Access and satisfaction among children in Georgia's Medicaid Program and SCHIP: 2000 to 2003.

The lagging reauthorization of the State Children's Health Insurance Program (SCHIP) affects States' plans for sustaining mature programs. This study used the Consumer Assessment of Healthcare Providers and Systems (CAHPS) program survey in order to assess changes in access and satisfaction for Georgia's SCHIP (PeachCare) and Medicaid children as PeachCare matured 2000 to 2003. Adjusting for family and child characteristics, PeachCare enrollees reported better access and higher satisfaction than Medicaid clientele initially, but access differences narrowed by 2003 while differences in satisfaction grew. This may point to cultural/language issues or treatment stigma for Medicaid clientele. Nonetheless, overall plan ratings remained high for both groups.

Introduction

SCHIP is a Federal-State partnership begun in 1997 that allowed States to create and finance health insurance programs for low-income children not already covered in their State Medicaid Programs. Currently up for reauthorization, SCHIP provides participating States with Federal funds through a matching grant up to a cap. Under SCHIP, States were given the flexibility to use funds for a Medicaid expansion program, to set up a separate program (SCHIP) or use a combination of both. States split evenly (15 States in each) between Medicaid expansions and separate programs while some 20 States used funds to implement combination programs (Rosenbaum, Markus, and Sonosky, 2004).

In Georgia, PeachCare began enrolling in January 1999 as a separate program. PeachCare followed the Georgia Medicaid Program in terms of provider reimbursement and network, use of a primary care case management (PCCM) delivery system and virtually all service coverage provisions (targeted case management and nonemergency transportation were omitted). Earlier work based on CAHPS® showed that in 2000, Georgia Medicaid children used fewer services and expressed lower satisfaction with services compared to PeachCare children (Edwards, Bronstein, and Rein, 2002). These authors thought such differences reflected less familiarity/comfort with PCCM, more non-program barriers to care, and/or different views of Medicaid versus PeachCare children held by physicians.

As PeachCare and other SCHIP's have matured, it is important to assess their performance and potential interaction with the traditional Medicaid Program. This is particularly true given the renewed growth in Medicaid enrollments from the economic downturn in the early 2000s, and the possible expansion of SCHIP under current congressional proposals. We use data from two CAHPS® surveys of Georgia Medicaid and PeachCare children (2000 and 2003) to address the following:

How have differences in reported access and perceptions of satisfaction between Medicaid and PeachCare changed as the PeachCare program matured?

Do differences in either year remain after adjusting for demographics, location, health status, and selected county characteristics?

Background

Georgia's PeachCare program grew from an early enrollment of 60,000 in January 2000 to over 280,000 in 2003 (Centers for Medicare & Medicaid Services, 2001, 2003). This more recent figure represents high penetration, approximately 71 percent, of the eligible uninsured children in the State (author tabulations of the 2002-2004 annual demographic supplement to the March CPS). Growth in PeachCare enrollment was fueled by an aggressive community outreach program accompanied by an increase in the State's income eligibility in 2001 from 200 to 235 percent of the Federal poverty level.

While Medicaid enrollment slowed during the latter 1990s it began growing again with the last economic downturn; total enrollment in Georgia grew at annual rates of 9.3 to 12.9 percent between June 2000 and June 2003 (Ellis, Smith, and Rousseau, 2005). At least part of this increase in Medicaid enrollment was a spillover effect from PeachCare outreach. That is, a significant portion of children applying for PeachCare were identified and enrolled as Medicaid-eligible children. Indeed, a parallel study to this one found 16 percent of Medicaid children continuously enrolled during the first 6 months of 2003 were so identified (Ketsche et al., 2007).

The age and family income of children served by each program vary due in large part to differences in the eligibility criteria; basic parameters of 2003 eligibility in Georgia are shown in Table 1.

Table 1

Georgia's PeachCare and Medicaid Program Characteristics: 2003

Characteristic	Infants		Age 1-5	Age 6-18
	Enrolled Under Pregnancy Expansion	Otherwise Enrolled
Medicaid Income Eligibility as Percent of FPL	<235	<185	<133	<100
PeachCare Income Eligibility as Percent of FPL	—	185-235	133-235	100-235

NOTES: FPL is Federal poverty level. As of July 2004, an infant born into Medicaid under the pregnancy expansions would be eligible up to 200 percent FPL.

