Since the Supreme Court decided that the Affordable Care Act's (ACA) Medicaid expansion is optional for the states, several have obtained federal approval to use Section 1115 waivers to expand Medicaid while changing its coverage and benefits design. There has long been concern that policy making for Medicaid populations may lack meaningful engagement with low-income constituents, and therefore the ACA established a new process under which the public can submit comments on pending Medicaid waiver applications. We analyzed 291 comment letters submitted to federal regulators pertaining to Medicaid Section 1115 waiver applications in the first five states to seek such waivers: Arkansas, Indiana, Iowa, Michigan, and Pennsylvania. We found that individual citizens, including those who identified as Medicaid-eligible, submitted a sizable majority of the comment letters. Comment letters tended to mention controversial provisions of the waivers and reflected the competing political rhetoric of “personal responsibility” versus “vulnerable populations.” Despite the fact that the federal government seemed likely to approve the waiver applications, we found robust public engagement, reflecting the salience of the issue of Medicaid expansion under the ACA. Our findings are consistent with the argument that Medicaid is a program of growing centrality in US health politics.
State Medicaid programs comprise the largest single source of health coverage in the United States, covering more than 70 million people and accounting for 15 percent of personal health spending (CMS 2015; Hartman et al. 2015). Created in 1965, Medicaid programs have long been an important source of health insurance coverage for certain impoverished populations, including pregnant women; children; parents; and those who are disabled, blind, or aged (Iglehart and Sommers 2015). Expanding state Medicaid coverage to low-income, nonelderly adults without dependent children was a cornerstone of the Affordable Care Act's (ACA) objective of ensuring universal health insurance coverage for citizens. In 2012, however, the Supreme Court ruled that the ACA's Medicaid expansion constituted a new program distinct from existing Medicaid programs and was therefore optional for states.1 Decisions on whether to expand Medicaid generally followed party lines; eighteen of the twenty-two states that did not expand Medicaid in 2014 had a unified Republican gubernatorial administration and legislature (Béland, Rocco, and Waddan 2015). As state policy makers grappled with the politics of whether or how to adopt the ACA's Medicaid expansion to nonelderly adults, a new policy option emerged: using a so-called Section 1115 waiver to modify the Medicaid expansion.
Although Section 1115 waivers have existed for decades, the ACA established a new process under which the public can submit comments on pending Medicaid waiver applications. There has long been concern that policy-making processes for Medicaid programs may be undermined by a lack of meaningful participation from constituents (Grogan and Gusmano 2007). The benefits of involving the public in health policy-making decisions can be conceptualized as intrinsic benefits (i.e., public involvement is consistent with participatory democracy), instrumental benefits (i.e., improved quality and outcomes for the public), and developmental benefits (i.e., improved public knowledge about policy) (Conklin, Slote Morris, and Nolte 2010). In the context of Section 1115 waivers, consultation might lead to intrinsic social benefits by engaging a group—low-income adults without children—that has typically lacked political power. Given that the new Section 1115 waivers are subject to independent evaluation, including prespecified hypotheses in documents that are publicly available, public engagement might be an important process by which to influence outcomes. However, the implementation of a meaningful engagement process may raise challenges in terms of whether relevant constituencies are adequately represented or whether or how policy officials are responsive to public participants (Abelson et al. 2004).
Our study sought to assess the role of citizens, particularly low-income citizens served by state Medicaid programs, in the process of policy changes to privatize state Medicaid programs. To this end we analyzed the source and content of 291 unique comments submitted to the Centers for Medicare and Medicaid Services (CMS) pertaining to Section 1115 waivers to expand Medicaid to nonelderly adults in five states during 2013 to 2014: Arkansas, Indiana, Iowa, Michigan, and Pennsylvania. Our analysis found that a sizable number of comment letters were submitted by citizens who identified as potentially Medicaid-eligible and that comments from both citizens and organized interests tended to mention specific, and often controversial, provisions of the waiver application. Although CMS ultimately approved all the waiver applications, the agency did limit specific provisions of the waivers that did not appear to advance the policy goal of providing access to health care for low-income populations. In the context of the political salience of Medicaid expansion under the ACA, our findings suggest a role for increasing public engagement as part of a strategy to build a constituency among low-income citizens. This article is organized as follows: we provide a brief overview of Medicaid Section 1115 waivers and changes to the waiver process under the ACA and then present our study data and methods, followed by study results and a discussion of the implications of our findings.
