PEDIATRICS Vol. 108 No. 5 November 2001, pp. 1135-1142
Differential Impact of Recent Medicaid Expansions by Race and Ethnicity
,
, ,, , and
From the * Albert Einstein College of Medicine, Children's
Hospital at Montefiore, Bronx, New York; Objective. Between 1989 and 1995, expansions in Medicaid eligibility provided publicly financed health
insurance to an additional 7 million poor and near-poor children. It is
not known whether these expansions affected children's insurance
coverage, use of health care services, or health status differently,
depending on their race/ethnicity. The objective of this study was to
examine, by race/ethnicity, the impact of the recent Medicaid
expansions on levels of uninsured individuals, health care service
utilization, and health status of the targeted groups of children.
Methods. Using a stratified set of longitudinal data from
the National Health Interview Surveys of 1989 and 1995, we compared
changes in measures of health insurance coverage, health services
utilization, and health status for poor white, black, and Hispanic 1- to 12-year-old children. To control for underlying trends over time, we
subtracted 1989 to 1995 changes in these outcomes among nonpoor
children from changes among the poor children for each race/ethnicity
group. Measures of coverage included uninsured rates and Medicaid
rates. Utilization measures included annual probability of visiting a doctor, annual number of doctor visits, and annual probability of
hospitalization. Health status measures included self-reported health
status and number of restricted-activity days in the 2 weeks before the
interview. Differences in means were analyzed with the use of
Student's t tests accounting for the clustering sample
design of the National Health Interview Surveys.
Results. Among poor children between 1989 and 1995, uninsured rates declined by 4 percentage points for whites, 11 percentage points for blacks, and 19 percentage points for Hispanics.
Medicaid rates for these groups increased by 16 percentage points, 22 percentage points, and 23 percentage points, respectively. With respect
to utilization, the annual probability of seeing a physician increased 7 percentage points among poor blacks and Hispanics but only 1 percentage point among poor whites (not significant) for children in
good, fair, or poor health. Among those in excellent or very good
health, the respective increases were 1 percentage point for poor
whites (not significant), 7 percentage points for poor blacks, and 3 percentage points for poor Hispanics (not significant). Significant
increases in numbers of doctor visits per year were recorded only for
poor Hispanics who were in excellent or very good health, whereas
significant decreases in hospitalizations were recorded for Hispanics
who were in good fair or poor health. Measures of health status
remained unchanged for poor children over time. The recorded decreases
in uninsured rates and increases in Medicaid coverage remained robust
to adjustments for underlying trends for all 3 race/ethnicity groups.
With respect to adjusted measures of utilization and health status, the
only significant differences found were among poor blacks who were in
good, fair, or poor health and who registered increases in the
likelihood of hospitalization and in poor Hispanics who were in
excellent or very good health and who registered decreases in the
numbers of restricted-activity days.
Conclusions. Recent expansions in the Medicaid program
from 1989 to 1995 produced greater reductions in uninsured rates among
poor minority children than among poor white children. Regardless of
race/ethnicity, poor children did not seem to experience significant
changes during the period of the expansions in either their level of
health service utilization or their health status.Medicaid, health insurance, health status, health services.
Baruch College, New York,
New York; § City University of New York, New York, New York; and
National Bureau of Economic Research, New York, New York.
ABSTRACT
Top
Abstract
Methods
Results
Discussion
References
Expansions in the Medicaid program enacted by Congress
during the past decade increased the number of poor and near-poor
children who are eligible for publicly financed health insurance.
Beginning with the 1902(r)(2) provisions of the Social Security Act,
passed in 1988 and amended in 1989 and 1990, states were permitted to extend Medicaid eligibility to groups of women and children using expanded income and asset thresholds. Initially targeted only at
families that earned between 100% and 133% of the federal poverty level (FPL), these and later Medicaid expansions eventually combined to
provide insurance to 7 million previously uninsured children from 1989 to 1995.1
In cross-section analyses, provision of health insurance has been shown
to be associated with increased use of primary care services.2-5 A more recent longitudinal study found
similar results.6 Compared with uninsured children, those
with insurance (public or private) are reported to have greater access
to primary care as measured by having a usual source of care, a regular
physician, access to care after normal business hours, and greater
satisfaction.7 Consistent with this finding, insured
children also have been shown to be more likely to receive medical care
from a physician for specific ambulatory conditions that are amenable
to primary care intervention,8 yet despite the accumulated
evidence regarding the relationship among insurance coverage, access,
and utilization, certain questions remain unanswered.
