Wisconsin's BadgerCare Plus Reform: Impact on Low-Income Families' Enrollment and Retention in Public Coverage

Abstract

Objectives

To examine the impact of a Wisconsin health care reform enacted in early 2008 on public insurance enrollment and retention.

Data Sources

Administrative data covering the period January 2007 to November 2009.

Study Design

We calculate unadjusted enrollment trends and exit rates stratified by age, income group, and enrollment mode. Kaplan–Meier curves and Cox proportional hazards models are estimated to assess the impact of the reform on program exits.

Principal Findings

Overall enrollment increased by approximately one-third and exit rates decreased by approximately one-fifth. The majority of new enrollment came from the previously income eligible.

Conclusions

Wisconsin's enactment of eligibility expansions coupled with administrative simplification and targeted marketing and outreach efforts were successful in enrolling and retaining low-income children and families in public coverage.

Wisconsin's BadgerCare Plus (BC+) program was designed to ensure access to health insurance coverage to virtually all Wisconsin children and to bolster coverage for parents and other caretaker adults. The program, launched in February 2008, expanded upon BadgerCare (Wisconsin's Children's Health Insurance Program) and Medicaid. BC+ combines its predecessor programs into one streamlined umbrella program and its introduction was characterized by the following reforms: an eligibility expansion augmented with an autoenrollment initiative; simplification of eligibility rules and enrollment and verification processes; and an aggressive marketing and outreach campaign.

This paper provides estimates of the impact of the reform effort on program enrollment and the continuity of coverage. We find that the introduction of BC+ was associated with increases in enrollment and decreases in exit rates, suggesting that the Wisconsin experience may be instructive for other states interested in achieving similar aims.

POLICY BACKGROUND AND KEY ELEMENTS OF REFORM

Eligibility Expansions

Panel A of Figure 1 illustrates the income eligibility criteria in place pre- and postintroduction of BC+, and Panel B displays the changes in premium requirements and crowd-out provisions enacted with the reform effort. BC+ eliminated the income eligibility ceiling for children. It operates as a single program with two insurance products: the Standard Plan, for enrollees with incomes <200 percent of the Federal Poverty Level (FPL), and the Benchmark Plan, for enrollees >200 percent FPL. The former is the traditional Medicaid plan and requires only minimal cost sharing, whereas the latter is comprised of a more limited set of covered services and requires copayments on nonpreventive services, similar to private insurance policies. The premium threshold for children was set at 150 percent FPL under BadgerCare and was raised to 200 percent FPL under BC+. Modest-income children (200–300 percent FPL) enrolled in the Benchmark Plan are subject to premium payments that increase with family income level; premiums start at U.S.$10 per month and are capped at 5 percent of total monthly income. The families of higher income children (>300 percent FPL) are required to pay the full cost of coverage in the Benchmark Plan, which amounted to approximately U.S.$100 per month in 2008.

Figure 1.

Income Eligibility, Premium, and Crowd-Out Provisions Pre- and Post-BadgerCare Plus

Recognizing that nontraditional family structures are highly prevalent among low-income households (Gorman and Braverman 2008), BC+ expanded the definition of parental eligibility to include low-income caretaker relatives. In contrast to the 200 percent FPL income threshold imposed for children, the sliding-scale premium begins at 150 percent FPL for parents and caretakers; again, total family premium contributions are capped at 5 percent of monthly income.

Before the launch of BC+, anti-crowd-out provisions were applied in the BadgerCare program but not in the Medicaid program. Under BC+, applicants with incomes over 150 percent FPL are subject to anti-crowd-out provisions. With good-cause exceptions, these individuals face a 3-month waiting period for dropped coverage and they cannot have been offered employer-sponsored insurance (ESI) during the past 12 months or have the opportunity to enroll in ESI during the upcoming 3 months. The employer must cover at least 80 percent of the premium for the crowd-out provisions to apply.

Autoconversion and Administrative Simplification

Wisconsin's Department of Health Services (DHS) carried out a one-time autoconversion effort at program launch, applying the new eligibility criteria to automatically enroll previously ineligible individuals for whom there was current information in the state's enrollment database. For example, the state “autoconverted” newly eligible siblings and parents of children who were already enrolled in state health programs, along with eligible children and parents whose applications for public coverage were pending.1 Approximately 44,000 individuals were enrolled through this autoconversion process.