SOURCE: Department of Community Health, Atlanta, Georgia, 2008.

The average PeachCare enrollee will be older and from a higher income bracket than the average Medicaid enrollee and, compared to children eligible under welfare-related (income support) criteria, more likely to be in a two-parent family. A key strength of the (CAHPS®) data is the inclusion of such information on family structure as well as health status and other sociodemographics known to affect access and utilization.

Earlier Literature

Given the role that physicians' participation in Medicaid plays in affecting enrollee access (Gruber, Adams, and Newhouse, 1997; Adams, 1994) it is important to understand the nature and capacity of the State's delivery system. The implementation of PCCM between 1994 and 1998 in Georgia was associated with reductions in physician participation and the use of primary, preventive, and especially, emergency room (ER) care among Medicaid children (Adams, Bronstein, and Florence, 2003; Bronstein, Adams, and Florence, 2005). Moreover, in the initial implementation period, increased PeachCare enrollment was associated with declines in the volume of care provided to Medicaid children by participating physicians in urban areas (Bronstein, Adams, and Florence, 2004). Participation of sufficient numbers of Medicaid providers to meet the continued growth in demand from PeachCare enrollees may have been restrained by a lack of growth in physician provider reimbursement levels; Georgia's Medicaid payments relative to Medicare stood at 0.81 in 1998 as well as 2003 (Zuckerman et al., 2004). In this study we are able to examine a broad set of access indicators from the CAHPS® survey for a time period after the PCCM program matured and during the 3 years in which PeachCare came to maturity.

In the broader literature, there is evidence that SCHIP enrolled children are not dramatically different from their Medicaid counterparts in terms of their substantial and wide-ranging health care needs (Brach et al., 2003), but there is less understanding of differences in actual utilization. One study that compared SCHIP and Medicaid-eligible children found SCHIP-eligible children to have more unmet need after enrollment (Feinberg et al., 2002), but adjustment for predisposing and enabling factors eliminated these differences; age and ethnicity remained significant determinants.

An earlier study that compared health care use between SCHIP and Medicaid-covered children in Georgia based on the CAHPS® survey found that Medicaid children were less likely to report very good or excellent health than children in PeachCare (Edwards, Bronstein, and Rein, 2002). Adjusting for special health care needs/demographics, this study found Medicaid children more likely than PeachCare children to report not having a personal physician, not having seen a physician in the last 6 months and having a problem getting necessary care. A study that used a group cohort of PeachCare enrolled, but Medicaid-eligible children as a control cohort found that part of the difference in satisfaction among PeachCare and Medicaid children related to differential stigma associated with the Medicaid Program relative to PeachCare (Ketsche et al., 2007).

Recent work compared PeachCare children with and without prior Medicaid enrollment to Georgia's near-poor children with family income above welfare levels, but below Medicaid thresholds (refer to Table 1) for their age (Bronstein, Adams, and Florence, 2006). Overall, PeachCare children were less likely to use well-child and non-urgent ER services than Medicaid children. PeachCare children with no prior Medicaid enrollment were no more/less likely to use primary care while those with prior Medicaid enrollment were more likely to receive primary care than Medicaid children. The latter result may indicate some self selection of sicker children among those who age out of Medicaid into PeachCare eligibility.

Given the evidence in the literature of an association between income and access/satisfaction and that PeachCare children are from higher income and more often, two-parent families, we hypothesized that:

PeachCare children/families are more likely than Medicaid children/families to report better access to services.

PeachCare children/families are more likely than Medicaid children/families to report satisfaction with their program.

Further, given the lack of growth in relative Medicaid reimbursements in Georgia during the time period studied, the rapid growth in PeachCare enrollment between 2000 and 2003, and the initial crowd out of Medicaid enrollees in Georgia's urban areas we also hypothesized:

Differences in access between PeachCare and Medicaid children widened during the study period.

Testing these hypotheses using self-reported access and satisfaction will provide valuable insight for a State that historically reached a large proportion of eligible SCHIP children and yet, faces budget uncertainties and severe financial pressures given the lack of reauthorization.

