of
the
newborn as good to fair; and they are least comfortable in data
on
risk factors (e.g., tobacco, alcohol), which 74 percent say are
either fair or poor.
Data on communicable diseases tend to be viewed as good (55
percent) or fair (36 percent). Virtually all respondents (50 of
51) say they use Center for Disease Control (CDC) definitions of
communicable diseases. Most also report that local disease
surveillance efforts use these definitions either always (41
percent) or most of the time (51 percent); the remainder
indicated
either that the definitions are sometimes used or that they do
not
know if they are used.
The sophistication of vital statistics systems varies by
state
(
Figure 4
). Forty-seven of the 51 jurisdictions have a formal
quality control procedure for birth certificates at the state
level. The percentage of birth certificates returned to
hospitals
to be corrected so they meet minimum reporting requirements is 5
percent or less for 64 percent of respondents, but about
one-third
returned more than 10 percent. Electronic birth certificates are
common (73 percent), but the percentage of certificates reported
electronically varies widely (e.g., less than half for 35 percent
and 90 percent or more for 16 percent). States also vary in the
timeliness of reporting (
Figure 5
). Twenty-six states have
published birth data for 1992 or later (as of early 1994), but
four states have published data only for 1989, and another four,
for 1990. The pattern for mortality data is similar.
It was common for respondents to indicate that their birth
certificate data are linked to one or more other files.
Ninety-four percent of the states link birth certificates to one
or more files. They are typically linked with Medicaid files (61
percent) or WIC data (50 percent), but linkages with hospital
discharge data (37 percent), AFDC (22 percent), and various other
public health data sets are also common. Only five states have
linked records with data from the Food Stamp Program. About half
of those with linked data reported that some effort is required
to
use the linked files in data quality studies.
Although not shown in the table, all states create a linked
live birth/death file that is used to build a national linked
birth/death file under a cooperative agreement with NCHS.
In monitoring the population's health, it is important to
obtain better information on morbidity and chronic as well as
acute diseases. As life expectancy increases, such data become
more important. In addition, morbidity data can signal new
health
or access problems, since these problems are more likely to be
reflected in illness than in death. Therefore, creating viable
monitoring systems is a critical current concern. The
considerable restructuring that is underway in health care
delivery as a result of the growth of managed care and some
health
reform initiatives makes this even more important, since as
systems change interest grows in understanding the extent to
which
performance improves or erodes. Reportable diseases and
registries, along with surveys of the population, are mechanisms
that generate data on chronic diseases and morbidity. Three
quarters of vital statistics respondents said they use reportable
disease files, 88 percent use registries, and 84 percent use
population-based surveys to obtain information on chronic
diseases. However, these sources were reported to vary in terms
of completeness.
Many respondents interpreted our questions on reportable
chronic disease files and registries differently and not all
files
and registries named appear to be chronic diseases. Cancer was
by
far the most widely developed data on both items. Other files
mentioned include tuberculosis, HIV/AIDS, birth defects, diabetes
and sexually transmitted diseases.
Data on Health Resources
Health resources include the kinds and distribution of
health
personnel, facilities, and services. We also include the output
of these resources in terms of health care use and spending.
Information on the resources available in a state, and their
characteristics and distribution can be important in developing a
diverse range of programs, ranging from support for personnel
training and incentives for location in underserved areas to
certificate-of-need and other regulatory programs. Information
on
utilization patterns and trends can be valuable in monitoring
health system performance and in developing initiatives to
address
any disparities between goals and current performance. Data on
spending are similarly useful, and essential, if the objective of
a given initiative is to contain or monitor spending against any
aggregate targets.
Information on health resources includes provider and
patient
characteristics, use of services, and finances (Table 5).13 Our
results indicate that required reporting varies considerably
across types
of providers. Facilities, especially acute care
hospitals, are more likely to have reporting requirements. Only
about half or fewer states collect information from other
providers, such as ambulatory care facilities or physicians, even
though they represent a growing share of spending and health care
delivery. For institutional providers, there are no clear
patterns in terms of the kinds of data collected. For individual
providers, data are more likely to be collected on provider
characteristics than on use or finances. We expected that this
might be the case, since these data are often collected as part
of
the licensing process. We used physicians and nurses as examples
in questions about the kinds of data states obtain on provider
characteristics (
Figure 6
). In both cases, considerably more
states collect data on training than on practice characteristics
of the providers (72 to 68 percent versus 30 to 15 percent).
Graduate training and board certification are examples of the
former, whereas setting of practice and hours worked are examples
of the latter. Thus, the pattern of results indicates that
states
should be better positioned to assess the quality of their labor
pool than the adequacy of its distribution and availability. And
states are much less well-positioned to monitor system outputs
(use and spending, as well as outcomes) than inputs (resources).
Trends in Funding for Health Data
Systems
Most respondents (68 percent) perceive that health data are
now better able to meet policy needs than they could three years
ago; all but 7 percent said their capabilities stayed at least
the
same (Table 6
). Those closest to the programs did not see as
much
improvement as those in central policy positions.