Administrative simplification was intended to increase initial enrollment of the target population, lower disenrollment rates among current members, and reduce the workload and costs associated with eligibility determination and recertification. Before the reform, members of the same household were often eligible for different programs, each with its own income counting rules and reporting requirements. Under BC+, these rules and requirements were streamlined, both reducing the reporting burden for families and making the program easier to administer. Income counting rules have been simplified and reporting requirements for changes in income and family structure among current enrollees have been reduced. Additionally, BC+ applicants have been relieved of the responsibility to document lack of access to ESI. This responsibility has shifted to a reporting relationship between employers and the state, under which employers of BC+ applicants are required to report health insurance offerings to a state-administered database.

DHS, in order to further ease barriers to program participation, enhanced and promoted an existing online application tool (available at https://access.wisconsin.gov/access/). Community organizations have been trained to assist potential applicants in using the online tool, further facilitating its take-up. In the year before the launch of BC+, approximately 15 percent of applications were submitted online; this percentage doubled during the first year of BC+ and continues to grow.

Marketing and Outreach

Wisconsin undertook a large marketing effort to promote the message that BC+ is for “all kids.” The brand purposely avoids any association between BC+ and traditional welfare programs, intending to reach eligible families who otherwise may have lacked knowledge of or perceived stigma with public programs. DHS partnered with over 200 community organizations to help enroll eligible families into the program and awarded mini-grants of up to U.S.$25,000 to 31 organizations, through which the organizations received a U.S.$50 “finder's fee” per approved application.

DATA, METHODS, AND RESULTS

The data for the study are from Wisconsin's CARES system, an administrative database maintained by DHS. The study population includes all child and parent/caretaker relative (collectively referred to as “adults” and/or “parents” throughout the paper) enrollees in BC+ and its predecessor programs. The study period is from January 2007 through November 2009.

The CARES system collects a variety of sociodemographic measures, which are used as both stratifying variables and covariates (displayed in Table SA1). The analytic sample is comprised of 1,310,812 unique enrollment spells contributed by 985,092 unique individuals.

Methods

The analytic aims are as follows:

1. Calculate enrollment growth subsequent to program launch.

2. Assess the extent to which the socioeconomic composition of program enrollees changed over the study period.

3. Assess whether the autoconverted enrollees' exit patterns differed from participants who enrolled through more traditional pathways.

4. Estimate the effect of program launch on the likelihood of disenrollment from public coverage.

Aims 1 and 2 are executed by computing total program enrollment by month over the course of the study, in aggregate and stratified by both age group and income group. To test for the presence of differential exit patterns between autoconverts and traditional enrollees—Aim 3—we estimate Kaplan–Meier survival curves over a 22-month period for both populations and test for differences in the curves using a log-rank test. We also estimate a Cox proportional hazards regression model to examine covariate-adjusted differences in exit rates between the two groups (Cox 1972).

Finally, we examine unadjusted and regression-adjusted exit rates in the time periods pre- and postreform. We estimate a Cox proportional hazards model that includes the vector of sociodemographic covariates available in the CARES data as well as monthly unemployment rates measured at the county level by the Bureau of Labor Statistics (covariates displayed in Table 2). To capture the changes in the BC+ program, we include three time indicators in the Cox specification: one reflecting the preperiod (January 2007–December 2007), which serves as the reference group; one reflecting the transition period (January 2008–March 2008); and one reflecting the postperiod (April 2008–November 2009). We include only new enrollment spells to avoid any bias arising from left censoring.

Table 2.

Cox Proportional Hazards Estimates of the Determinants of Disenrollment, January 2007–November 2009

Variable	Time Period Only (1)	w/Demographics (2)	Demographics, Socioeconomics (3)	Kids 1–17 <100% FPL (4)
Transitional	1.053**	1.064**	1.034**	0.925**
(January 2008–March 2008)	[0.008]	[0.008]	[0.008]	[0.014]
Post	0.755**	0.750**	0.824**	0.721**
(April 2008–November 2009)	[0.004]	[0.004]	[0.005]	[0.008]
Youngest household member 1–5 years		0.673**	0.628**	0.730**
		[0.007]	[0.007]	[0.010]
Youngest household member 6–12 years		0.660**	0.574**	0.796**
		[0.007]	[0.006]	[0.010]
Youngest household member 13–18 years		0.662**	0.553**	0.864**
		[0.007]	[0.006]	[0.011]
No children in household		0.751**	0.611**
		[0.008]	[0.007]
Female		0.991*	0.989**	0.974**
		[0.004]	[0.004]	[0.007]
Number of adults in household		1.205**	1.141**	0.987*
		[0.003]	[0.003]	[0.006]
Number of kids in household		0.928**	0.945**	0.978**
		[0.001]	[0.001]	[0.003]
High school graduate		0.963**	0.961**	0.920**
		[0.004]	[0.004]	[0.008]
More than high school		1.033**	0.996	0.977*
		[0.006]	[0.006]	[0.011]
Rural		0.366**	0.338**	0.175**
		[0.099]	[0.092]	[0.067]
Income under 150% FPL			0.874**
			[0.009]
Income between 150% and 200% FPL			1.128**
			[0.010]
Premium required			2.779**
			[0.021]
County unemployment rate			0.919**	0.918**
			[0.001]	[0.002]
County		X	X	X
Observations	8,096,411	8,096,411	8,096,411	2,965,289
Log pseudo-likelihood	−3,679,962.9	−3,674,051.3	−3,650,113.8	−890,840.67