Data and Methods

The CAHPS® survey was initially developed in 1997 with the support of the Agency for Healthcare Research and Quality and has been widely used across populations, settings, and health plans. These data assess perceived satisfaction among parents of enrolled children based on experience in accessing primary, specialty, emergency, and dental care. In April 2000, the Georgia Health Policy Center (GHPC) provided a report to the Department of Community Health (DCH) on the satisfaction of parents and guardians of children drawn from the CAHPS® survey. On behalf of DCH, the GHPC fielded a second round of surveys in late 2003 and early 2004 to re-evaluate parental/guardian satisfaction. In this article we used the results from April 2000 and 2003 CAHPS® survey to compare measures between SCHIP and Medicaid recipients over those years.

The potential respondents for the 2003 survey were selected from children continuously enrolled in either program for 6 or more months prior to May 2003. Random samples of about 5,000 were drawn; from these original samples, a mail-back survey form was sent to parents of 3,000 children in each program. Non-respondents and parents of other children from the original samples were contacted by telephone to reach the target number of 1,000 completed surveys for each group. A nationally recognized survey firm specializing in health care related research administered both mail and telephone surveys. Table 2 describes the response rate for each of the two groups.

Table 2

CAHPS® Survey Response Rates, by Medicaid and PeachCare Programs: 2003

Response Rate	PeachCare	Medicaid
Total Attempts	3,504	3,370
Total	1,003	1,086
Mail	706	608
Telephone	297	478
Total Response Rate	28.60	32.20

SOURCE: Department of Community Health, Atlanta, Georgia, 2003.

While the previously mentioned response rate is lower than that reported by Edwards, Bronstein, and Rein (2002), this reflects a change in methodology by the survey firm that was designed to facilitate rapid completion of the requisite number of surveys in 2003. For example, repeat contacts to increase response rates were limited to a single mailing and at most, two telephone attempts in 2003, while the previous survey process included at least two mailings and up to seven telephone attempts for completion. The relative newness of PeachCare and high degree of participant enthusiasm in 2000 may also have contributed to the higher response rate.

We note that the response rates for this survey were generally consistent across enrollee groups, suggesting that satisfaction differences between the groups of respondents were unlikely to be attributable to non-response bias. We also controlled for survey response mode in our analyses. When we compared the survey respondents with the continuously enrolled populations in each program we found small, but statistically significant (p = 0.10 or below) differences between the survey respondents and all enrollees based on age and location. Differences in race/ethnicity of respondents compared to enrollees were more substantial, particularly among Medicaid respondents. We controlled for all demographic variables in the analyses that follow using survey reported age, race, and ethnicity variables.

Given that there was a different mix of telephone and mail surveys in 2000 and 2003 we ran separate cross-sectional regressions. We used primarily the same set of dependent variables examined in the earlier year of (CAHPS®) data and a set of independent variables available in both years to enable comparisons. We did test three additional independent variables reflective of access or service need within the county: (1) number of pediatricians per 1,000 children, (2) presence of a federally qualified health center (FQHC) or rural health center, and (3) log of per capita income in both the 2000 and 2003 regressions.

Results

Table 3 shows key differences between PeachCare and Medicaid children in 2000 and 2003 based on the survey data.

Table 3

Demographics of Georgia's Respondents, by Program: 2000 and 2003

Demographic	2000		2003
	PeachCare	Medicaid	PeachCare	Medicaid
	Percent
Child's Health Status1
Excellent	34	29	51	41
Very Good	33	30	34	32
Good	29	29	13	22
Fair	3	10	2	5
Poor	1	1	0	1
Relationship—Survey Respondent1
Mother or Father	98	85	97	85
Grandparent	1	10	2	10
Aunt or Uncle	0	3	0	2
Other Relative or Guardian	1	2	1	3
Child's Location2
Atlanta	31	28	41	34
Rural Georgia	54	50	45	49
Other Metro	16	22	14	16
Child's Race1
White	67	31	64	39
Black	28	62	27	50
Hispanic	4	5	5	10
Other	1	2	4	1

Difference between Medicaid and PeachCare is significant at the p<= 0.01 level in both 2000 and 2003.