One question is whether the expansion of Medicaid to near-poor children
had equivalent effects on white and nonwhite eligible populations. In
past cross-section analyses, race/ethnicity has been shown to be a
powerful predictor of utilization of health care services independent
of insurance status.9-13 Minority children have been
shown to have lower rates of health insurance coverage, less
utilization, and worse health status than their white counterparts. The
question of whether the Medicaid expansions resulted in relatively
larger improvements in utilization of health care services for
disadvantaged minority children is, therefore, an important one with
relevant public policy implications. Equally important is the question
of whether adoption of these new policies had differential effects on
the health status of members of the affected populations depending on
their race/ethnicity.
In this study, we used a stratified set of longitudinal data to examine
the differential impact of recent Medicaid expansions by
race/ethnicity. Using a before-and-after comparison of health care
service utilization and health status of poor white, black, and
Hispanic children relative to their nonpoor counterparts, we estimate
the differential impact of recent changes in Medicaid policy on these
outcomes among poor families.
Data
We used the 1989 and 1995 National Health Interview Surveys
(NHIS)14 to compare changes in the health insurance coverage, health care utilization, and health status of white, black,
and Hispanic poor and nonpoor children. The NHIS is a national probability sample of the civilian noninstitutionalized population that
collects information on family background, health conditions, and
health care utilization. We chose 1989 and 1995 because these years
bracket the largest increase in Medicaid enrollment of children.
In the early part of this period, Medicaid eligibility was decoupled
from the family structure requirements of state Aid to Families with
Dependent Children (AFDC) programs and states were permitted to extend
eligibility for Medicaid coverage beyond AFDC levels up to 100% of the
FPL. By July 1, 1991, states were required to cover all children
younger than 6 years from families with incomes up to 133% of the FPL
(and were allowed to cover those in families with incomes up to 185%
of the FPL). Additional coverage requirements included all children up
to 19 years of age who were born after September 30, 1983, and whose
family incomes were below 100% of the FPL. These changes resulted in
an increase in Medicaid enrollments of almost 7 million, or 70%,
between 1989 and 1995.15
Our sample included children ages 1 to 12 years. We eliminated infants
because they were eligible for Medicaid coverage under the expansion
thresholds for pregnant women that were more generous than those for
children.16 Moreover, enrollment of infants was
facilitated by streamlined administrative procedures such as
presumptive eligibility. We excluded children between 13 and 17 years
of age because the expansions in Medicaid eligibility applied only to
children who were born after September 30, 1983. The initial sample of
children ages 1 to 12 years for the years 1989 and 1995 numbered
42 140.
Poverty, Race, and Ethnicity
Given the legislative changes that occurred between 1989 and
1995, the populations that were affected by the Medicaid expansions varied from state to state and year to year. Ultimately, the children who were made newly eligible by the expansions included all those in
families with incomes between the existing 1989 AFDC thresholds and
133% of the FPL as well as some children in families with incomes up
to 185% of the FPL.
To capture this entire targeted group, we identified as "poor" the
families in the NHIS sample that had incomes of <200% of the FPL
using the NHIS indicator for poverty (income as a percentage of the
FPL). For comparison with the targeted group, we identified nonpoor
families that had incomes between 300% and 400% of the FPL. These
families were not eligible for Medicaid either before or after the
expansions. We eliminated families that had incomes between 200% and
300% of the FPL to avoid misclassification between targeted and
comparison groups resulting from misreporting of income. Families with
incomes above 400% of the FPL were excluded from the analysis to
maintain as much comparability as possible between the targeted
populations and the comparison group. We imputed family income for
children when this variable was missing (n = 1598) and
assigned them to the poor or nonpoor category. The combined number of
children in the poor and nonpoor categories for the 2 years of the
study was 27 246.
We used the NHIS definitions to categorize children by race/ethnicity
into 3 groups: non-Hispanic white, non-Hispanic black, and Hispanic.
Children who fell into another race/ethnicity category were excluded
from the analysis (n = 1031). The remaining sample consisted of 26 215 children divided into 3 race/ethnicity categories (white, black, and Hispanic), 2 poverty levels (poor and nonpoor), and
2 years (1989 and 1995).
