States are required to conduct annual Medicaid redeterminations. How these redeterminations are undertaken is crucial to determining the nature of Medicaid coverage. There can be wide variations in the proportion of clients disenrolled, with potentially large numbers of people disenrolled each year. This case study of Illinois Medicaid shows how, as the Affordable Care Act added people, redeterminations were taking people off the rolls—about 25 percent of all Medicaid clients were disenrolled in one year. Many of these people were no longer eligible, but it appears that a larger number were in fact eligible but simply failed to comply with administrative requirements in a timely way. Balancing between the two imperatives of program integrity and continuity of care is a difficult act for Medicaid programs. The Illinois experience also illustrates impacts on information technology and outsourcing of eligibility functions, not to mention budget considerations.
Editor's Note: JHPPL has started an ACA Scholar-Practitioner Network (ASPN). The ASPN assembles people of different backgrounds (practitioners, stakeholders, and researchers) involved in state-level health reform implementation across the United States. The newly developed ASPN website documents ACA implementation research projects to assist policy makers, researchers, and journalists in identifying and integrating scholarly work on state-level implementation of the ACA. If you would like your work included on the ASPN website, please contact web coordinator Phillip Singer at email@example.com. You can visit the site at http://ssascholars.uchicago.edu/jhppl/.
JHPPL seeks to bring this important and timely work to the fore in Report on Health Reform Implementation, a recurring special section. The journal will publish essays in this section based on findings that emerge from network participants. Thanks to funding from the Robert Wood Johnson Foundation, all essays in the section are published open access.
—Colleen M. Grogan
The Affordable Care Act (ACA) has been responsible for the addition of about six hundred thousand Medicaid individuals in Illinois. Simultaneously, the Illinois Medicaid program dramatically increased efforts to redetermine existing clients, with the result that a large number of clients were removed from the rolls at the same time. The causes of renewed vigor in redetermination are familiar to any student of Medicaid: budget problems, inadequate infrastructure, and an interest in privatizing certain state functions. There are, however, questions about whether this was the best way to approach these issues and how, going forward, program integrity should be balanced with the desire for health care continuity.
The specific concern is this: for all the discussion of potential problems of churn (as it is called) between Medicaid and the ACA marketplace (Guerra and McMahon 2014; Rosenbaum 2015), the larger disrupter of continuity is churn within the Medicaid population that happens because of the federal requirement for annual redetermination of continued eligibility. Many of the people who are removed from eligibility are, in fact, eligible. This is a well-known problem (Ku, Steinmetz, and Bruen 2013), although it has not received as much academic attention since the advent of the ACA raised new problems. Nor has it received media attention—because the issue is complicated and because it is drowned out by perpetual claims of fraud by Medicaid opponents. But Medicaid churn remains important. There are very large numbers of individuals affected—in Illinois up to 25 percent of Medicaid clients are being disenrolled each year, the majority of whom are most likely eligible. Moreover, there is concerted attention from some groups touting the possibility of reducing Medicaid rolls through more stringent redeterminations, and several states have recently adopted legislation to do that.
The past several years in Illinois provide a case study of how this process—which is typically, but mistakenly, seen as primarily a technical bureaucratic issue—becomes a battleground over the fundamental nature of Medicaid coverage. This issue is all the more pressing given the increasing reliance on Medicaid as a primary source of insurance, especially in expansion states.
In the spring of 2012, Illinois was facing a serious budget problem in its Medicaid program, and the executive branch worked with the general assembly to adopt a series of measures to address the crisis. One of the sixty-two initiatives was the use of an outside vendor to help catch up on overdue Medicaid redeterminations of eligibility. This process was championed by Republican members of the general assembly, and it embodied not only the need to save funds but also their antipathy to public employee unions, their preference for outsourcing, and their philosophical objections to Medicaid.