Evolution of Medicaid Section 1115 Waivers
Section 1115 of the Social Security Act, established in 1962, allows the federal government to approve state-level experiments. In contrast to programmatic waivers, authorized under Section 1915 of the act, Section 1115 waivers are aimed at encouraging “laboratories of democracy” and require scientific evaluations of the their effects on outcomes (Rosenberg and Zaring 1995). Approximately fifty Medicaid waivers were approved before the 1990s, but these waivers proliferated under both the Clinton and the George W. Bush administrations (Thompson and Burke 2007). After Clinton's national health reform effort failed, the Clinton administration moved to streamline and simplify the waiver process for states, essentially allowing states to take the lead on health reforms that would expand coverage (Schneider 1997). Under the George W. Bush administration, Section 1115 waivers continued to be approved, with an increased focus on states' abilities to implement less generous coverage and require patient cost sharing (Coughlin et al. 2006). A common theme is that both the Clinton and the George W. Bush administrations viewed Section 1115 waivers as a way to respond to states' desires to shape their Medicaid programs, as opposed to a vehicle to promote policy learning through Medicaid experiments conducted in the laboratory of the states (Thompson and Burke 2007).
The ACA made several changes to the Medicaid Section 1115 waiver process. Notably, it required that the Department of Health and Human Services promulgate regulations to ensure meaningful opportunities for the public to comment in the approval process. In 2012 CMS issued a final rule delineating both state and federal processes for the approval of Section 1115 waivers under the ACA (CMS 2012). Prior to submitting the waiver application, states are required to provide the public with a detailed description of the waiver, hold at least two public hearings, and provide a thirty-day state comment period. In the final waiver application submitted to CMS, states are required to document the public comment process and responses to public input. At the federal level, CMS implements a thirty-day federal comment period within fifteen days of receipt of a state's waiver application. The CMS federal comment process, including all comments submitted, must be publicly available online. Although the rule does not require CMS to explicitly respond to public comments in the federal comment period, the new Section 1115 approval policies may have a significant impact on the implementation of the Medicaid waivers under the ACA (Watson 2015). This policy change to allow public comments provides an opportunity to study such comments regarding state Medicaid programs that was not previously available.
The five states included in our study (Arkansas, Indiana, Iowa, Michigan, and Pennsylvania) were the first states to obtain federal approval to implement the Medicaid expansion to nonelderly adults via a Section 1115 waiver in 2013 or 2014. However, the characteristics of each state's coverage for low-income populations prior to the waiver application varied substantially. Notably, while Indiana, Michigan, and Pennsylvania had significant experience with Medicaid managed care plans, Arkansas and Iowa had virtually no Medicaid beneficiaries in managed care prior to the waivers (see appendix A.1 for a description of Medicaid programs in the five states in 2012, prior to the Section 1115 waiver applications).
In the context of the ACA Medicaid expansion to nonelderly adults in the five states, Section 1115 waiver applications generally have sought to make coverage more closely resemble the private health insurance plans that are available via the state or federal marketplaces or via employers (table 1).
To this end, the states' applications included provisions such as requiring premiums or premium-like contributions (all five states); disenrolling individuals for nonpayment of premiums (Indiana, Iowa, and Pennsylvania); or waiving coverage of nonemergency transportation to obtain medical care (Arkansas, Indiana, Iowa, and Pennsylvania). Arkansas and Iowa requested using the federally facilitated marketplace to provide coverage to Medicaid-eligible individuals. In contrast, Indiana, Michigan, and Pennsylvania sought changes to Medicaid coverage that moved it away from the feeling of a social welfare program, such as the use of personalized health care spending accounts.
Role of Low-Income Citizens in Medicaid Policy Making
Prior research has found mixed results on whether the public comment process facilitates meaningful consultation from politically weak individuals or groups (Golden 1998; Yackee and Yackee 2006). Previous public comment processes have been found to be dominated by organized interests from the business community (Yackee and Yackee 2006), suggesting that individual citizens, and in particular low-income citizens, lack the political power to alter the trajectory of the rule-making process (Schneider and Ingram 1993). The new Medicaid waivers under the ACA provide a compelling opportunity to study the role of citizens, especially those of low income, in the Medicaid policy-making process for several reasons. First, as described above, the new transparency process under the ACA provides an opportunity for citizens to directly communicate with policy makers. Second, Medicaid expansion under the ACA was and continues to be a highly salient issue. Third, a state's Medicaid expansion decision will likely be not only politically salient but personally relevant to those individuals likely to be eligible for Medicaid coverage.