Health Insurance Coverage
We used reported coverage at the time of interview to categorize
children into 3 insurance strata: Medicaid, private insurance, and
uninsured. Families that were receiving AFDC were coded as being on
Medicaid as were children in the 1989 sample who had any type of public
assistance health insurance coverage or for whom the respondent had a
current Medicaid card and was not covered through the military. Similar
definitions were used for the 1995 sample. Children who had military
coverage or who were reported as having both private and Medicaid
coverage were classified as private. For 1500 children, the insurance
status was missing or unknown. The final sample of children aged 1 to
12 years in the 2 income categories of interest for whom race/ethnicity
and insurance status was known numbered 24 118.
Health Care Services Utilization and Health Status
The NHIS contains a variety of measures of utilization and
health status. An important distinction in utilization measures concerns the difference between visits that are patient initiated and
those that are contingent on a practitioner's
decisions.12 The former often are referred to as measures
of "realized access."10 We used the probability of a
doctor visit in the past 12 months as a measure of realized access to
care. The measures of utilization contingent on access that we used
included number of doctor visits and hospitalizations in the past 12 months. Both of these measures are in common use in previously
published cross-sectional analyses of health care service utilization
and its relationship to either race and
ethnicity9,11,12,17 or health insurance
coverage.5,10,18,19 Finally, health status measures
included a respondent-directed evaluation of the child's health status
as excellent, very good, good, fair, or poor and number of
restricted-activity days in the previous 2 weeks. In addition to
reporting changes in measures of health status, we used self-reported
health status to stratify the changes in utilization over time. The
2-week recall period was used for restricted-activity days to enhance
recall accuracy.
Statistical Model
Our objective was to measure the differential impact of the
Medicaid expansions on the insurance status, service utilization, and
health status of different race/ethnicity groups. We therefore used a
stratified before-and-after model to compare changes from 1989 to 1995 in these outcomes for whites, blacks, and Hispanics across 2 income
brackets.
Our approach began by calculating 1995 Stratification of the analysis by race/ethnicity and income explicitly
controlled for the effects of these factors on the outcome variables of
interest. The before-and-after design controlled for the influence of
unobservable characteristics of the stratified groups that were
presumed not to change over time (eg, aggregate health endowments,
distances to providers, other health-promoting or health-impairing
behaviors). Because many of these unobservable characteristics are
associated not only with the outcome variables but also with the
likelihood of obtaining Medicaid coverage, they represent potential
confounders for any cross-sectional analysis. In a before-and-after
design, however, comparisons are made not between levels of Medicaid
coverage and outcome variables but between changes in levels of
Medicaid coverage and outcome variables over time. The presumption that
unobservable characteristics such as health endowment, behaviors,
distances to providers, and so forth do not change in the aggregate
suggests that their effects on outcome variables are the same in the
earlier time period and in the later time period. They do not,
therefore, confound the analysis. This feature of the before-and-after
analysis represents a crucial design advantage over the single
cross-sectional approach in addressing the issue of selection bias.
For bivariate comparisons within income group strata, we used
Student's t tests to compare the magnitude of changes in
the outcome variables. The data were analyzed with the use of SAS version 8.0 and Stata Release 6.0 (Stata Corp., College Station, TX) to
generate means and standard errors accounting for the clustering sample
design of NHIS data.
Table 1 depicts the numbers of
observations in the 1989 and 1995 NHIS samples of 1- to 12-year-old
children by income level, race/ethnicity, and insurance status. When
correctly weighted, these observations represent more than 50 000 000
children throughout the United States in the 2 years of study.
TABLE 1
METHODS
Top
Abstract
Methods
Results
Discussion
References
1989 differences in coverage,
utilization, and health status outcomes for nonpoor children by
race/ethnicity to establish underlying trends in these variables
independent of the impact of Medicaid expansions because nonpoor
children were, by definition, ineligible for Medicaid in either year.
We next calculated similar 1995 1989 differences for poor
children. Subtracting the nonpoor trends from those of the poor
children revealed the impact of Medicaid expansions on poor children by
race/ethnicity. The assumption underlying this approach is that the
time trends observed among the nonpoor indicate the experience also
that would have occurred among the poor had there been no expansions in
Medicaid during this period. In a final step, we compared these
Medicaid effects across race/ethnicity groups to identify the
differential impact of the Medicaid expansions.