The resulting program has spawned a number of commentaries lauding the ability of these kinds of initiatives to generate large budget savings, including an article in the Wall Street Journal (Matthews 2014; Ingram 2013; Foundation for Government Accountability 2014). Some of these commentaries garble the basic facts, and even those that don't fail to appreciate a number of key contextual factors. Those factors have significant Medicaid policy implications.
The specific background of this program is a matter of record. In 2012 Illinois Medicaid was not doing a particularly good job of carrying out Medicaid redeterminations. Performance was never as bad as critics claimed, but it was insufficient and it was a contributor to rising Medicaid costs.
From 2003 to 2007, Medicaid experienced persistent disenrollment of about 1.5 percent of its covered population each month, or about 18 percent over the course of a year.1 Subsequently, three factors converged:
■ The Great Recession severely damaged family economies, particularly those in the lower part of the income spectrum, causing people to stay on Medicaid longer.
■ The Illinois General Assembly and the governor, in dealing with previous budget problems, let the number of caseworkers who determine eligibility dwindle from twenty-eight hundred at the beginning of that period to about eighteen hundred by the end of 2012, despite a huge increase in eligible clients over the same period.
■ Illinois chose to allow portions of the Medicaid population, primarily children in homes with income below 200 percent of the federal poverty level, to be “passively” redetermined. That meant that clients were sent a letter asking if their circumstances had changed. Even if there was no response, the children on the case were continued—although in the absence of a response adults were disenrolled. (Many, in fact, did respond and were redetermined accordingly.)2
The combination of these factors caused the average persistent disenrollment rate to slide from 1.5 percent to about 1 percent per month by 2012. So in addressing the 2012 Medicaid budget crisis, the general assembly mandated the use of an outside vendor to redetermine cases. However, as often happens, the general assembly had not worked through the operational details. Among other things, the law did not take into account that federal regulations require the use of state employees to make the final eligibility determinations or that a large number of Medicaid cases were redetermined at the same time as eligibility for the Supplemental Nutrition Assistance Program (SNAP), which resulted in a different processing flow for these cases.
Illinois Medicaid Redetermination Project
The law required that a vendor be selected and a contract signed within ninety days. Maximus, a national firm based in Virginia, was selected and proceeded to create a system for reviewing Medicaid cases that did not include SNAP. It had proved virtually impossible to coordinate with an outside vendor the multiple regulatory and technical constraints on simultaneously redetermining Medicaid and SNAP. Moreover, the data were reasonably clear that SNAP redeterminations were being done on a timely basis, and, since Medicaid was part of the process, these cases were generally getting redetermined appropriately. The major concerns were those cases for which the only benefit was Medicaid.
As the specifics of the process were developed, it also became clear that, despite many claims to the contrary, Maximus did not have materially more access to additional databases than the state had. Indeed, virtually all the data the firm used to make recommendations were data previously available to the state and were simply routed to Maximus for its use.
The process of reviewing was arranged so that Maximus ran all cases through the databases and then prioritized the order of its detailed review of cases according to the likelihood that a case, based on available data, seemed likely to be ineligible. If the databases did not provide sufficient information to make a recommendation, Maximus requested that the client provide additional information.
Predictably, this process led to Maximus recommending a large number of cases to the state for cancellation. The vast majority of those cases were recommended for cancellation not because of any specific information showing the cases ineligible but because of the clients' failure to answer the request for additional information in the very tight frame written into the law. No doubt, some people did not respond because they were ineligible. But a large number did not respond because they did not receive the notice, did not receive it in time, couldn't read the notice, or were otherwise simply incapable of response.
For about 40 percent of those cases that Maximus recommended for disenrollment, the clients subsequently provided information to state caseworkers before the cases were cancelled. Of those actually canceled, about one-third were reinstated in the next three months when clients provided information verifying eligibility.
Nevertheless, this process resulted in the removal of a substantial number of clients from the Medicaid rolls. How this situation actually relates to budgetary saving is harder to quantify, although it seems that certain savings were never as much as legislative sponsors had predicted.3 Among other things, in the first six months of this project, roughly half the people removed from the rolls had not used any Medicaid services in the previous six months. Even those removed clients who had used services had used markedly fewer than those who stayed.