Difference between Medicaid and PeachCare is significant at the p<=0.05 level in 2000 and at the p<=0.01 level in 2003.

SOURCE: Georgia's Department of Community Health: Data from the 2000 and 2003 CAHPS® survey administered by Pegus, Inc.

The descriptive data on PeachCare and Medicaid children in general, showed PeachCare children were healthier, more likely to be living with their parents, far less likely to be minorities, and more likely to be in urban areas. While these patterns held in both years, there were notable changes. First, both groups of children appeared healthier; the percentage that replied they were in excellent health increased in 2003 by over 10 percentage points in both programs. This was likely the result of aggressive outreach/enrollment of children who would not have otherwise been enrolled. Although these changes should have affected overall access and satisfaction measures, they would not necessarily have affected the comparative experience of the two groups of children.

The growth in enrollment in both programs and the changing demographics in the State also appears to have brought in more Hispanics, especially within the Medicaid Program. Whereas 62 percent of Medicaid children in 2000 were Black, this dropped to 50 percent in 2003, while the percent reporting Hispanic ethnicity grew from 5 to 10 percent. There was also a marked increase in the percentage of children living in the Atlanta urban area, especially among PeachCare children, offset by a decline in the share from other urban or rural areas. These changes may reflect, in part, the increase in the income eligibility to 235 percent in 2001.

Before moving to multivariate analysis, we examined unadjusted differences in the access and satisfaction measures for children enrolled in PeachCare versus Medicaid in 2000 and 2003 (Tables 4 and 5).

Table 4

Comparison of Georgia's Responses Regarding Access, by PeachCare and Medicaid Program: 2000 and 2003

Selected Access Measure	2000		2003
	PeachCare	Medicaid	PeachCare	Medicaid
	Percent
Do you have one person you think of as your child's personal doctor or nurse? (Answer: No.)	19***	31	20*	23
When you called during regular office hours, how often did you get the help or advice you needed for your child? (Answer: Sometimes or Never.)1	7*	10	6***	12
How often did your child get an appointment for regular or routine health care as soon as you wanted? (Answer: Sometimes or Never.) 2	9***	16	11	13
In past 6 months, how many times (not counting times your child went to an emergency room) did your child go to a doctor's office or clinic? (Answer: None.)	13***	26	23	21
In past 6 months, how much of a problem did you have, if any, in getting a referral for a specialist that your child needed to see? (Answer: A big problem.) 2, 3	4***	15	4**	10
In past 6 months, did your child see a specialist? (Answer: No.) 2	79***	86	80	82
When your child needed care right away for an illness or injury, how often did your child get care as soon as you wanted? (Answer: Sometimes or Never.) 1	9	13	7	9
In past 6 months, how many times did your child go to an emergency room? (Answer: None.)	78	79	79	77
In past 6 months, how much of a problem did you have, if any, to get care for your child that you or a doctor believed necessary? (Answer: A big problem.) 3	2	2	1	2
(Answer: A small problem.)3	3***	10	5**	8

Difference is significant at the p<=0.1 level.

Difference is significant at the p<=0.05 level.

Difference is significant at the p<=0.01 level.

Asked only if respondents had said that they had called for advice, made an appointment, or had an illness or injury that needed care right away. Response categories were always, usually, sometimes, never.

Asked only if respondent has said that they or their doctor thought their child needed to see a specialist.

Response categories were big problem, small problem, not a problem.

SOURCE: Georgia's Department of Community Health: Data from the 2000 and 2003 CAHPS® survey administered by Pegus, Inc.

Table 5

Comparison of Georgia's Responses Regarding Satisfaction, by PeachCare and Medicaid Program: 2000 and 2003