RESULTS
Top
Abstract
Methods
Results
Discussion
References
Numbers of 1 to 12-Year-Old Children in NHIS Sample by Income,
Race/Ethnicity, and Insurance Status, 1989 and 1995
Insurance Coverage
Tables 2 and 3 summarize the experience of children from poor and nonpoor white, black, and Hispanic families during the period 1989 to 1995 with respect to their health insurance coverage. Before the Medicaid expansions in 1989, minority children had higher rates of being uninsured than did their white counterparts, and Hispanics had higher rates of being uninsured than did blacks. From 1989 to 1995, significant declines in rates of being uninsured were registered for all poor children. For nonpoor children, there were no significant changes in the likelihood of being uninsured.
|
|
Evidence that the Medicaid expansions reached their intended targets also is illustrated in the results (Table 2). Large increases in Medicaid coverage are documented among poor children from all race/ethnicity groups during this period with increases reaching as high as 23 percentage points for poor Hispanics. That Medicaid rates increased more than the observed declines in uninsured rates suggests that some poor or near-poor families were losing private insurance as time went on, perhaps because of an explicit substitution of Medicaid for the more costly private insurance. The small but significant levels of Medicaid receipt among the nonpoor shown in Table 2 reflect the presence of a population of children who are eligible for Medicaid independent of their family income (medically needy) as well as some children from families with income-reporting inaccuracies.
Table 3 shows adjusted changes between 1989 and 1995 for poor children. To adjust for unmeasured trends unrelated to Medicaid expansions, we subtracted the changes between 1989 and 1995 among the nonpoor from the changes between 1989 and 1995 among the poor. The uninsured rates among poor whites controlling for underlying trends observed among the nonpoor declined by 3 percentage points. By contrast, poor blacks experienced a decline of 11 percentage points and poor Hispanics experienced a decline of 23 percentage points in their rates of being uninsured compared with the nonpoor.
Poor families from all race/ethnicity groups recorded significant increases in Medicaid coverage, controlling for trends among the nonpoor, but only a modest differential impact by race/ethnicity was discernible (Table 3). Increases in Medicaid uptake ranged from 15 percentage points among whites to 18 and 21 percentage points among black and Hispanic families, respectively. These data may represent 2 distinct trends that have coincided. On the one hand, minority children seemed to be gaining Medicaid eligibility and shifting out of uninsured status; on the other hand, children from many white families may have been gaining Medicaid coverage through substitution of private insurance as changes in the eligibility rules allowed them this option for the first time or because of loss of private coverage for other reasons.
Utilization
The experience by income and race/ethnicity with regard to changes over time in various measures of utilization is summarized in Tables 47. These measures include the likelihood of having a visit to a physician in the previous year, the number of physician visits per year (contingent on having visited the doctor), and the probability of hospitalization during the year. Tables 4 and 5 summarize data for children who were characterized as being in good, fair, or poor health, and Tables 6 and 7 contain data for children who were characterized as being in very good or excellent health.
|
|
|
|
As would be expected, for all 3 measures, children who were in good, fair, or poor health regardless of race/ethnicity, income level, or year had higher rates of utilization than did children who were in excellent or very good health. In addition, white children were reported to have more doctor visits than minority children regardless of income group, health status, or year.
During the course of the Medicaid expansions, all poor children increased the probability of visiting a physician. For poor black children, regardless of health status, the increase in this measure of 7 percentage points was statistically significant, as it was for poor Hispanic children who were in good, fair, or poor health (Table 4). Poor white children showed almost no increase in this measure of realized access from 1989 to 1995. During the same period of time, however, nonpoor children also were recording increases in the probability of visiting a physician. Once these underlying trends are accounted for (Tables 5 and 7), poor minority children who were in good, fair, or poor health experienced an increase in the likelihood of seeing a physician relative to nonpoor minority children, although these increases did not reach the level of statistical significance.
As with the probability of seeing a physician during the course of the year, all poor children recorded increases in annual numbers of visits to the doctor during the time of the Medicaid expansions. This trend reached statistical significance, however, only for Hispanic children who were in excellent or very good health (Table 6). Many nonpoor children also experienced increases in number of doctor visits during the same period (Tables 4 and 6). Once the underlying trend among nonpoor children is taken into account (Tables 5 and 7), none of the increases in annual numbers of doctor visits among poor children was found to be significant.
|
|
Similar trends can be seen in the change in hospitalization rates over time. The only group of poor children to experience a significant decrease in the probability of being hospitalized was Hispanic children who were in good, fair, or poor health. Their rate of hospitalization fell from 8% to 4% during the period of the Medicaid expansions (Table 4). Among nonpoor children, the rate of hospitalization declined 10 percentage points among blacks during the same period. Tables 5 and 7 indicate that, controlling for trends among the nonpoor, poor black children who were in good, fair, or poor health experienced an increase in their use of the hospital during the period of Medicaid expansions, unlike their white and Hispanic counterparts. These findings among black children are driven by the decrease in hospitalization rates among nonpoor members of this ethnic group rather than by large, absolute increases among the poor.