This finding is not surprising. In Illinois Medicaid, a small number of clients account for the vast bulk of the costs. These clients are very sick and, almost without exception, are in fact eligible. Given the size of the skew in the data, the overall average costs are particularly unreliable guides to potential savings. Neither is it surprising that those who most needed health care were the ones who took the most pains to respond.4
There is also the question of the administrative cost. The American Federation of State, County and Municipal Employees (AFSCME) filed a grievance that the use of Maximus violated the existing collective bargaining act, and the arbitrator agreed. Part of the reasoning behind the arbitrator's decision was that the state could not prove that Maximus was a more cost-effective alternative than hiring state workers to do the same thing, particularly since Maximus was using the same data and the caseworkers still had to work the cases to some degree, creating a clear duplication of effort.
All of this is not to say that no money was saved or that employing Maximus was a waste. Even after netting out Maximus's cost (and the cost of the necessary additional caseworkers to actually process their recommendations), the overall rate of cancellations increased sufficiently to show some net savings.
Much more important in the larger scheme of things is that the Maximus contract precipitated major changes in the processing of redeterminations by the state, some of which would have eventually happened, but this certainly accelerated the process. The two most important changes were as follows:
■ The creation of two large hubs specifically devoted to processing Medicaid redeterminations for cases not participating in SNAP. These were the cases that had been receiving the least scrutiny before the Illinois Medicaid Redetermination Project (IMRP) was launched. These were staffed with new caseworkers, the first net increase in caseworker count in more than five years.
■ The realization that better call center capabilities were required, leading to the creation of the first state call center of scale to support Medicaid (and SNAP and cash assistance) in the fall of 2013. Similarly, the Maximus mail room was a useful example of the efficiencies of a centralized, technology-enabled mail center—something that the state lacked but is clearly necessary.
In general, Maximus brought a focus to one sector of redeterminations that been slighted and a process technology that the state's antiquated computer system could not match. (Lack of technology was another result of insufficient investment in the eligibility process. The state was still using a thirty-five-year-old COBOL-based mainframe. Most of the computers on caseworkers' desks were insufficient to open any but the smallest portable document format [PDF] files.) The new eligibility system that the state had already started would incorporate those features, but Maximus created a head start.5
Illinois Medicaid Redetermination Project—“Hybrid” Version
To return to the arbitration of the AFSCME-filed grievance, the state disagreed with the arbitrator's decision that the contract with Maximus was a violation of the collective bargaining agreement but made the decision that eligibility integrity was better served by negotiating a settlement rather than having the issues tied up in court and facing ongoing guerrilla activities from AFSCME. Moreover, an adverse court outcome would have upheld the arbitrator's order to simply end the Maximus contract, which would have had even worse consequences.
Therefore, the state negotiated with Maximus and with AFSCME to develop a “hybrid” solution that met the terms of the original legislation but removed the extra level of review by Maximus workers. Maximus continued to provide process technology for managing the flow of cases and maintained its call center and mail room. The new agreement reduced the payment to Maximus by roughly 50 percent but did not hinder the redetermination process at the newly established hubs.
Under this negotiated arrangement, the state continued to remove people from the rolls at about the same level as before. The results of both phases of the IMRP have been a temporary spike in the number of clients being canceled and a return to an underlying persistent cancellation rate slightly above what was seen prior to 2007—about 1.6 percent per month.
