CHIPRA Mandated Evaluation of the Children's Health Insurance Program: Final Findings. Duration of Coverage


Children typically stay enrolled longer in Medicaid than in CHIP.

The median time that children remained enrolled in coverage varied considerably by type of program. Using data on the 10 study states together, new Medicaid spells were estimated to last about twice as long as new spells in CHIP (26 months for Medicaid versus 13 months for Medicaid expansion CHIP and 14 months for separate CHIP).43 These pooled measures of duration remained fairly stable over the study period, with a slight decline in the median length of Medicaid spells toward the end of the period.44

Duration of coverage varied across states, particularly among state Medicaid programs.

The overall trend of longer duration in Medicaid than in CHIP was consistent for all 10 states except California, where median duration in the separate CHIP—at 22 months—was higher than median duration in Medicaid (at 15 months, one of the two lowest Medicaid durations in the sample) (Figure VI.1). In most separate CHIP programs, median duration ranged from 12 to 14 months. Although California did not introduce automated renewal methods to the extent of other states during the study period, it took other steps to support renewal in its separate CHIP, including mailing preprinted renewal forms, sending reminder cards, and making telephone calls to nonresponders. The shorter median duration observed in Louisiana’s small separate CHIP for children with incomes between 200 and 250 percent of FPL is consistent with the expectation that this component provides a temporary bridge between Medicaid/CHIP and private coverage.

We see larger differences in the median duration for Medicaid programs, from lows of 15 months in California and Utah and 17 months in Texas to a high of 44 months in Louisiana. The higher Medicaid duration in Louisiana’s Medicaid and Medicaid expansion programs likely reflects the state’s use of passive renewal and other administrative processes that keep most children enrolled unless the family reports changed circumstances. Louisiana had adopted several simplifications that, in combination, largely eliminated the need for families to fill out renewal forms (Kellenberg, Duchon and Ellis, 2010). In contrast, the Texas and Utah Medicaid programs did not offer 12-month continuous coverage and required redeterminations more frequently than once every 12 months during the study period. The lower duration of enrollment in California’s Medicaid program is likely indicative of the lack of automated renewal processes and a decentralized approach that relied on county department of social service offices to process renewals (Hill et al., 2013).

Figure VI.1. Median Duration of New Coverage Spells, by State and Program Type

Figure VI.1. Median Duration of New Coverage Spells, by State and Program Type

Source: Mathematica analysis of state-provided Medicaid/CHIP administrative data.

Note: Duration measures are calculated by using data on spell length for new enrollment spells starting in the period August 2007 to July 2012. The numbers reported in this figure take a simple average of median duration values across months. Only months where the end data for at least 50 percent of spells can be observed in the data for each state are included.
The median duration of public coverage in Louisiana is on average at least 60 months; most new spells ended after July 2012, so the end date was censored for more than 50 percent of these spells throughout the time period.
M-CHIP = Medicaid expansion CHIP; Public = Medicaid and CHIP combined; S-CHIP = separate CHIP; Combination CHIP = state administers both M-CHIP and S-CHIP.

The longer duration of public coverage in Louisiana and Ohio (on average at least 60 months in Louisiana and 41 months in Ohio) likely relates to the design of the state CHIP programs; Ohio is the only pure Medicaid expansion program among the 10 study states, and while Louisiana is a combination program, 94 percent of CHIP enrollees are in the Medicaid expansion component. Movement between Medicaid and Medicaid expansion programs is often seamless and typically associated with fewer administrative or paperwork difficulties. It is difficult to pinpoint a specific policy or program design reason that explains Michigan’s and Virginia’s longer duration (35 to 38 months, on average). Both states operate combination CHIP programs with upper income limits of 200 percent of FPL, offer 12-month continuous eligibility (in Virginia, only for CHIP), and allow enrollees to self-declare their income at renewal. Michigan offers online renewal, and Virginia uses preprinted renewal forms populated with the most recent eligibility-related information in the state’s data system. Florida and New York share many of the same characteristics and their duration rates were also higher than average for much of the study period (26 to 29 months, on average).

We gain additional insight by examining the distribution of spell length and how this varies by state (Figure VI.2). Most striking is that the majority of new spells of public coverage in every state lasted at least 16 months, typically taking the child through an annual renewal period. As expected, the percentage with longer spells is highest in the states where the median duration of public coverage is higher (Louisiana, Michigan, Ohio, and Virginia). In most states fewer than 10 percent of enrollment spells were shorter than 6 months; these short spells were more common in California (24 percent) and Utah (12 percent). Some portion of these short spells was likely cases where a child is initially presumed eligible but later found ineligible (or eligible for Medicaid instead of CHIP).

Figure VI.2. Distribution of the Average Length of New Spells of Public Coverage, by State, 2007-2011

Figure VI.2. Distribution of the Average Length of New Spells of Public Coverage, by State, 2007-2011

Source: Mathematica analysis of state-provided administrative data

42 This chapter draws on data presented in two more detailed memoranda: Harrington et al. (2013) contained in Harrington and Kenney et al. (2014), and Orzol et al. (2013).

43 The duration of Medicaid and CHIP enrollment was estimated using state administrative data on children with a new spell of enrollment anytime during the 5-year period from July 2007 to July 2012. For this analysis, a child is considered a new enrollee if he or she is enrolled in the program for two consecutive months, but not in the previous two months. The new enrollment month is the first month of the enrollment spell. New program-specific enrollees include children transferring directly from another program, whereas new public coverage enrollees are those with no public coverage in the preceding two months.

44 The data used for this analysis span August 2007 through June 2012. Median duration may be computed only when the end dates are observed for at least 50 percent of spells. On average, we observed the end of at least 50 percent of spells for a roughly three-year period for the separate CHIP spells and just over two years for Medicaid and Medicaid expansion spells.

View full report


"rpt_CHIPevaluation.pdf" (pdf, 1.01Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®