The objectives of our study were to: (1) characterize the type of commenters under the new Section 1115 waiver comment process, (2) determine specific comment elements and types of arguments employed by different types of commenters, and (3) assess whether the type of commenter or content of the comment letters were associated with support or opposition to the Medicaid waiver applications.
Participants are more likely to influence policy through comment processes when they provide useful, substantive information, especially at the early stages of regulatory decision making (Naughton et al. 2009; Costa, Desmarais, and Hird 2016). A better understanding of the commenters and the arguments they offered during the public comment process might provide insight into how the new policy is working and the extent to which consulting the public influences CMS approval of the waivers. In the long term, understanding the quality of arguments provided in public comments may provide insight into whether Medicaid expansion decisions are building a mobilized constituency.
We obtained 1,067 comment submissions that were submitted to CMS pertaining to Medicaid Section 1115 waiver applications in five states (Arkansas, Indiana, Iowa, Michigan, and Pennsylvania). Although Pennsylvania in mid-2015 ended its waiver program and moved to the ACA Medicaid expansion model, we included these comments because the waiver was approved and implemented for a short time. We included comments submitted to federal regulators because we were interested in the new comment process established under the ACA. CMS has responsibility for approving waivers and ensuring that the states are overseeing the scientific evaluation of these waivers. Therefore, we did not include comments submitted within each state prior to the waiver application submission.
The data we received contained all comment submissions via an online text field or via an attached document. We reviewed all comment submissions and excluded those in which the comment submission field was empty of text, contained fewer than two sentences, or contained text unrelated to Medicaid waivers. Empty comment fields or those with only a small amount of text or unrelated text may have represented incomplete attempts at submitting comments during the online submission process. Indiana and Pennsylvania in particular had a large number of identical form letters submitted, suggesting that advocacy groups in these states were successful in organizing a comment-writing campaign. Ultimately, form letters were counted as one unique comment because their content was identical, meaning that counting them as individual comments could bias results. Additionally, public officials tend to group comment themes together in responses to public comments, rather than responding to individual considered comment letters. Because of a limit on length in the online submission fields, we collapsed multiple comment submissions that were part of the same comment letter into one unique comment (see appendix A.2 for counts of included and excluded comment submissions by state). Our final analytic sample included 291 unique comment letters (Arkansas = 4; Indiana = 103; Iowa = 19; Michigan = 7; and Pennsylvania = 158). The number of comment letters in each state varied widely. We note that Arkansas, which was the first state to submit a waiver application (it was approved in September 2013), had the fewest comments submitted to CMS. Iowa and Michigan, which followed Arkansas and had waivers approval in December 2013, similarly had fewer comments submitted to CMS. However, Pennsylvania and Indiana, which did not submit waiver applications until 2014, had far more comments submitted to CMS. Given that CMS imposed special terms and conditions on waivers, such as prohibiting cost sharing for people with incomes of less than 100 percent of federal poverty, it is possible that advocacy groups and/or citizens recognized that comments could potentially sway federal regulators. Therefore, the increasing number of comments submitted over time might reflect the growth of advocacy around these Medicaid waivers.
We used summative content analysis methods to develop a coding instrument to study the content of the comment letters (Hsieh and Shannon 2005). Under this approach we sought to quantify discrete characteristics of the content of the letters, as well as to understand the meaning of content (Potter and Levine-Donnerstein 1999). We developed a fifteen-item coding instrument that included codes for content linked to provisions of the waiver applications related to coverage design, to provisions related to coverage limits, to rhetoric pertaining to the Medicaid expansion population, or to anecdotal or data-driven arguments. Finally, the coding instrument identified whether the comment gave an overall impression of supporting the waiver application, supporting the application conditional on specific changes, or opposing the application. Each item in the instrument was a binary measure (“yes” if the item was mentioned in the comment letter; “no” if the item was not mentioned in the comment letter).