Despite the rhetoric to the contrary, the reason for cancellation is rarely fraud. For instance, in the most recent quarter with available data (quarter 2 of 2015), the information Maximus collated from the electronically available databases showed reason to suggest that the client was ineligible for less than 8 percent of the cases. And about one-sixth of those subsequently provided qualifying information—such as showing that the electronic data were from a period prior to job loss before any disenrollment. Far and away, the most common reason was, still, that the client failed to return the required information. More than 80 percent of all cancellations are caused by failure to return the information required. And for the 2015 fiscal year, more than one-third of the people canceled were reinstated in the succeeding three months when they provided the information.6 Others are reinstated over a longer period. It is always possible that people fail to return information because they know they are not eligible. And, clearly, caseworkers reviewing data collected by Maximus and from clients canceled cases for other reasons. But none of this supports a pattern of wide-scale fraud. In fact, the number of cases referred for fraud investigation is negligible. Rather, this situation reflects ongoing fluctuations in income among clients in this group combined with their material difficulties in managing the administrative tasks of providing timely information to caseworkers.
Although simple conclusions are elusive, there are several observations to be drawn from this experience.
Role of Privatization
The broadest generalization is that privatization of the sort practiced in this example is not a magic bullet. But it does have its uses.
First, the vendors did not have access to special or appreciably different data than the state had. In the Illinois case, Maximus did develop a better process for managing the case flow than the state had.7 Had the state not already been in the process of building its own system, purchasing the Maximus system would have been useful.
But the real payoff in Illinois was from segregating “Medicaid-only” clients and adding caseworkers to the redetermination process. Whether those caseworkers were provided by the state or by Maximus did not seem to make much difference. Generally speaking, state caseworkers had stronger backgrounds (few Maximus workers could meet the standards required for state caseworkers) and received more rigorous training. The cost between adding state caseworkers and hiring Maximus was roughly equivalent. Maximus's eligibility specialists were paid a lot less than state caseworkers, but when the entire costs of hiring the outside firm were compared with the costs of hiring caseworkers (including benefits and support costs), the differences were not significant. The bottom line is that in the current environment “boots on the ground” are required. If a sufficient number of caseworkers are not authorized, then some things don't get done.
Yet Maximus clearly has attributes that are harder for the state to replicate. It was able to staff up and open its facilities more quickly than the state would ever be able to do. Not only was it able to work outside all state procurement and personnel rules, but it had templates for quick action. It had a cadre of technical people who could be immediately focused and dedicated to working on the issues without distraction from their “day jobs.” That Maximus had much more flexibility than the state with regard to employees affected both the firm's ability to remove unproductive workers and its ability to rearrange business processes without having to enter into time-consuming negotiations about job descriptions.8
Eligibility versus Ineligibility
There can be no dispute that in Illinois a large number of people were removed from the Medicaid rolls—between April 1, 2014, and March 31, 2015, more than 25 percent of all Medicaid clients were disenrolled. This amounted to the removal of more than 800,000 individuals, although 288,000 (36 percent) of these returned in less than three months (HFS 2015). But whether this is a good thing or a bad thing is much more difficult to determine.
For openers, there is the question of public trust in Medicaid. Too many conversations about the proper role of Medicaid get sidetracked by the assumption that many people are on the rolls fraudulently. This view is not unrelated to the general lack of trust in government. From this perspective, an effective and fair redetermination process is an unalloyed good thing. After all, it is clear that there are many ineligible people on Medicaid.
However, here is where things start to get murky. No one really knows how many of the people removed from Illinois Medicaid were in fact ineligible, let alone how good are the processes for separating the eligible from the ineligible. Available evidence suggests that at least half the people removed were actually eligible, perhaps a lot more than half. The process identified people less on the basis of actual eligibility and more on their degree of proficiency at receiving and returning pieces of paper. There are many reasons why poor people are not good at complying with this request even if they are totally eligible for Medicaid.
One reason has to do with the lack of appreciation that many Medicaid clients have for the idea of Medicaid as ongoing coverage for health care. Rather, they see Medicaid, and perhaps all health insurance, simply as a mechanism for reimbursing health care expenses when necessary. Consequently, they do not avail themselves of preventative services, do not proactively seek health care, and, to the point at hand, do not worry about Medicaid eligibility until they are sick. At that point, the system—the physician's office, the hospital emergency room, the federally qualified heath center—intervenes to smooth the path to getting coverage and the claim “taken care of.”