Selected Satisfaction Measure	2000		2003
	PeachCare	Medicaid	PeachCare	Medicaid
	Percent
With choices (the program) gave you, how much of a problem was it to get a personal doctor or nurse that you were happy with? (Answer: A big problem.) 1	9*	4	5	4
Rated child's personal doctor or nurse 6 or less on a 10-point scale?	9*	13	6*	8
How often did your child wait more than 15 minutes past the appointment time to see physician? (Answer: Usually or Always.) 2	32	33	33	31
How often were office staff as helpful as you thought they should be? (Answer: Sometimes or Never.) 2	8*	12	6***	10
In past 6 months, how often did doctors or other providers spend enough time with you and your child? (Answer: Sometimes or Never.) 2	10***	17	7***	14
Rated specialist 6 or less on a 10-point scale? 3	13	16	14	12
Rate primary care provider 8 or less (scale 0 to 10).	33**	39	32	31
Rate overall health care 8 or less (scale 0 to 10).	32**	39	26***	34
Rate overall health plan 8 or less (scale 0 to 10).	29	29	16***	26

Difference is significant at the p<=0.1 level.

Difference is significant at the p<=0.05 level.

Difference is significant at the p<=0.01 level.

Response categories were big problem, small problem, not a problem.

Response categories were always, usually, sometimes, never.

Asked only if respondent saw a specialist.

SOURCE: Georgia's Department of Community Health: Data from the 2000 and 2003 CAHPS® survey administered by Pegus, Inc.

There were two key patterns. First, there were generally advantages in access reported by PeachCare versus Medicaid children in both years and second, the gap in these advantages narrowed markedly over the study period. For example, almost one-third (31 percent) of Medicaid children, but only 19 percent of PeachCare children reported not having a personal doctor or nurse in 2000. By 2003, however, there was no significant difference between PeachCare (20 percent) and Medicaid (23 percent) children on this measure. PeachCare children reported better access than Medicaid children in 2000 for seven of the nine measures reported in Table 4 but by 2003, for only four of these measures. A key change was the decline from 26 to 21 percent of Medicaid families who said their child did not have a non-ER visit in the past 6 months; this measure increased from 13 to 23 percent for PeachCare families. The only measure for which the gap grew was the percent of parents responding that they sometimes/never got the help they needed when they called for help or advice. In 2000, there was only a 3-percentage point gap in this access problem by program, whereas by 2003, 12 percent of Medicaid families versus only 6 percent of PeachCare families experienced this problem.

Access problems may or may not translate directly into reported dissatisfaction with the program and health care received. In Table 5 we report percentage response on selected measures of satisfaction for families in PeachCare versus Medicaid in 2000 and 2003.

As these data show, there were generally high levels of satisfaction in both programs but as with access, satisfaction measures were generally higher for PeachCare enrollees. For summary measures—rating primary care providers or health care at 8 or less on a 10-point scale—PeachCare families reported higher satisfaction than Medicaid families in 2000. Only 33 percent of PeachCare families (versus 39 percent of Medicaid families) ranked their child's provider at less than 8 in 2000, but this difference in satisfaction eroded by 2003. As in 2000, Medicaid enrollees were more likely to express dissatisfaction with their overall health care than PeachCare enrollees in the more current year. While both groups were equally likely to rate overall satisfaction with the plan at 8 or lower in the 2000 survey, there was a substantial increase in satisfaction with the plan among PeachCare enrollees that did not occur among Medicaid enrollees, creating a gap in overall satisfaction in 2003. This appears related to a lower ranking among Medicaid families of their personal doctor/nurse, dissatisfaction with helpfulness of office staff, and time providers spent with them and their child.

Multivariate Results

While the descriptive data indicate significant improvement in terms of access for Medicaid versus PeachCare enrollees as noted, the composition of enrollees changed. To more clearly understand how families assessed these programs we ran multivariate logistic regressions on each of the measures presented in Tables 5 and 6. In each regression (full results available on request from the author) we included measures of: (1) location (rural, non-Atlanta metropolitan statistician area versus Atlanta); (2) age plus age squared; (3) health status (good, very good, or excellent versus fair or poor); (4) sex; (5) race/ethnicity; (6) English spoken by parent/guardian; (7) education level of parent/guardian (less than high school, some college, or more versus high school only); (8) mail versus telephone response; (9) number of pediatricians per 1,000 children in county; (10) presence of an FQHC or rural health center in county; and (11) log of per capita income in county. Relative odds of PeachCare versus Medicaid children for access measures, confidence intervals around the odds, and statistical significance are presented in Table 6.