We pilot tested the coding instrument on a random sample of twenty-five comment letters that had been submitted to the states prior to the states' application to CMS. Therefore, comments included in the pilot testing were not included in our final analytic data. Two authors (AK, NG) independently coded five comments per round of pilot testing and then met with a third author (MJ) to discuss instrument functioning, content, and clarity. The instrument was refined, with items edited, added, and deleted during this iterative process.
After pilot testing of the instrument was complete, two authors (AK, NG) then independently coded a random sample of forty comment letters. We used prevalence- and bias-adjusted kappa statistics, which provide a measure of intercoder reliability that is adjusted to assess reliability for binary items where yes and no values are not evenly distributed (Byrt, Bishop, and Carlin 1993). Intercoder reliability was moderate to excellent, with K ranging from 0.50 to 0.90 and a mean K = 0.72 (see appendix A.3 for kappa scores for specific items). A third author (MJ) met with the coders to adjudicate all items for which there was disagreement. The remaining 251 unique comments were then randomly divided and independently coded by the two coders (AK, NG).
First, we measured whether comment letters mentioned any specific features of coverage design under the Medicaid waivers. These were based on features included in states' waiver applications and included: private or market-based health insurance coverage, beneficiary cost sharing, behavior incentives, a lack of nonemergency transportation, or a requirement that beneficiaries participate in employment activities (i.e., the work requirement). We also measured whether comment letters mentioned limits on coverage proposed in the waiver applications, defined as: a lockout from coverage for nonpayment of cost sharing, limits on utilization, or limits on the type of medical provider.
Second, we measured whether comment letters included rhetoric that assigned certain characteristics to the Medicaid expansion population. We conceptualized rhetoric as persuasive writing that incorporated the use of symbolic speech to invoke a specific sentiment, either in support of or in opposition to the waiver application (Russell et al. 2008). We examined whether the comment letters used the rhetoric of “personal responsibility,” defined as using the exact phrase, discussing beneficiaries' responsibility to manage their own care, or referencing consumer choice or consumer-driven care. We also examined whether letters used the rhetoric of “vulnerable populations,” defined as using the exact phrase or mentioning the needs of specific populations identified by the Centers for Disease Control and Prevention as vulnerable (e.g., homeless individuals, migrant workers, or persons with disabilities) (CDC 2015). These arguments are conceptualized as symbolic arguments.
Third, we measured whether comment letters included an anecdotal or research-driven argument. Anecdotal arguments were defined as mentioning the letter writer's personal experience with Medicaid, either as a patient, friend or family member, provider, or advocate. Research-driven arguments were defined as mentioning a research study, referring to clinical or public health guidelines, or using statistics to illustrate a point. These arguments are conceptualized as technical arguments.
Finally, we measured whether each comment letter gave the overall impression of supporting the waiver application, supporting the application with specific changes, or opposing the application.
Type of Commenter
For each comment letter, we recorded the type of commenter. We identified four broad types of commenters. Citizen commenters were defined as letters from individuals not identified as affiliated with any organization. To identify citizen commenters who would be personally affected by the waivers, we recorded whether they identified themselves as being uninsured or lower income, potentially eligible for Medicaid under the expansion, or currently or formerly enrolled in Medicaid. Health care provider letters were defined as those from any organization representing those providing care directly to patients, including medical societies, health systems, hospital or long-term care facilities or associations, or insurance plans. Nonhealth industry letters were defined as those from organizations representing industries that did not provide patient care (e.g., chambers of commerce or health care industry consulting firms). Advocacy group letters were defined as letters from organizations identified as representing a certain constituency (e.g., disease-specific patient advocacy groups, community organizations, or elected officials).
We used frequencies and descriptive statistics to analyze the content of the comment letters, overall and by type of commenter. We used cross tabulations to describe associations between specific content in the comment letters and whether the inclusion of content was associated with support, conditional support, or opposition to the waiver applications. Chi-square tests were used to assess the differences in the content and the overall sentiment of comment letters by type of commenter and to assess the differences in the overall sentiment of letters mentioning different specific provisions of the waiver applications. Although the data are a census of all nonduplicated comment letters submitted to CMS for the five states' Medicaid waiver applications, results from chi-square tests may be useful if comment letters are considered a sample of public sentiment.
