Data-Driven Strategies to Improve Transitions Between Medicaid and Exchange Coverage

February 28, 2018

Julie Silas of Social Interest Solutions

Mining Data for Insights

Collaborating and Coordinating Across Agencies
Correlating Eligibility Status to Other Characteristics Comparing Data Changes Over Time
Identifying Missing or Conflicting Data and Unexpected Results

Introduction

The Affordable Care Act (ACA) offered more health insurance options to low- and moderate-income individuals and families by expanding Medicaid and creating more affordable coverage options through health insurance exchanges. Because the options available to individuals or families depend on factors that can change over time — their income, age, family size, and residence — people may need to transition between different health coverage options as their circumstances change. While the intent of the ACA is for these transitions to happen seamlessly, people are not always able to transition smoothly and sometimes people experience gaps in coverage.

Transitions between Medicaid and exchange coverage can be particularly challenging because typically separate agencies are responsible for the sources of coverage. In most states, the federal government runs the exchange, while state agencies oversee the Medicaid program. Even in states that have established state-based exchanges, the exchanges are not usually run by the same agency that runs the Medicaid program. As a result, the systems and processes that support the two sources of coverage may be bifurcated, leading to inefficiencies and confusion at transition points for consumers and administrators alike.(1)

Notwithstanding these challenges, states can adopt strategies to improve transitions for individuals moving between Medicaid and exchanges. One such strategy is to make better use of data to gain insights about how people transition — or fail to transition — between health coverage programs. States already have a wealth of data that they collect and create about the people they serve. Even without significant investment, states can use the data at their disposal to discern patterns to better understand what happens during transitions, why coverage disruptions happen, and what can be done to promote seamless transitions.

The first step in developing a data-driven strategy is to identify relevant data which the state already has. The data may inform their understanding of the challenges the transition population faces. The second step is to dig into the data, often from disparate sources. By analyzing the data, the state may be able to identify gaps, patterns, and inconsistencies that can provide insights into the nature of transition challenges, inform approaches to address identified problems, and provide metrics for assessing improvement.

Identifying Relevant Data

Medicaid and exchange agencies already collect extensive data from people applying for or renewing benefits. At a minimum, they collect detailed information required to determine eligibility, such as address, date of birth, gender, immigration status, family size, and income. States also collect optional data for the purpose of serving their customers better, such as information about their race, ethnicity, written and spoken language preferences, preferred method of contact, and how they were referred to the programs.

In addition to data that agencies collect from people, they create and capture data about people’s experiences as they move through the application, enrollment, and renewal processes. For example, agencies often know how a person applied (e.g., online or through a paper application); whether an agent, broker, or other assister was involved; the status of the application, and the frequency and mode of contact between the individual and the agency. Also collected are various time-sensitive actions, such as when the application was completed, documents were due and submitted, eligibility was established, plan selection was due and completed, coverage was effectuated or terminated, and/or renewal was due and completed.

Agencies routinely rely on data to produce budgets, determine staffing needs, claim federal matching funds, and comply with reporting requirements.(2) The data in these reports, and the underlying data compiled to produce the reports, can be highly useful in identifying and describing the transition population.

For an in-depth understanding of the transition population and insights into their experiences as they move through the transition process, states will likely need to enhance the routine data sources with a broader and richer mix of data analyses. States can likely meet the need without collecting additional information, but instead by making more creative use of what they already collect. States can use these data to identify specific challenges and barriers that may disproportionately impact different sub-groups within the transition population. While data alone likely will not allow the state to completely diagnose the causes of transition challenges, they will provide agencies with clues as to where they could investigate further.

The following questions may help states begin to identify relevant sources of data:

What data does the state capture when customers interact with their programs, such as through online portals and accounts, call centers, and in-person visits?
Is the information collected for all customers, including those who do not ultimately apply for or receive benefits and/or for those whose benefits have ended?
What data does the state share with other agencies and third parties, such as sharing of data between Medicaid and exchange agencies, with health plans, and with verification data sources?
Do the exchanged data include information about people transitioning between Medicaid and exchange coverage?
What information relevant to the transition population is captured in the state’s many information systems, including systems that support call centers, eligibility workers, case management, help desks, claims processing, health plan management, grievances, and appeals? For example, which of these systems include information about how and when people initiate, terminate, and reinstate benefits through Medicaid and/or exchange coverage?
What units or staff in each agency are responsible for data monitoring and reporting, including for purposes of program planning, budgeting, program integrity, and oversight?
What insights do they have about data available to the state that is relevant to the transition population and their experiences?