Health Status
Tables 8 and 9 present changes in the numbers of restricted-activity days that children experienced in the previous 2 weeks stratified by self-reported health status. Children who were reported to be in good, fair, or poor health experienced greater numbers of restricted-activity days regardless of income level than did children whose health was reported as being excellent or very good. In addition, poor children from each race/ethnicity group (with few exceptions) had more restricted-activity days than their nonpoor counterparts. Although the numbers of restricted-activity days generally decreased between 1989 and 1995 among children who were in good, fair, or poor health, none of the changes for any race/ethnicity group reached statistical significance. Among children who were in excellent or very good health, nonpoor Hispanic children recorded increases in the number of restricted-activity days during the study period. When trends among the nonpoor are controlled for as shown in Table 9, poor Hispanic children who were in excellent or very good health demonstrated significant decreases in the numbers of restricted-activity days experienced during the course of the Medicaid expansions. For all other race/ethnicity groups in either health status stratum, adjusted changes in this measure of health status did not reach statistical significance. Regarding self-reported health status, no significant change in percentage of children classified as being in excellent or very good health is detectable for any group during the period of the Medicaid expansions (data not shown).
| |
DISCUSSION |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
We characterized by race and ethnicity changes in insurance coverage, health care service utilization, and health status experienced by poor and nonpoor children from 1989 to 1995 during the course of expansions in the Medicaid program. We found that changes in the probability of having health insurance varied significantly by race/ethnicity, whereas changes in utilization and health status did not.
The race/ethnicity differential changes in the probability of having health insurance indicate that the Medicaid expansions not only increased the number of poor children with health insurance coverage but also decreased racial disparities with respect to coverage. Poor black and Hispanic children who had worse health insurance coverage at the beginning of the period studied achieved greater reductions in uninsured rates than did their poor white counterparts. The white/Hispanic uninsured rate differential of 23 percentage points that existed in 1989 was cut by more than half during the subsequent 7 years. The white/black uninsured rate differential was reversed: by 1995, there were fewer uninsured poor black children than poor white children.
Increases in Medicaid coverage rates exceeded the uninsured rate declines among the poor during the same period for all 3 race/ethnicity groups. New Medicaid enrollees originate either from the pool of uninsured or from those losing private insurance. Medicaid increases above uninsured rate declines suggest, therefore, that some families came into the Medicaid program between 1989 and 1995 from those with private insurance before the expansions occurred. How much of this shift from private to Medicaid enrollment represents "crowd-out" and how much represents families opting for Medicaid after losing private insurance is an important question. Recent studies investigated the magnitude of the crowd-out effect of the Medicaid expansions.20-22 One study indicated that only 15% of the shift from private to Medicaid coverage was attributable to crowd-out; the rest came from families who lost private insurance for a variety of reasons.23 Others have pointed to such reasons as the recession of the early 1990s, increases in the price of medical care, and shifts away from manufacturing jobs as explanations for the net decline in employer coverage rates independent of expansions in the Medicaid program.24
Shifts to Medicaid from private insurance would be expected to occur more frequently in populations with higher private coverage (and therefore more opportunity to either lose or substitute private insurance) at the beginning of the period of Medicaid expansions. The race/ethnicity differentials seen in our study confirm this pattern. Poor black and white families with higher initial coverage took up Medicaid at rates that were 11 and 12 percentage points higher than the rate at which their being uninsured declined. By contrast, poor Hispanic families with less initial coverage experienced Medicaid take-up rates only 4 percentage points higher than the declines that they experienced in uninsured rates. Potential explanations for this pattern beyond simply the differences in the opportunity for substitution warrant additional scrutiny.