Of the 291 unique comment letters pertaining to Medicaid Section 1115 waiver applications in five states, 186 (64 percent) were submitted by citizens, of which 55 identified as Medicaid-eligible citizens; 65 (22 percent) were submitted by advocacy groups; 25 (9 percent) were submitted by health care providers; and 14 (5 percent) were submitted by nonhealth industry groups.
Table 2 shows the percentage of specific content elements of comment letters overall and by type of commenter. Overall, a majority (60 percent) of comment letters mentioned at least one specific waiver provision in the application, and a majority (53 percent) of letters mentioned at least one type of coverage limit in the waiver application. While 13 percent of letters included the rhetoric of personal responsibility, 41 percent included the rhetoric of vulnerable populations. Thirty-six percent of letters included an anecdotal (19 percent) or research-driven (18 percent) argument. There were statistically significant differences in the proportion of comment letters across different types of commenters that mentioned any specific waiver provision (χ2 = 26.9; p < 0.01); any benefit limit (χ2 = 25.1; p < 0.001); and any rhetoric (χ2 = 17.1; p < 0.01).
The mention of specific waiver provisions varied meaningfully by type of commenter. Consider, for example, beneficiary cost sharing, which was the most frequently mentioned specific provision overall. Cost sharing was mentioned in 82 percent of advocacy group letters, compared to 21 percent of letters from nonhealth industry groups. The controversial work requirement provision was mentioned in 51 percent of letters from advocacy groups, but no letter from a nonhealth industry group mentioned the work requirement provision. The frequency of specific waiver provisions mentioned in letters from citizens and health care providers was more similar, however.
Large differences by type of commenter also appeared with respect to the use of an anecdote or a research-driven argument. Anecdotes were employed in letters from citizens (28 percent), particularly Medicaid-eligible citizens (67 percent), but were rarely to never mentioned in letters from advocacy groups (3 percent) or health care providers or nonhealth industry groups (0 percent).
We counted form letters as one unique comment each; however, we assessed the content, frequency, and overall sentiment of form letters separately as they likely lack the same influence as unique letters. As shown in table 3, form letters described general opposition to the waiver, addressed concerns with specific benefits, or addressed concerns for specific populations. Form letters opposed or conditionally supported the waiver applications.
Figure 1 shows that overall, 20 percent of comment letters supported the waiver application, 26 percent supported the application conditional on specific changes, and 54 percent opposed the waiver application. The level of support for waiver applications varied by type of commenter (χ2 = 93.7; p < 0.001). Citizen commenters who identified themselves as Medicaid-eligible overwhelmingly opposed the waiver applications (75 percent opposed, 19 percent supported, 13 percent conditionally supported). This compares to 68 percent of citizen commenters overall opposing the waivers, 27 percent of health care provider commenters opposing the waivers, 0 percent of nonhealth industry groups opposing the waivers, and 35 percent of advocacy groups opposing the waivers. Citizens largely expressed opposition to the waivers, health care providers were somewhat evenly divided, nonhealth industry groups supported the waivers, and advocacy groups generally offered support conditional on changes.
Figures 2 and 3 present the associations between the content of the comment letters and overall support or opposition to the waiver application. Figure 2 shows that comment letters that mentioned specific waiver provisions or benefit limits tended to conditionally support or oppose the waiver application, with the distribution of overall sentiment differing significantly by which domains of content were mentioned in letters.
Comment letters including the rhetoric of personal responsibility frequently supported the waiver application (51 percent), while letters including the rhetoric of vulnerable populations frequently opposed the waiver application (52 percent).
Figure 3 shows that of the comment letters mentioning specific waiver provisions, support was highest, at 38 percent, among those mentioning behavior incentives. In contrast, support for the waiver application was very low among letters mentioning other specific provisions, including cost sharing (12 percent), lack of nonemergency transportation (0 percent), or a work requirement (3 percent).
The ACA's Section 1115 comment process is new, and therefore we were unable to compare our findings in the present study to the approval of previous Section 1115 waivers. However, we examined some public comments pertaining to the first proposed federal rule implementing Medicare Part C (CMS 2004). Like the Medicaid waivers, the implementation of Part C created a greater role for privately delivered health care services in what had previously been a purely public insurance program. We found that 145 comments were submitted about Medicare Part C, from a range of individuals, advocacy groups, and health industry groups that was consistent with the type of commenters we observed for Medicaid waivers. In general, citizen commenters expressed general concerns, whereas organized interests provided more specific requests of changes to the proposed rule. In contrast, the Medicare Part C proposed rule received twenty comments from individual citizens (14 percent of all comment letters), which was lower than what we observed for the Medicaid waiver applications.