Mining Data for Insights

Once a state has identified data relevant to the transition population, it can analyze the data to discern patterns and explore the need for changes in policy, business processes, and/or technology to make it easier for people to transition between Medicaid and exchange coverage. For example, states may discover that a significant share of the transition population have frequent income changes that are causing them to churn between Medicaid and exchange coverage, resulting in high administrative costs, multiple opportunities for gaps in coverage, challenges in coordinating benefits, under and overpayment of premiums, and similar concerns. A review of data may lead states to explore opportunities to adjust their policies regarding continuous eligibility, reporting requirements for changes in circumstances, or other eligibility standards. The analyses may help states identify particular populations who are likely to need assistance through the transition process and employ strategies to assist them. For example, states may find that people who prefer written or spoken information in languages other than English are less likely to successfully transition between Medicaid and exchange coverage. As a result, states may devise in-language outreach strategies to address their needs.

States are increasingly investing in statistical and mathematical analysis of data sets to more easily and routinely learn from the stories their data can tell. With or without those advanced capabilities in place, however, there are many near-term strategies states can use to get the most from their data, including the data they have identified as relevant to the transition population. Four strategies that agencies can consider are discussed below.

1. Collaborating and Coordinating Across Agencies

As noted above, one of the biggest challenges that prevents successful transition between Medicaid and exchange coverage is the fact that the programs are not fully integrated under one agency — and neither are the data. However, if the two agencies coordinate with one another — including sharing data across systems regularly, systematically, and frequently — together they can overcome some of the obstacles of having two different systems. Suggestions to improve collaboration and coordination that can be put in place include:

Develop a shared data dictionary so that both agencies have a common understanding of the data elements they each collect, making it easier to compare, track, and analyze data over time.
Update the data dictionary as the data sharing evolves, to capture new elements as staff learn more through the data analysis process.
Develop a common understanding and agreements about data-sharing practices, which can help ensure that data sets are complete and comprehensive, are shared regularly, and cover a consistent period of time.
Track data sources across the two agencies to illuminate where problems may be arising.
Establish a common client identifier that is shared between the state-based exchange (if applicable) and Medicaid.
Align different data elements so that they are measuring similar attributes to be able to make fair comparisons.

AN ILLUSTRATION: Incongruous FPL Brackets. Eligibility determinations for Medicaid and subsidized exchange coverage are based on slightly different annual federal poverty levels (FPLs). Medicaid uses the FPL in effect at the time the person’s eligibility is being determined, while exchanges use the FPL in place at the time of open enrollment (which is usually the year prior to the coverage year). For example, a person applying in June 2018 will be determined eligible or ineligible for Medicaid based on the 2018 FPL guidelines, but will be determined eligible or ineligible for subsidized exchange coverage based on the 2017 FPL guidelines. Medicaid also uses different FPL thresholds, depending on various eligibility characteristics (e.g., states generally have different eligibility thresholds for children, pregnant women, parents, and childless adults), while exchanges have uniform FPL thresholds for all applicants. Mid-calendar year, when a person is renewing or experiencing a change in circumstances, these differences in Medicaid and exchange FPL thresholds may impact eligibility. If the data for FPL thresholds are collected in different formats for each program — that is, if they are tracked by program eligibility standards that are not comparable across the two programs — it can be very difficult for staff to ensure people aren’t falling through the cracks (mistakenly determined ineligible for either program) or paying exchange premiums when they are actually eligible for the lower cost Medicaid program. While the two entities may have different FPL thresholds, they can still configure and report data into their systems by FPL brackets that are comparable, making it easier to analyze and pinpoint inaccurate eligibility determinations.

2. Correlating Eligibility Status to Other Characteristics

Examining available data about eligibility status changes over time can help identify transition challenges and how they impact specific populations. Once all federal law requirements are met, including ensuring that all categories of Medicaid eligibility are considered and that beneficiaries are provided an opportunity to confirm findings before a letter is sent to discontinue coverage, people can transition between Medicaid and the exchange. Once people rightly transition, they may have circumstances that result in multiple transitions in a very short time period. For example at annual renewal, a Medicaid agency may determine, based on an ex-parte review, that a person’s income has changed sufficiently to make them ineligible for Medicaid and eligible for subsidized exchange coverage. They are sent a letter informing them that they are now exchange eligible, providing them an opportunity to confirm findings, and then encouraging them to set up an exchange account and pick and pay for a plan, which they do. Under this scenario, three months later they may report to the exchange that they have lost their source of income, making them likely eligible for Medicaid once again.