Notes. Sample: spells originating in or after January 2007.

“Premium required” is a time-varying dummy variable that reflects whether a premium was required in a given month.

Robust standard errors in brackets, clustered at the individual level.

Coefficients reported as hazard ratios.

Preperiod (January 2007–December 2007) is excluded category in all columns.

Youngest household member 0 years, male, less than high school education, urban residence are excluded categories in Columns 2–4.

Income over 200% FPL, premium not required are excluded categories in Column 3.

Columns 2–4 include controls for county of residence, as denoted by an “X.”

FPL, federal poverty level.

^**p<.01

^*p<.05.

Results

Table 1 displays pre and postperiod enrollment numbers for children (Panel A) and adults (Panel B). Month-by-month enrollment and exit figures are available in Figure SA1. The number of children enrolled in the program increased 29 percent between December 2007 and November 2009, from 339,256 to 436,389; adult enrollment increased by 46 percent, from 176,674 to 257,066. Notably, over half (59 percent) of the increase in child enrollment came from the ranks of children living in families with incomes below 150 percent FPL, all of whom were income eligible for public coverage before the reform.

Table 1.

BadgerCare Plus Enrollment by Federal Poverty Level

% FPL	December 2007	November 2009	Change	% of Change
Panel A: Children
All	339,256	436,389	97,133	100
<150	320,076	377,615	57,539	59
150–200	18,339	41,375	23,036	24
200–300	685	14,493	13,808	14
300+	156	2,906	2,750	3
Panel B: Adults
All	176,674	257,066	80,392	100
<150	166,393	226,009	59,616	74
150–200	9,983	27,307	17,324	22
200+	298	3,750	3,452	4

FPL, federal poverty level.

The autoconverts comprised roughly 60 percent of new enrollees entering into the program in February 2008. Approximately two-thirds of both autoconverts and other new enrollees had incomes that were <150 percent FPL and similar proportions in both groups were subject to premium payments (sociodemographic profiles displayed in Table SA2). Both the Kaplan–Meier (Figure SA2) and Cox proportional hazards (Table SA3) specifications suggest that program retention was considerably poorer for autoconverts subject to premium requirements than for other new enrollees subject to premium requirements; perhaps more important, both models demonstrate that the likelihood of disenrollment was very high for all new enrollees subject to premium payments, regardless of enrollment mode. In contrast, disenrollment among lower income (i.e., those not subject to premium payments) autoconverts was only modestly higher than traditional enrollees with comparable incomes. The adjusted hazard of disenrollment was approximately 13 percent higher for low-income autoconverts relative to their peers who enrolled via an alternate mode (Table SA3, Column 4).

Children had lower exit rates over the study period than adults; an average of 2.6 percent of children exited each month compared with an average of 4.2 percent of adults. Both groups experienced similar trends in exits: Exit rates rose slightly throughout 2007, rose during program transition (sharply so for parents), and declined in a nonlinear manner after April 2008 (Figure SA1, Panel B). The average monthly unadjusted exit rate fell 22 percent for children and 15 percent for adults from the pre to the postperiod.

The Cox regression results on the sample of 809,050 unique spells (contributed by 653,754 unique individuals) originating during the study period suggest that the hazard of disenrollment fell by 18 percent in the postperiod relative to the preperiod, net of the influences of socioeconomic covariates and county unemployment rates (Table 2, Column 3). The adjusted hazard of disenrollment during the transition period (January 2008–March 2008) was roughly 3 percent higher than in the preperiod. These percent differences are statistically significant at the 0.01 level.