And, of course, there are good reasons for a Medicaid program to reduce the number of ineligible clients in the current structure of American health care. Budgets are tight, and payments made to people who are not eligible reduce funds available to those who are eligible. In Illinois, some of this issue took care of itself under the old fee-for-service system. When clients moved out of state or got other insurance, they simply stopped using Medicaid and therefore stopped incurring costs. However, under the coordinated care approach that Illinois (like most other states) has adopted, Medicaid continues to make capitation payments for clients who are not actively removed from the rolls—even if they use no services. No Medicaid program wants to pay a managed care entity for a client who is no longer receiving its services.
Redetermination is a form of testing, and as with any other testing there are trade-offs between sensitivity and specificity. A state can dramatically reduce its Medicaid rolls by making the redetermination process particularly onerous, but it is not likely to be particularly fair. Or it can make it easier to reestablish eligibility, with the likely consequence of paying for people who are not in fact eligible.
It is worth mentioning that determining eligibility—either initially or at redetermination—is not an exact science. Over the years, eligibility requirements have become endlessly complicated. The ACA's introduction of the use of modified adjusted gross income (MAGI) has increased the complexity. Just as different Internal Revenue Service agents can give different answers to complicated tax questions, Medicaid eligibility decisions are likewise variable.
In any event, the large number of people determined ineligible by Medicaid programs creates a substantial source of churn. Most of these will not show up as churn between Medicaid and the marketplace because the client is not automatically referred to the marketplace if the reason for disenrollment is “lack of cooperation”—that is, failure to return information. Clients can still apply to the marketplace, but for the very reason that many failed to return the information that would have shown their Medicaid eligibility, they are not likely to apply to the marketplace. Typically, they will simply drop into the ranks of the uninsured until circumstances impel them (or their health care provider) to reestablish eligibility. This system is probably not the worst imaginable—payments to managed care entities are minimized and health care can be made available when individuals are sick. But it does fly in the face of attempts to create a culture of health care coverage that will, over a longer period, improve health status.
Suggestions for Action
There are questions to be raised about the social value of using means-determined eligibility for various classes of health care, particularly as the changing health care financing landscape has reduced the amount of difference in health care coverage available to citizens.9 But in the current polarized environment it is hard to imagine much likelihood of a reasonable consideration of this broader issue. Accordingly, it makes sense to address more modest proposals that will reduce the amount of unnecessary churn. Cycling eligible people off and on Medicaid is not a great use of administrative resources and is incompatible with the more continuous vision of population-based health care.
The most important step to achieve a responsible balance in redetermination is through a process that places as few demands as possible on the clients but also weeds out clients with a high likelihood of being ineligible. Hallmarks of such a process would include the following:
■ There would be extensive use of available electronic information to detect circumstances very likely to lead to ineligibility, such as an out-of-state address or earned income apparently in excess of eligibility limits. Clients should, of course, be allowed to dispute such data, but evidence of this sort would shift the burden of proof.
■ However, for clients with no electronic evidence of likely ineligibility, the default option should be to continue eligibility. Clients should be clearly warned that failure to report changes in circumstances could be a basis not only for ineligibility but even for criminal prosecution.
■ For all circumstances, client communication with the responsible state agency must be relatively easy. This includes the ability to communicate with the eligibility entity in person, by phone or mail, or electronically.10 A sufficient number of caseworkers is a sine qua non.
■ Some periodic validation of the electronic databases must be made by actual audits. People who commit out-and-out fraud will be hard to detect. While these are the people society would most want to catch, if they have made sufficient effort to keep information off electronic databases, they will be hard to find. In any event, the number of people in this category is unlikely large. More typical will be people who have lost eligibility between redeterminations but do not report in a timely manner. A state can decide how important it is to obtain timely reporting; some states will opt for a policy to assume twelve months' continuous eligibility as a way of avoiding unnecessary churn and improving coordination with the marketplace. After all, in most cases, the client would then become eligible for the marketplace.