Analyzing available data about subsets of the transition population based on their initial eligibility status and changes in their status over time may help states identify how characteristics of the transition population may affect their likelihood of successfully transitioning between the two programs. While the correlations will not themselves pinpoint the root causes of transition challenges, data will help states diagnose problems and develop strategies to address them.

States may discover linkages between those unable to complete a transition and such factors as race, ethnicity, spoken and written language preferences, geography, age, gender, and other demographic factors. By exploring such linkages, agencies may gain insights such as:

Delineate patterns that indicate when certain groups are easily making it through the system while others are not.
Identify if certain populations are disproportionately impacted by the differences between program rules of Medicaid and exchange coverage, for example self-employed individuals with fluctuating income may be inadvertently transitioning between programs because Medicaid measures income based on the current federal poverty level (FPL) and exchanges use FPL levels in effect during open enrollment based on annual income to determine eligibility.
Analyze changes in status that occur within a specific time period to determine if looking at shorter periods of time might provide greater insights into whether the transition process is working as intended.

AN ILLUSTRATION: Demographic Differences. Each Medicaid agency identifies a group of individuals at renewal whose circumstances have changed — often because of changes in family size or household income — making them no longer eligible for Medicaid, but rather eligible for exchange coverage. Unlike Medicaid, however, exchange coverage does not begin until someone picks a health plan and pays their first month’s premium, all of which is supposed to happen within 60 days.(3) To explore linkages between eligibility status and demographic factors, state Medicaid agencies can construct monthly data cohorts — aggregate data about all of the people in a month who have been determined ineligible for Medicaid and eligible (or potentially eligible) for exchange coverage — and share the data with the exchange. The exchange can then match the Medicaid data with the exchange’s own eligibility status data by using common identifiers, probabilistic matching, or other techniques as appropriate. By reviewing demographic factors within and across these cohorts, the exchange can see which groups transitioned seamlessly to exchange coverage (i.e., they selected plans and paid premiums) and compare them to the group of people who have not. The exchange can further analyze the group that was unable to transition within the 60 days and find patterns in the demographic data of the group to help target their outreach efforts. Perhaps those failing to pay their premiums are individuals and families in a low-income bracket. Perhaps they are families who speak English, but who expressed a preference for written notices in other languages. Perhaps they are mostly children under 18, whose parents are already enrolled in exchange coverage. Discerning demographic differences between those who successfully transition within 60 days and those who do not can help states develop targeted outreach and education strategies.

3. Comparing Data Changes Over Time

Enrollment reports that track both demographics and unique individual’s status can contain hundreds of thousands of records. Reports can be designed to track each interaction each person has in the system, but can also store and report duplicate records that indicate the person is still in the system with no changes (a comprehensive data set that contains every single record of all the interactions in the system allows staff to review the entire experience of those transitioning between the two programs). However, agencies can also benefit from configuring data sets into discrete components that can facilitate gaining insights — and help differentiate static data from data changing over time. It can also help states develop a baseline that can be used to confirm assumptions that states may make regarding challenges that the transition population may face.

For example, in addition to the full data set, staff can:

Refine data to only look at unique individuals within the system, which can narrow hundreds of thousands of records into a few hundred thousand and hide the duplicate records where no changes have occurred during the period of time in question.
Restrict a small data set to an even smaller one, narrowing the review to encompass the last available status for each individual in the system at a given point in time, which will allow staff to pinpoint each individual’s last interaction in the system and their most recent eligibility status.

Staff can use the different data sets to set a baseline; that is, to establish an understanding of the overall general transition population, which will allow them to make more accurate and meaningful comparisons when looking at the more restricted data sets. The ability to reconfigure the comprehensive data set into discrete components can enable staff to discern problems with the data — allowing for comparisons across data sets and using those comparisons to identify things like mathematical errors and problems with calculations (often notable when one data element is built from other data elements). Data set comparisons also can confirm or contradict assumptions and surface changes across reports (including where changes should not be seen on static data elements like race or gender), which can indicate that the data might be compromised.

AN ILLUSTRATION: Changes Mid-Reporting Period. In most states, the Medicaid eligibility threshold for children and young adults changes between 18 and 19 years of age. In a state where the eligibility threshold drops dramatically when a person is no longer a child, there may be a large number of 19 year olds in the transition population unfamiliar with the need to pick and pay for a plan in order to get exchange coverage. A large monthly data set with multiple records for each unique individual in the transition population will include multiple records for those changing from 18 to 19 years old during the month. A data set with multiple records for the same individual will be harder for the agency to analyze how the age change may be impacting those transitioning. If the agency configures a data set that only looks at unique individuals and only those individuals who had an actual change in their data during the reporting period, the data will show the agency whether the change in eligibility threshold has resulted in a higher proportion of 19 year olds transitioning from Medicaid to the exchange; information that can be used to consider policy changes or customize outreach and education efforts.