States' political decisions about whether and how to expand Medicaid eligibility to nonelderly, nondisabled adults under the ACA have enormous health and economic implications (Sommers, Baicker, and Epstein 2012; Price and Eibner 2013). In our content analysis of 291 unique comment letters under the Section 1115 approval process for Medicaid expansions under the ACA, we observed comments from individual citizens as well as organized interests. Both contained substantive content, as measured by the frequent mention of specific provisions contained in the Medicaid waiver applications. There appeared to be a distinction between the content most frequently mentioned in letters from citizens or health care providers versus that mentioned in letters from nonhealth industry groups. Citizens, particularly those who identified as Medicaid-eligible, overwhelmingly opposed the waiver applications. Perhaps not surprisingly, advocacy groups most often included requests for specific policy changes in their letters. Those who opposed the waivers' approval tended to mention specific, and in some cases controversial, waiver provisions such as a lack of nonemergency transportation for Medicaid beneficiaries. Those who supported the waivers tended to use rhetoric, and the inclusion of research-driven arguments was associated with support or conditional support of the waiver applications. Although 54 percent of all comment letters opposed approval of the waivers, the waiver applications in all five states were ultimately approved.
Our analysis of the new Medicaid Section 1115 waiver comment process differs from previous studies of political participation in the rule-making process in two notable ways. First, we found that citizen commenters submitted more than 60 percent of comments, whereas prior studies of federal notice-and-comment processes and our own comparison to the Medicare Part C rule discovered that business interests submit the majority of comments (Yackee and Yackee 2006). Second, in contrast to prior work suggesting that the substantive quality of citizen comments is low, especially compared to comments from business interests (Krawiec 2013), we found that the majority of comment letters included substantive content mentioning specific provisions of the waiver applications. Mentions of specific waiver provisions or proposed benefit limits were associated with opposition to the waivers or support conditional on specific changes. The fact that the use of waivers to expand Medicaid represented an issue of high salience in the context of different media framing of the ACA rollout may explain these differences (Gollust et al. 2014). There was also some evidence of an advocacy learning process, with advocacy groups learning over time that public comments could be used as tools to request federal regulators to place limitations on, if not outright reject, the waivers.
Findings of our study, however, are in line with a growing body of research arguing that Medicaid, although a means-tested program that serves low-income individuals and families, is of growing importance to US health policy (Iglehart and Sommers 2015). In particular, the present findings are consistent with the idea that the Obama administration encouraged advocacy groups to assert pressure on state governments regarding Medicaid expansion (Thompson and Gusmano 2014). It is too early to draw conclusions about whether Medicaid expansions broadly—and public engagement with state Medicaid program decisions specifically—will result in core Medicaid constituencies similar to Social Security or Medicare recipients (Grogan 2015). The present findings suggest that at least in the first two years of the implementation of the ACA, organized interests and individual citizens were mobilized to attempt to influence the outcomes of Medicaid Section 1115 waivers.
Findings from this study should be interpreted in light of several limitations. First, our comment letters are from five states that sought Medicaid Section 1115 waivers in 2013 or 2014, and results may not be generalizable to other states or future time periods. Second, we did not analyze comment letters submitted to state agencies and are therefore unable to make any claims about the Section 1115 approval process as it pertains to intrastate processes. In our study Arkansas and Michigan submitted relatively few comment letters to federal regulators, and we cannot definitively state why the comment counts differed so across states. However, we note that as more states submitted waiver applications, advocates might have inferred a chance to change the special terms and conditions issued by CMS, if not the waiver approval itself. Third, our measures of the content of comment letters involved the creation of a coding instrument to identify and analyze data. We have attempted to maximize the validity of the coding instrument through a rigorous pilot testing process, and we have achieved good intercoder reliability. Fourth, the nature of the present analysis does not permit us to measure the effects of the comment process on the implementation of the Medicaid waivers, health care access, or quality measures among Medicaid beneficiaries.