During a time when state budgets were often cut, state data
systems fared better or at least no worse than other programs,
generally speaking. However, funding for data systems, like
other
functions, was cut back, which contributed to more pressure on
staff, reduced scope of and less timely analysis, and less
commonly, cut backs in the scope of systems and actual data
collection.
As can be seen in Table 6, about half the respondents described
the
recent trend in state funding for health data systems as
increasing, and most of the rest described it as staying the
same.
However, a less positive profile emerges if we look
at
how respondents described the trend in state funding to support
health data systems compared with the state budget in general.
Thirty-two percent said the trend shows an increase in funding,
20
percent observed a decrease, and about half (48 percent) said
funding has stayed the same. Program staff are less than likely
central policymakers to have perceived increases, and public
health and vital statistics staff are least likely to have
observed an upward trend.
More than two-thirds of respondents reported recent budget
cuts in their state (Table
7
). Respondents in these states most
commonly reported that funding to develop health statistics was
cut proportionately with other funding, but about half the
respondents indicated that the budget for health statistics was
either cut less than that for other areas or actually increased.
This runs somewhat counter to the common perception that data
functions lose out to programs in funding decisions.
Nonetheless,
while data systems were potentially somewhat protected from the
fiscal environment, respondents still perceived that the fiscal
climate contributed to erosion in their statistical systems. The
most common manifestation of this squeeze was to put more
pressure
on staff to produce output (84 percent). Less timely analysis
was
also common (49 percent), as was reduced scope of analysis or
reporting (34 percent). The actual elimination of data systems
was more rare (9 percent), though cutbacks in the scope of data
systems were more common (26 percent). Not surprisingly,
cutbacks
were often targeted at functions that involve general purpose
data, like vital statistics and health resource/database systems,
rather than program-related data. Respondents responsible for
these functions were also the most likely to report
disproportionate cuts in their budgets.
Importance of Alternatives for Improving
Policy Usefulness of Data
The task of making data more policy useful is
multidimensional, involving both technical improvements and
better
translating data to the policy audience and applying it to policy
issues (Table 8
). The nature of this task is reflected in the
three steps most likely to be viewed as very important to
improving the ability of state data systems to support policy
needs: linking databases across programs and services (85
percent), increasing policymakers' knowledge about using data for
policy (83 percent), and improving training for and ability of
staff to apply data to policy issues (75 percent). Defining data
in a consistent way, ensuring the timeliness of information,
creating new and relevant data, increasing the skill level of
data
production staff, developing policy-relevant briefings, and other
steps are also viewed as important. Even though it is common to
fault federal data requirements for inhibiting the development of
useful state systems, respondents were least likely to view
reducing such requirements as very important (17 percent). This
is consistent with previously discussed findings.
Respondents believe that the single most important thing
the
federal government can do to improve states' ability to
generate
useful population-based health data is to provide funding that
would support state efforts both in general and for specific
enhancements to their systems (50 percent). Also reported as
being necessary was a more coordinated and uniform approach to
data collection by federal agencies, including standard
definitions and reporting requirements (15 percent). Technical
assistance was also considered important (12 percent).
CONCLUSIONS
State data systems and the confidence policymakers have in
them reflect the history of state health programs and sources of
support. States appear relatively confident in data that
supports
ongoing operations of stable programs: monitoring Medicaid,
operating individual public health
programs, tracking vital
statistics, and licensing health personnel. However, state data
systems are not perceived as being well-suited to supporting
assessments of program needs or to guiding decisions about
restructuring health care systems in a changing environment. It
is noteworthy that, in our example of Medicaid policy, there is
greater confidence in diagnostics than in interventions. At a
time when many states are moving to managed care, it is revealing
that Medicaid systems are perceived as being least likely, in our
illustrative examples, to help policy officials isolate capitated
expenditures, and that almost half of all respondents are not
very
or not at all confident in managed care data, or they do not have
it at all. In addition, because public health data lack the
analytical flexibility that would allow policymakers to look
across jurisdictions, programs, or individuals, these data are
inadequate to guide a policy response to a changing health care
environment and provide the information needed to appropriately
restructure the use of resources.
Our analysis shows that data systems fall shortest in terms
of
providing the information needed to assess systemwide performance
in today's environment. Even though health care is clearly moving
to an ambulatory setting, the number of systems that provide
information about this setting and the providers who operate
within it are much fewer than those connected with the
institutional and inpatient settings. At a time when
"report
cards" on access, quality, and cost are a major subject
interest,
it is of some concern that half or more of the respondents
perceive that data on cost containment, access, prevention and
health promotion, and quality of care are either unavailable or,
if they are available, that respondents are not very or not at
all
confident in them.
For those concerned about these shortfalls, it is
encouraging
that our survey findings indicate that health statistics fared
relatively well in the current climate of fiscal austerity.