4. Identifying Missing or Conflicting Data and Unexpected Results

A valuable piece of analysis of any comprehensive set of data will look at missing, confusing, or unexpected data. Missing or unexpected information can lead to additional research outside the data report to recognize the source of the problem and surface fundamental issues with the eligibility and enrollment system. When analyzing the data itself, agencies may find that the information that is missing or confusing is not hard to identify; when organized and sorted, the data will show values that are missing or that do not make sense. For example, states may discover that there is missing or blank data for the majority of records, such as income or household size information. Upon further investigation, states may discover that their user interface or other data collection mechanisms are not prompting collection of the information, that a database is not capturing information properly, that the queries they use to generate data are faulty, that there are problems with data entry or in the data collection mechanisms, that the system is improperly deleting or overriding the data, or any of a number of other potential challenges.

Noting the missing or faulty data can lead staff to ask further questions such as:

Are the issues related to the raw data or how they were collected and reported?
Are the data confusing because of the underlying data query?
Are the data missing because required data elements were coded into the system as “optional”?
Are users entering data inaccurately?
Does the data reflect work arounds, where users are entering default data to skip required questions?

As staff uncover the source of the data issues, they can initiate system, process or policy solutions that can be put in place to ensure data integrity and improve transitions.

AN ILLUSTRATION: Looking Beyond the Report. When reports produce unexpected results that cause confusion among program experts, it may be time to look beyond the aggregate data and focus on individual cases to recreate the consumer journey across time and case management systems. Agencies can pick several cases from the report and check across different elements within the case to understand how key eligibility factors such as income and household size change over time. They can investigate how the system or systems assigned eligibility and how that eligibility was communicated to beneficiaries via notices. Any unexpected results can be investigated further to identify system, process or policy issues that are driving any transition issues.

Conclusion

For people whose eligibility for health coverage changes due to circumstances such as income and household makeup, seamless transitions between Medicaid and exchange-based coverage can be challenging. State agencies looking to reduce gaps in coverage have powerful tools at their disposal in the data sets they collect, track, and evaluate. Without collecting more data than is minimally necessary, agencies can become more expert about the populations applying for benefits by performing advanced data analysis. These strategies include collaborating and coordinating data sharing across agencies; correlating eligibility status to other characteristics; comparing data changes over time; and uncovering missing or conflicting data and unexpected results. Examining more closely the individual data collected, analyzing it not only through the lens of demographics, but also by tracking how individuals and families progress through the system, agencies can identify barriers to coverage and surface potential changes that could be made to better serve those in need.

Endnotes

For example, people transitioning between Medicaid and exchange coverage may get separate notices from each program with inconsistent information about next steps and lack of clarity about who to contact to help resolve the inconsistencies.
For example, states work with the Centers for Medicare and Medicaid Services (CMS) on the Transformed Medicaid Statistical Information System (T-MSIS), which includes enhanced information about beneficiary eligibility. Available at www.medicaid.gov.
45 C.F.R. 155.420(c).

Julie Silas is the Director, Policy Innovation, at Social Interest Solutions, where she works on issues at the intersection of policy and technology. In this role, Julie provides policy research, analysis, technical assistance, and thought leadership. She is the primary author of issue briefs, white papers, and consumer fact sheets on a variety of health and human services programs, including free or reduced-price meal programs, SNAP, Medicaid, and other health insurance affordability programs.

Acknowledgement
This report was produced with the support of the California Health Care Foundation (CHCF).

CHCF is dedicated to advancing meaningful, measurable improvements in the way the health care delivery system provides care to the people of California, particularly those with low incomes and those whose needs are not well served by the status quo. We work to ensure that people have access to the care they need, when they need it, at a price they can afford. CHCF informs policymakers and industry leaders, invests in ideas and innovations, and connects with changemakers to create a more responsive, patient-centered health care system.

For more information, visit www.chcf.org.

About Social Interest Solutions
Social Interest Solutions (SIS) is a national nonprofit dedicated to improving access to quality health and social services through policy and technology solutions. We work with federal, state, and local agencies, service providers, community-based organizations, and researchers to better connect millions of low-income Americans to health care, nutrition, and other programs.