A second useful step will be to fully involve managed care entities in maintaining people's eligibility as part of their ongoing responsibility in a continuing and coordinated process of achieving and maintaining health. The coordinated care entities are very aware of this issue and are expecting Medicaid to work with them to help keep eligible clients enrolled continuously. Obviously, this effort is not purely altruistic; they have significant financial stakes in maintaining continued cash flow. Nevertheless, it is consistent with the underlying goals of coordinated care. Reducing inappropriate eligibility churn and maintaining continuous, coordinated care is a win for all involved. This dimension, however, underscores the need for a good enough surveillance function to identify people obviously not eligible.
In short, developing appropriate Medicaid redetermination policies is not simple—let alone easy. It is an ongoing balancing effort among a number of different imperatives in the strangely fragmented American health care system. The example of Illinois clearly outlines some of the dilemmas involved.
1. In this period, initial disenrollment was in the 1.8–2 percent range, but about 25 percent of the disenrolled were reinstated within three months, for a persistent disenrollment rate of about 1.5 percent.
2. Although there were always legislative complaints about this process, it was strongly supported by the federal government; in fact, Illinois received State Children's Health Insurance Program (SCHIP) bonus funds for this practice.
3. Proponents of the legislation claimed that it would save $350 million, although they did not specify a methodology. The Medicaid agency never accepted this figure and estimated potential savings of $125 million. Estimates of actual savings differ—it is always hard to agree on how much wasn't spent that might otherwise have been spent. The most detailed internal estimate made by Medicaid is that the program avoided $115 million in expenses in fiscal year 2014, but at a cost of $45 million, for a net savings of $70 million. Earlier internal estimates had suggested slightly larger savings but nowhere near the $350 million target. This has not kept various legislators and others who supported the program from claiming savings at the $350 million level (Foundation for Government Accountability 2014).
4. Estimating savings is also complicated because during this period Illinois was dramatically increasing the amount of care reimbursed via capitations. The savings proposition from removing people from the rolls is very different for capitated environments. This issue is addressed later in this article, but it should be pointed out that the savings going forward (i.e., in a largely capitated environment) would be greater than immediately after enactment (i.e., in a still fee-for-service environment).
5. Illinois Medicaid was taking advantage of the ACA to replace its ancient eligibility system. It had started in 2010 and was well under way by spring 2012 when this legislation was passed. The legislation, in fact, noted that the outside vendor was only a “bridge” to the new system, although not all legislators accepted that notion. The new system, which is being implemented in phases, is on schedule for an early 2016 completion.
6. Medicaid posts statistics on the redetermination process each quarter on the Illinois Department of Healthcare and Family Services website (HFS, n.d.).
7. Maximus did not have a “package” system that it could immediately use. Because of the very short time frame written into the law to get the system up and running, it felt compelled to use a system from another firm to manage the case flow. The arrangement did not work well. It was subsequently patched to acceptability and then, about nine months later, completely replaced with a much better system.
8. People could reasonably question how much of the flexibility that Maximus exercised is socially desirable. But there is no question that from a management standpoint Maximus is able to achieve desirable ends in much less time than would be necessary within the framework of state policies and procedures. This proficiency is particularly crucial in an environment where new systems are being introduced and things are changing rapidly.
9. The overall government subsidy for a person going from Medicaid to the marketplace is most likely not that different, so from one perspective the question is mostly about what percentage of the subsidy is picked up by the federal government versus the state.
10. Historically, at least in Illinois, it was fairly difficult to disenroll oneself from Medicaid because of the difficulty of reaching caseworkers. It was much, much easier to simply stop using Medicaid, under the assumption that the system would eventually catch up. Sometimes it did, sometimes it did not. But as noted elsewhere, when capitated payments come to predominate, keeping the rolls current becomes much more important.