However, the statistical infrastructure has been somewhat
eroded--particularly in terms of the scope of analysis,
timeliness
of reporting, and burden on staff. General purpose data on the
population's health and health resources, which serve a variety
of
policy needs, were also relatively more likely to suffer as a
result of cutbacks, with reductions in scope or, potentially,
elimination of entire systems. In the face of these divergent
trends, it is encouraging that more than two- thirds of the
respondents perceive that health data are better able to meet
policy needs than they were three years ago.
The challenge facing states in generating more policy-useful
data is complicated by the factors that determine utility. It is
revealing that 11 of the 17 measures we gave for improving the
ability of state data systems to support policy needs were judged
as "very important." These measures span a range of
technical and
strategic considerations that may influence the utility of
information in the policy process. In our previous work on the
topic of enhancing data for state health policy, we identified
some ways in which states can confront strategic issues. We also
found that policymakers see funding as one of the top three
barriers to enhanced data, at least for health reform.14 Federal
funding has been useful in enhancing state systems but it also
has
had some negative effects. Opinions are split on the effects of
categorical funding on data systems.
Clearly, it will not be easy to address the challenge of
generating policy-useful information. Yet, the experience of the
past few years appears to show that it is possible to improve
data. These improvements will be particularly valuable in
helping
states respond both to a changing economic and social climate and
to the pressures for programmatic and systemwide monitoring that
these changes create.
END NOTES
See Penny Feldman,
Marsha Gold, and Karyen Chu, "Enhancing
Information for State Health Policy." Health
Affairs, Summer 1994,
pp. 236-250.
The earlier paper by
Marsha Gold, Lauren Burnbauer, and Karyen
Chu entitled "Half Empty or Half Full: The Capacity of State
Data to
Support Health Reform" was presented on October 31, 1994 at
the
American Public Health Association in Washington, DC. The
manuscript,
which is being reviewed for publication, may be obtained from the
authors.
Additional details on
survey methods are included in Lauren
Burnbauer, "The Survey Methodology Report," March 1994.
This document
is available from the author.
We used the State
Yellow Book to identify an appropriate
initial contact in the governor's office who was asked to verify
(where previously information existed) or suggest (where it did
not)
the appropriate person for each function in the sample. To this
end,
we developed definitions for each function and decision rules for
handling unique circumstances (e.g., preference priority for type
of
legislative staff, when one person is appropriate for more than
one
function, etc). In some cases, we were referred to people
outside the
governor's office who could identify respondents for particular
functions such as legislative staffing. In scheduling
interviews, we
again used these definitions to confirm the appropriateness of
the
respondent.
There were 34 multiple
respondents (16 percent of cases), so
this involves 406 different respondents. Multiple respondents
were
interviewed in 29 states, with only two cases involving a single
respondent for more than two functions.
This finding is
consistent with the conclusions drawn in our
earlier analysis of the capacity of state data systems to support
health reform. Op. cit., note 3.
U.S. General Accounting
Office, "Medicaid-States Turn to
Managed Care to Improve Access and Control Costs",
Washington, DC:
March 1993.
See, for example, John
Holahan, David Liska, and Karen
Obermaier, "Medicaid Expenditures and Beneficiary Trends
1988-1993,"
The Kaiser Commission on the Future of Medicaid, Washington, DC,
September 1994.
Ann Cherlow, personal
communication, November 17, 1994.
See U.S. Department
of Health and Human Services, "Healthy
People 2000: National Health Promotion and Disease Prevention
Objectives." DHHS Publication No. (PHS) 91-50213.
Washington, DC.,
September 1990.
The respondent set
was narrower in one respect. We did not ask
the budget department official these questions. Since these data
are
not as directly tied to programs as were other areas of inquiry,
we
felt that budget officials may not be very knowledgeable in these
areas.
The responsibility
for these kinds of data is spread out in
some states. Therefore, it is possible that some kinds of data
are
underreported, particularly if our respondents were not aware of
what
might be collected by licensing agencies, boards, or insurance
agencies.
Within each state, we interviewed eight to nine senior state
officials. The following individuals and/or functions were
identified: the governor's health aide, health analyst's for
legislative committees (one to two individuals, central budget
department staff dealing with health programs; the lead agency or
task force charged with health reform; Medicaid agency policy
analysis; public health agency policy analysis; vital and
population-based health statistics; and the database commission
or
other agency responsible for health resource utilization
analysis.
Responses to items on health reform are the subject of a separate
paper.
While each is viewed as a high-priority issue by well over half
of
the respondents, the data on Medicaid clearly generated the
greatest confidence among state officials, a finding consistent
with other survey results discussed later (60 percent very
confident, with 33 percent somewhat confident).
See, for example, D.E.
Altman and D.H. Morgan, "The Role of
State and Local Government in Health." Health
Affairs, Winter 1983,
pp. 7-31.
Return to Top
Return to INFOSHP Home Page