When we considered changes in health service utilization, race/ethnicity differentials were most discernible among children who were in good, fair, or poor health. In this group, poor minority children experienced greater increases in the probability of seeing a doctor than did their poor white counterparts. The differential of 11 percentage points in the probability of whites/Hispanics seeing a doctor that existed in 1989 among these less healthy children was halved by 1995, and the differential of 8 percentage points in whites/blacks was reduced to 2 percentage points. The gains experienced by poor black and Hispanic children in the probability of seeing a doctor outstripped those of nonpoor minority children by 7 and 12 percentage points, respectively. During the same period, poor white children were falling further behind their nonpoor counterparts in this measure of realized access. Such trends are additional evidence of the differential impact that Medicaid expansions seemed to have had on the utilization patterns of minority children. Conversely, no race/ethnicity differential changes were discernible in other measures of utilization or in measures of health status.
Why did the substantial race/ethnicity differential changes in health insurance coverage induced by the Medicaid expansions of 1989 to 1995 not translate into larger, more consistent differential changes in service utilization or health status? Economic theory and empirical observation suggest that if individuals are supplied with free access to medical care, then they should use more of it. Other things being equal, this increased utilization might be expected to improve health. Because poor black and Hispanic children experienced greater increases in coverage than poor whites after the Medicaid expansions, an inability to document more consistent utilization increases and improved health status among these children is anomalous. Either race/ethnicity differential changes in utilization and health status were as limited as they seem, or the full magnitude of the actual changes was obscured by elements of the study's design or the available data.
Design elements that may have had an impact on the ability to detect actual differences include the decision to stratify the analysis, the decision to analyze the outcomes on the basis of income class, and the choice of outcome variables to consider.
The number of observations in the NHIS data from the targeted and comparison populations for the 2 years of this study totaled 24 118. For many purposes, this number was sufficient to generate precise estimates of outcome measures of interest. Stratification of the sample by race/ethnicity, income, year, and health status, however, generated smaller cells with less power to detect actual differences that might exist. The relatively wide confidence intervals illustrate this difficulty (Table 5). Despite this limitation in the precision of some of the estimates, however, the NHIS data set remains the most comprehensive representative sample of the noninstitutionalized US population with which to evaluate the effects of the Medicaid expansions by race/ethnicity.
A second decision
to use income class as the level of analysisalso
had important implications for how the outcomes would be viewed.
Cross-sectional analysis of Medicaid's effect on utilization and
health status cannot distinguish between the respective effects of
health insurance coverage and those of unobservable characteristics of
individuals. To the extent that families with less healthy children are
more likely both to obtain insurance coverage and to use services than
are families with healthier children, cross-sectional comparisons of
utilization attribute to insurance coverage (whether public or private)
effects that derive from characteristics of the children. An important
strength of the before-and-after comparison that we conducted on the
basis of the outcomes measured at the level of the entire income class
is that it avoids this self-selection bias under the reasonable
assumption that the unobservable characteristics of the entire income
class in aggregate are invariant over time. If these characteristics do
not change for the class as a whole, then comparing the outcomes of
interest across 2 time periods will isolate the effects of the
expansions in Medicaid eligibility.
The use of income class as the level of analysis avoids selection bias, but it does so at a cost. It does not permit the analysis to show the actual effect on an individual of acquiring health insurance. Consider, as an outcome, the likelihood of seeing a physician. For any income class in a given year, this outcome will be the sum of 2 products: 1) the percentage of children with health insurance in the class multiplied by the probability that insured children will see a physician plus 2) the percentage of children without health insurance in that class multiplied by the probability that uninsured children will see a physician.
We recorded changes in the percentage of covered children and changes in the outcome. From these observations, it is possible to infer how the probability of seeing a physician varies by insurance status. If the probabilities by insurance status remain constant over time, then modest increases in aggregate coverage may result in modest increases in the aggregate probability of seeing a physician even if the probability varies widely by insurance status. Consider poor Hispanics who were in good, fair, or poor health. Their coverage increased by 19 percentage points from 1989 to 1995, and their likelihood of seeing a physician increased by 7 percentage points. This suggests that there was a difference of 37 percentage points between the likelihood of seeing a physician among the insured compared with the uninsured [(0.19)*(0.37) = 0.07]. Our findings that Medicaid expansions resulted in relatively modest increases in utilization at the level of the income class should be viewed in this context. The aggregate effects at the class level do not fully capture the impact that gaining insurance coverage had on any given individual.