Table 6

Comparison of Adjusted Odds Ratios for Georgia's PeachCare Versus Medicaid Program on Selected Access Measures: 2000 and 2003

Odds Ratio for Programmatic Effects (PeachCare versus Medicaid) on Selected Access Measures	Odds Ratio PeachCare Relative to Medicaid
	2000	2003
Do you have one person you think of as your child's personal doctor or nurse? (Answer: No.)	0.996(0.700-1.334)	0.916(0.704-1.192)
When you called during regular office hours, how often did you get the help or advice you needed for your child? (Answer: Sometimes or Never.) 2	0.499**(0.271-0.918)	0.627*(0.378-1.038)
How often did your child get an appointment for regular or routine health care as soon as you wanted? (Answer: Sometimes or Never.) 2	0.576**(0.346-0.961)	1.083(0.706-1.661)
In past 6 months, how many times (not counting times your child went to an emergency room) did your child go to a doctor's office or clinic? (Answer: None.)	0.522***(0.367-0.743)	1.217(0.944-1.568)
In past 6 months, how much of a problem did you have, if any, in getting a referral for to a specialist that your child needed to see? (Answer: A big problem.) 3	0.322*(0.102-1.011)	0.461*(0.197-1.075)
In past 6 months, did your child see a specialist? (Answer: No.)	0.362***(0.219-0.601)	1.047(0.807-1.358)
When your child needed care right away for an illness or injury, how often did your child get care as soon as you wanted? (Answer: Sometimes or Never.) 2	1.02(0.524-1.986)	0.928(0.484-1.779)
In past 6 months, how many times did your child go to an emergency room? (Answer: None.)	0.832(0.607-1.142)	0.912(0.713-1.165)
In past 6 months, how much of a problem did you have, if any, to get care for your child that you or a doctor believed necessary? (Answer: A big problem.) 3	0.694(0.235-2.053)	0.772(0.307-1.942)
(Answer: A small problem.) 3	0.325***(0.165-0.641)	0.798(0.491-1.298)

Significant at the p<=0.1 level.

Significant at the p<=0.05 level.

Significant at the p<=0.01 level.

Adjusted for: location, age, health status, sex, race/ethnicity, English spoken at home, parent/guardian education, mail or telephone response, pediatricians per 1,000 children in county, whether there is a federally qualified health center or rural health center in the county, and log of per capita income in the county.

Response categories were always, usually, sometimes, never.

Response categories were big problem, small problem, not a problem.

SOURCE: Georgia's Department of Community Health: Data from the 2000 and 2003 CAHPS® survey administered by Pegus, Inc.

In the multivariate analyses the number of measures for which PeachCare families reported greater access than Medicaid families was smaller (six) than in the descriptive data and only two measures were significantly different by 2003. Specifically, in 2000 PeachCare families were significantly less likely to report issues with getting the help/advice needed for their child (odds = 0.499) or to report a problem getting a referral to a specialist (odds = 0.322) or a (small) problem getting necessary care (odds = 0.325). Moreover, they were about one-half as likely (odds = 0.522) as Medicaid families to report that their child did not have a non-ER visit with a doctor or clinic during the past 6 months. By 2003, some of these differences disappeared. In particular, PeachCare children were no longer less likely to report that their child did not go to a non-ER doctor or clinic within the past 6 months.

The descriptive data also indicated that the advantages in access for PeachCare families were reflected in higher satisfaction. Differences in demographics can affect satisfaction as not only actual access, but also expectations regarding access and provider treatment are often shaped by these factors and past experience. Table 7 presents the adjusted odds ratios using the same controls as before.

Table 7

Comparison of Adjusted Odds Ratios for Georgia's PeachCare Versus Medicaid Program on Selected Satisfaction Measures: 2000 and 2003