As displayed in Figure 1, members in different age and income strata were subject to a variety of different policy interventions, making it difficult to isolate the influence of any given intervention on exit rates. However, there is some heuristic evidence that administrative simplification efforts—net of the influences of crowd-out and premium provision changes—were associated with improved retention. Specifically, we estimate the Cox specification on a subsample of children who experienced no changes in crowd-out, eligibility, or premium provisions (children ages 1–17 with family incomes <100 percent FPL) and find that the decline in the hazard of disenrollment is of even larger magnitude than the estimate for the larger sample (Table 2, Column 4). Thus, we can rule out that changes in the crowd-out and/or premium provisions were the sole drivers behind the decreases in disenrollment.

DISCUSSION

Our findings suggest that substantial enrollment increases and retention improvements can be achieved through the mutually reinforcing combination of program elements adopted by Wisconsin's BC+ program: eligibility expansions coupled with autoenrollment; administrative simplification; and targeted marketing and outreach campaigns.

The early experience of BC+ is particularly applicable to those concerned with reaching the large numbers of uninsured children nationally who are currently eligible but not enrolled in Medicaid or CHIP. The majority of new enrollees in BC+ were children and adults who would have been income eligible under the previous program rules. This is consistent with previous studies showing a “welcome mat” or “woodwork effect” following other states' CHIP expansion and outreach initiatives (Rosenbach et al. 2003; Center for Children and Families 2007; Arjun and Guyer 2008;). This woodwork effect was enhanced by the autoconversion process, which enrolled previously ineligible siblings and parents.

Challenges and Limitations

The BC+ program has not been without its challenges and setbacks. Milwaukee County, Wisconsin's most populous area and the area with the most potential enrollees has historically experienced widespread delays in its ability to accept and process public insurance applications, causing difficulties in the local launch of BC+. A class action lawsuit regarding such delays was settled in April 2009, a result of which was a state takeover of the enrollment process for the Milwaukee area (Shultze 2009).

There have been additional challenges in some areas of program implementation. BC+ applicants subject to premium payments often fail to complete enrollment or drop out of the program. Also, the database that tracks ESI offerings is not yet fully populated, and given the dynamism of the private insurance market keeping this database up to date will be an ongoing challenge.

Going Forward: Implementing National Health Care Reform

The recently passed federal Patient Protection and Affordable Care Act relies heavily on Medicaid and CHIP expansions to cover children and adults under 133 percent FPL, an eligibility threshold already exceeded by Wisconsin's BC+ program. While BC+ was designed and initiated before the passage of federal health care reform, its lessons remain relevant for reaching, enrolling, and retaining children and newly covered adults in public and subsidized coverage going forward.

Wisconsin, like all states, will be recalibrating its approach to coverage as it implements the federal health reform law. Among the major changes in play for the state in 2014 is the transfer of childless adults with incomes below 133 percent FPL into traditional Medicaid from their currently separately operated plan. The state also faces the option of moving all currently enrolled parents and childless adults who are above 133 percent FPL off of BC+ into subsidized coverage plans administered through the health insurance exchange. These choices will need to be informed by a full accounting of potential trade-offs—particularly as lower income populations may opt to pay the noncoverage penalty or qualify for exemptions from the individual mandate if they find themselves ineligible for traditional public programs.

The changes ahead also could potentially separate what has been a single coverage route for Wisconsin's low-income children and their parents through BC+. Coverage of parents along with their children has been an important element associated with strong program take-up and also with quality family health care (Aizer and Grogger 2003; Dubay and Kenney 2003; Wolfe and Scrivner 2005; Sommers 2006;). The choice to separate parents' coverage, keeping children above 133 percent FPL in BC+ while placing their parents in the exchanges, will bring new enrollment challenges. With separate health plans covering low-income children and their parents, states will be further challenged to achieve another goal of national health care reform for the vulnerable populations that they serve: clear messaging and consistent goals in health care utilization and clinical quality.

SUPPORTING INFORMATION

Additional supporting information may be found in the online version of this article:

Appendix SA1: Author Matrix.

Figure SA1: Monthly Enrollment and Exit Rates in BadgerCare Plus, January 2007–November 2009.

Figure SA2: Kaplan-Meier Estimates of Disenrollment: Auto-Converted Enrollees Versus Other New February 2008 Enrollees.

Table SA1: Descriptive Statistics: Enrollment Spells of the CARES Study Population.

Table SA2: Descriptive Statistics: Auto-Converts vs. Other New February 2008 Entrants.

Table SA3: Cox Proportional Hazards Estimates of Disenrollment, by Auto-Conversion Status. Sample: February 2008 BadgerCare Plus Entrants.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.