A second limitation of the NHIS data for the years that we compared concerns the available measures of utilization and health status. The probability of visiting a physician, the number of physician visits contingent on any visits, and the likelihood of hospitalization all are useful measures of utilization. Additional comparisons of changes in access measures, in the inability to obtain needed care, in the type of practitioner, or in the site of visit over time would have enriched this aspect of the analysis. Other studies have documented that children who receive publicly financed health insurance use less ambulatory care, dental, vision, and prescription services when compared with children who are privately insured.25 Moreover, those who are covered by Medicaid are half as likely to report having a regular source of care.26 Unfortunately, the NHIS data do not contain these elements for either 1989 or 1995, so these comparisons were unavailable to us. In addition, although self-reported health status and restricted-activity days are widely accepted measures of health status, they do not capture the full range of this elusive metric.
Beyond the methodological considerations outlined above lies an equally
important substantive issue. It can be argued that the lack of findings
with respect to increased utilization and improved health status after
the Medicaid expansions is not simply an inability to discern changes
that actually occurred but accurately reflects the modest changes
induced by this policy shift. Under this interpretation, most children
were receiving needed care even before the expansions occurred. What
improved Medicaid eligibility accomplished was to provide a financing
mechanism to allow a greater percentage of these families to have the
government pay for care that otherwise would have gone unreimbursed. If
this were the case, then it would not be surprising to find little
change in health status and only modest changes in utilization.
Although increased coverage may be a necessary condition to improve
utilization and health status, it may not be sufficient. Future changes
in supply-side issueswho provides Medicaid patients with services, in
what settings, and with what quality of caremay be at least as
necessary to change utilization patterns and health status as improving
coverage.
The Medicaid expansions in the late 1980s and early 1990s had a beneficial impact on minority children by increasing disproportionately their access to health insurance coverage. Additional studies will help to determine what future policy initiatives ought to be considered to build on these achievements.
| |
FOOTNOTES |
|---|
Received for publication Dec 18, 2000; accepted Mar 26, 2001.
Reprint requests to (A.D.R.) Albert Einstein College of Medicine, Children's Hospital at Montefiore and National Bureau of Economic Research, 1621 Eastchester Rd, Bronx, NY 10461. E-mail: aracine{at}montefiore.org
| |
ABBREVIATIONS |
|---|
FPL, federal poverty level; NHIS, National Health Interview Surveys; AFDC, Aid to Families With Dependent Children.
| |
REFERENCES |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
- US Dept of Health and Human Services. Health Care Financing Review: Medicare and Medicaid Statistical Supplement, 1997. Baltimore, MD: US Department of Health and Human Services; 1997:191. Publ. No. 03399
-
St. Peter RF,
Newacheck PW,
Halfon N
Access to care for poor children:
separate and unequal?
JAMA
1992;
267:2760-2764
[Abstract/Free Full Text] - Freeman HE, Corey CR Insurance status and access to health services among poor persons. Health Serv Res 1993; 28:531-541 [Medline]
- Currie J, Thomas D Medical care for children: public insurance, private insurance and racial differences in utilization. J Hum Res 1995; 30:135-162
- Marquis MS, Long SH Reconsidering the effect of Medicaid on health care services use. Health Serv Res 1996; 30:791-808 [Medline]
-
Keane CR,
Lave JR,
Ricci EM,
LaVallee CP
The impact of a children's
health insurance program by age.
Pediatrics
1999;
104:1051-1058
[Abstract/Free Full Text] -
Newacheck PW,
Stoddard JJ,
Hughes DC,
Pearl M
Health insurance and
access to primary care for children.
N Engl J Med
1998;
338:513-519
[Abstract/Free Full Text] -
Stoddard JJ,
St. Peter RF,
Newacheck PW
Health insurance status and
ambulatory care for children.
N Engl J Med
1994;
330:1421-1425
[Abstract/Free Full Text] -
Newacheck PW,
Hughes DC,
Stoddard JJ
Children's access to
primary care: differences by race, income and insurance status.
Pediatrics
1996;
97:26-32
[Abstract/Free Full Text] - Newacheck PW, Stoddard JJ, Hughes DC, Pearl M. Children's access to health care: the role of social and economic factors. In: Stein REK, ed. Health Care for Children. New York, NY: United Hospital Fund of New York; 1997:53-76
-
Lieu TA,
Newacheck PW,
McManus MA
Race, ethnicity, and access to
ambulatory care among US adolescents.