Odds Ratio for Programmatic Effects on Selected Satisfaction Measures	Odds Ratio PeachCare Relative to Medicaid
	2000	2003
With choices (the program) gave you, how much of a problem was it to get a personal doctor or nurse that you were happy with? (Answer: A big problem.) 2	1.166(0.469-2.902)	1.353(0.532-3.441)
How often did your child wait more than 15 minutes past the appointment time to see physician? (Answer: Usually or Always.) 3	1.082(0.797-1.469)	1.065(0.827-1.373)
How often were office staff as helpful as you thought they should be? (Answer: Sometimes or Never.) 3	0.705(0.438-1.135)	0.621**(0.395-0.976)
In past 6 months, how often did doctors or other providers spend enough time with you and your child? (Answer: Sometimes or Never.) 3	1.002(0.611-1.642)	0.553***(0.367-0.833)
Rated specialist 6 or less on a 10-point scale? 4	0.887(0.389-2.022)	1.435(0.729-2.822)
Rate primary care provider 8 or less (scale 0 to 10).	0.784(0.577-1.066)	0.888(0.693-1.138)
Rate overall health care 8 or less (scale 0 to 10).	0.730**(0.543-0.982)	0.592***(0.459-0.764)
Rate overall health plan 8 or less (scale 0 to 10).	0.649*(0.407-1.035)	0.461***(0.357-0.596)

Significant at the p<=0.1 level.

Significant at the p<=0.05 level.

Significant at the p<=0.01 level.

Adjusted for: location, age, health status, sex, race/ethnicity, English spoken at home, parent/guardian education, mail or phone response, pediatricians per 1,000 children in county, whether there is a federally qualified health center or rural health center in the county, and log of per capita income in the county.

Response categories were big problem, small problem, not a problem.

Response categories were always, usually, sometimes, never.

Asked only if respondent saw a specialist.

SOURCE: Georgia's Department of Community Health: Data from the 2000 and 2003 CAHPS® survey administered by Pegus, Inc.

Even after controlling for all other factors, PeachCare families were less likely to rank their overall care and plan at less than 8 in both 2000 and 2003. After controls, other differences between PeachCare and Medicaid seen in the 2000 descriptive data disappear. However, even after controlling for other factors in 2003, the PeachCare families were less likely (odds=0.621) to report the office staff as not helpful and less likely (odds=0.553) to report problems with time spent by providers.

Decomposition

It is clear from the data in Table 3 that the characteristics of the two enrollee groups changed 2000 to 2003 (more healthy, and more White and Hispanic). To better understand whether changes in characteristics versus changes in the provider system/other factors explained the narrowing of the gap between the two groups, we tried to tease apart the two effects. To do this, we first used the 2000 coefficients from the regression models and the 2003 enrollee characteristics to model what access would have been for each enrollee group as if the prior programmatic effects remained. We compared these to predicted values using the 2000 coefficients and enrollee characteristics. Changes in the gap from these two sets of predicated values reflect changes explained by changing enrollee characteristics. These changes are then compared to the changes in actual gaps (Table 4) which reflect the combined effects of changing enrollee characteristics and programmatic effects. If the modeled changes were similar to the actual changes we concluded that changes in enrollee characteristics explained all, or most, of the change in the gap in access measures between groups.

The results indicated that changes in the characteristics of Medicaid and PeachCare enrollees explained little or none of the changes in the gap for most of the access measures. This included whether enrollees got appointments when needed, had one or more non-ER visits, saw a specialist, or perceived problems in getting care when they or their physician felt it was needed. For one access measure, however, whether the enrollee had a person they thought of as their own doctor or nurse, enrollee characteristics explained all of the change. Overall, these results suggested that changes in the provider system or other factors, perhaps even how families accessed the system (e.g., more seeking of care from safety net providers) were behind the narrowing of gaps seen between Medicaid and PeachCare children 2000 to 2003. This finding is supported by a growth in the number of providers with paid claims from under 28,000 in fiscal year (FY) 2000 to over 40,000 in FY 2004 (Georgia Department of Community Health, 2000, 2004).

Discussion

Our hypotheses asserted that there were differential advantages in access and satisfaction for PeachCare versus Medicaid children, and that these differences would widen as PeachCare expanded and demand grew. Consistent with our hypothesis we found an advantage for PeachCare over Medicaid children in access and satisfaction in 2000, but contrary to our hypotheses of exacerbated differences in access, these gaps were diminished by 2003. Related to this, there was a decrease in a key utilization measure (number of children with no non-emergency visits) among Medicaid children and an increase among PeachCare children. Overall, both groups appear better off in 2003 than 2000, especially since both groups appear to have enrolled a greater share of children reporting excellent health. At the same time, while levels of satisfaction with the both programs improved, the gap between PeachCare and Medicaid enrollees actually increased.