Implications for Health Policy
Although states have strong financial incentives to expand Medicaid to nonelderly adults, the Medicaid expansion decisions have become politically difficult given the bifurcation of public opinion along political party lines (Béland, Rocco, and Waddan 2016). There is some speculation that Section 1115 waivers might not be about policy innovation but rather are a way for conservative governors to have it both ways—that is, to capitalize on the benefits of Medicaid expansion under Obamacare while claiming to reject Obamacare. Obama administration officials publicly stated their support for waiver applications, raising the prospect that negative comments were likely to have little impact on the approval process. One question might be who commented and why, given that approval of the waivers seemed destined to occur. Our analysis of the public comments reflects some ambivalence among organized interests in that they seemed to endorse Medicaid expansion but raised issues with its implementation. This suggests that advocacy groups may have been willing to accept what they saw as an imperfect first step in expanded access to health insurance. Advocacy groups tended to request specific changes to the waivers and interestingly, often employed both rhetorical and technical arguments within the same comment. Organized interests appeared to have correctly inferred that specific policy suggestions could affect waiver approval and subsequent evaluations, as evidenced by CMS' issuance of special terms and conditions. This ambivalence did not occur to the same degree among citizen commenters, however, who were not legislative or advocacy professionals. Citizens were largely opposed to the waivers, favored traditional Medicaid expansion, and tended to focus on patient cost sharing and access to care.
A broader question that our analysis raises is whether, or to what extent, the submission of comment letters affected the structure or implementation of the waivers and consequently, whether public benefits were realized because of the new waiver comment process. Although our study design does not permit causal inference on the effect of the comment letters, the findings hint at a mixed answer to this question. We observed that commenters provided input on changes to waiver policy that they deemed necessary to serve the needs of low-income populations. However, there was a clear disconnect between citizen opposition to the waivers in favor of traditional Medicaid expansion and the fact that CMS ultimately approved all waivers.
Given the national health policy goal of achieving near-universal health insurance coverage, it was perhaps unlikely that CMS would disapprove any of the Section 1115 waivers in its entirety. CMS could and did, however, place limitations on some state provisions that arguably did not advance the program objectives of Medicaid, such as disapproving co-payments for those with incomes below 100 percent of the federal poverty level and restricting the waiver of nonemergency medical transportation to only one year.
Our findings have implications for knowledge about the creation of constituencies around social welfare programs. In recent decades Medicaid has evolved from a targeted medical welfare program linked to the receipt of cash assistance to a program serving an expanding portion of lower-income Americans (Grogan and Patashnik 2003). The data we present highlight the range of constituency groups that submitted comments, including, notably, a sizable response from citizens who identified themselves as Medicaid-eligible. It is not clear that a reciprocal relationship yet exists between participation and Medicaid policy, in contrast to the well-documented relationship between participation and social programs for seniors (Campbell 2003). The construct of Medicaid continues to connote a stigma associated with poverty and welfare (Allen et al. 2014), potentially hindering Medicaid policies from building constituencies in the same way that other policies might. Future research is needed to investigate how traditional or waiver-based Medicaid expansions may be correlated with political participation among lower-income populations. Although the five states in this study showed different baseline levels of managed care penetration, surprisingly, the private provision of Medicaid benefits in and of itself was not an issue highlighted in comment letters. An important empirical question is whether states that opt to expand Medicaid coverage via the state or the federal exchanges, such as Arkansas or Iowa, might build stronger citizen constituencies via increased solidarity between the lower- and middle-income populations.
In the early years of implementation of the ACA, much health policy attention has appropriately focused on quantifying the extension of insurance coverage and access to health care under state Medicaid expansions. Our study provides information about how Medicaid expansions via Section 1115 waivers were a salient issue that drew a sizable number of comment letters from individual citizens as well as organized interests. These results are consistent with the notion that Medicaid is emerging as a major social program with a sizable advocacy base. As Medicaid continues to evolve with increasing managed care penetration and coverage that includes characteristics more like private health insurance plans, it will be important to monitor federal and state policy makers' responsiveness to constituents. The new transparency in approval and evaluations of Section 1115 waivers, although not a panacea to improve political participation among impoverished populations with typically low levels of civic involvement, may be an effective way to engage the public in the Medicaid policy-making process.
1. National Federation of Independent Business v. Sebelius, 132 S. Ct. 2566 (2012).