Am J Public
Health
1993;
83:960-965
[Abstract/Free Full Text] - Guendelman S, Schwalbe J Medical care utilization by Hispanic children: how does it differ from black and white peers? Med Care 1986; 24:925-940 [CrossRef][Medline]
- Weinick RM, Weigers ME, Cohen JW Children's health insurance, access to care, and health status: new findings. Health Aff 1998; 17:127-136 [Abstract]
- National Center for Health Statistics (Producer). Public Use Data Tape Documentation: Part I. National Health Interview Survey, 1989 and 1995. Springfield, VA: National Technical Information Service, US Department of Commerce; 1994, 1998
- US Department of Health and Human Services. Health Care Financing Review: Medicare and Medicaid Statistical Supplement, 1977. Baltimore, MD: US Department of Health and Human Services; 1977. Publ. No. 03399
- US General Accounting Office. Health Insurance for Children: Private Insurance Coverage Continues to Deteriorate. Washington, DC: US General Accounting Office; 1996 Health Education and Human Services Division. Publ. No. GAO/HEHS-96-129
-
Weitzman M,
Byrd RS,
Auinger P
Black and white middle class children
who have private health insurance in the United States.
Pediatrics
1999;
104:151-156
[Abstract/Free Full Text] - Currie J, Gruber J Health Insurance eligibility, utilization of medical care, and child health. Q J Econ 1996; 111:431-466 [CrossRef]
-
Rosenbach ML,
Irvin C,
Coulam RF
Access for low-income children: is
health insurance enough?
Pediatrics
1999;
103:1167-1174
[Abstract/Free Full Text] - Cutler DM, Gruber J Does public insurance crowd out private insurance? Q J Econ. 1996; 111:391-429 [CrossRef]
- Dubay L, Kenney G Did Medicaid expansions for pregnant women crowd out private coverage? Health Aff 1997; 16:185-193 [CrossRef][Medline]
-
Shenkman E,
Bucciarelli R,
Wegener DH,
Naff R,
Freedman S
Crowd out:
evidence from the Florida Healthy Kids Program.
Pediatrics
1999;
104:507-513
[Abstract/Free Full Text] - Yazici E, Kaestner R Medicaid expansions and the crowding out of health insurance. Inquiry 2000; 37:23-32 [Medline]
- Newacheck PW, Hughes DC, Cisternas M Children and health insurance: an overview of recent trends. Health Aff 1995; 14:244-254 [Abstract]
- Lefkowitz DC, Monheit AC. Health Insurance, Use of Health Services and Health Care Expenditures. National Medical Expenditure Survey Research Findings 12. Rockville, MD: AHCPR; 1991. Publ. No. 92-0017
-
Wood DL,
Hayward RA,
Corey CR,
Freeman HE,
Shapiro MF
Access to
medical care for children and adolescents in the United States.
Pediatrics
1990;
86:666-673
[Abstract/Free Full Text]
Pediatrics (ISSN 0031 4005). Copyright ©2001 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  K. G. Duderstadt, D. C. Hughes, M.-J Soobader, and P. W. Newacheck The Impact of Public Insurance Expansions on Children's Access and Use of Care Pediatrics, October 1, 2006; 118(4): 1676 - 1682. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. P. Greenwald, S. O'Keefe, and M. DiCamillo Why Employed Latinos Lack Health Insurance: A Study in California Hispanic Journal of Behavioral Sciences, November 1, 2005; 27(4): 517 - 532. [Abstract] [PDF]
|
|
|
|
 |
|
 S. Pati, R. Keren, E. A. Alessandrini, and D. F. Schwarz Generational Differences In U.S. Public Spending, 1980-2000 Health Aff., September 1, 2004; 23(5): 131 - 141. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Hadley Sicker and Poorer--The Consequences of Being Uninsured: A Review of the Research on the Relationship between Health Insurance, Medical Care Use, Health, Work, and Income Med Care Res Rev, June 1, 2003; 60(2_suppl): 3S - 75S. [Abstract] [PDF]
|
|
|
|
|
|
T. A. Lieu, P. Lozano, J. A. Finkelstein, F. W. Chi, N. G. Jensvold, A. M. Capra, C. P. Quesenberry, J. V. Selby, and H. J. Farber Racial/Ethnic Variation in Asthma Status and Management Practices Among Children in Managed Medicaid Pediatrics, May 1, 2002; 109(5): 857 - 865. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