As the PeachCare and other SCHIPs matured over the latter 1990s and moved into the new decade, States may have struggled not only with expanding enrollments and transitioning children between the two programs, but also maintaining adequate provider networks to serve each population well. Our results indicate that Georgia, a State with major increases in enrollment concurrent with stable provider reimbursements, kept levels of satisfaction at generally high levels even while differences in access for PeachCare, relative to Medicaid, children were diminished.

One question for policymakers as the programs continue to grow is what role inherent family characteristics versus provider characteristics play. In focus groups from an earlier study (Bronstein, Adams, and Florence, 2006) parents expressed appreciation for coverage, but SCHIP parents were more vocal and felt SCHIP coverage did not have the same stigma as Medicaid. Providers noted that patients had a difficult time understanding PCCM and that Medicaid families frequently used emergency rooms. Georgia's SCHIP parents tended to have their previous provider as their assigned gatekeeper under PCCM and placed a high value on this while some Medicaid families resented restricted provider choice. Growing differences in satisfaction between the programs seen in our data may be associated with the differential stigma of Medicaid (Ketsche et al., 2007a) even while access appears to have been facilitated by growth in provider participation/volume. The DCH in Georgia felt this increase in access was due to doubling of dentists' payment rates, a separation of child and adult therapy programs, and a requirement that providers register each location in which they provided Medicaid services.

It is important to note that healthier, perhaps more compliant children/families enrolled in these programs as the pool of uninsured children was further penetrated. This may have enabled participating providers to meet children's needs without significant physician fee increases. While the surprising large expansion in Atlanta (outreach was greater in rural areas) may have strained provider capacity, it is quite possible that safety net providers, along with the growth in total providers previously noted, played a role in improving access for Medicaid children. We did include measures of safety net providers (presence of FQHC) in the regressions and found a positive effect on access. Hence, it is the private provider network and/or how families accessed the two provider systems that appear to have improved access. Finally, even while Medicaid/Medicare fees were stable, the relative decline in private fees relative to Medicare 2000-2005 (Tu and Ginsberg, 2006) may have made Medicaid clientele somewhat more attractive as both families and providers adjusted to the new program.

Overall Summary

From the policy perspective, the goal of the SCHIP and Medicaid Program is to improve overall health, for those eligible to have access to quality care, and for both populations to be satisfied with care received. While the increases in health status seen in the descriptive data cannot be directly attributed to program expansion, this is a positive situation for Georgia and could bode well for further expansion. Moreover, differences in access were shown to decrease between the two programs, in part due to improvements in provider availability and access for Medicaid children. Where differences remain, both groups appear better off, although there are persistent differences in Medicaid families access to help/advice during office hours calls and, perhaps related to this, a greater tendency of Medicaid families to rate their overall health care plan lower.

Overall, these results suggest that States can run a separate SCHIP that provides high quality care and that does not negatively interact with its traditional program. This is further bolstered by the observation that the overall mean ratings of Georgia's public programs, 8.8 for Medicaid and 9.1 for PeachCare, compares well with mean ratings of 7.6 for Medicaid (and 7.46 for commercial) enrollees (Roohan et al., 2003). The State's policies aimed at improving provider participation appeared effective and should be sustained, but other policies may be needed to diminish remaining differences between Medicaid and PeachCare families' satisfaction (Ketsche et al., 2007a). Other research on Georgia's publicly insured children indicate that policies that help Medicaid children transition into PeachCare when eligibility changes at age 6 could assure continuity of coverage and enroll children in a program with lower treatment stigma (Ketsche et al., 2007b).

Other policies might include increasing physician payment rates as well as family education regarding how managed delivery systems should be used. Education concerning timeliness for appointments, review of transportation, and other non-financial barriers, such as language and literacy should also be considered. If reauthorization provides sufficient revenues for Georgia to continue serving children in the Medicaid Program and SCHIP policymakers may be able to test and evaluate such policies. They will need to incorporate these concerns/issues into ongoing dialogue and contract language with the private managed care companies now serving much of the Georgia Medicaid child clientele.