Medicaid Data Increasingly Used to Improve State’s Maternal and Child Health Programs
Posted December 17, 2015 by Rachel Kremen
|A growing number of states are using Medicaid data to spot trends in maternal and child health outcomes.
With America’s infant mortality rate almost three times higher than other industrialized countries and low-income families affected most significantly, a growing number of states are using Medicaid data to spot trends in maternal and child health (MCH) outcomes.
“Nearly half of all births are paid for by Medicaid, so Medicaid data has enormous potential to tell a story about the health and well-being of many of the country’s most vulnerable citizens,” says Sabrina Selk, ScD, a senior analyst at NICHQ.
Some states have already found success combining Medicaid claim and enrollment data with information from other sources, such as vital statistics and state immunization records, to yield new insights into MCH matters.
“One of the first issues we identified with linked Medicaid-vital records data was the high rate of smoking among pregnant women on Medicaid,” says Laurie Cawthon, MD, MPH, a public health epidemiologist in the Division of Research and Data Analysis of the Washington State Department of Social and Health Services. “This helped convince state leaders to maintain our cessation hotline even as budget cuts ended some smoking cessation programs for pregnant women.”
In Ohio, in Franklin County, almost half of the women get their care from small physician practices, and may require several referrals before their first appointment at a major health center, explains Mary Applegate, MD, medical director for the Ohio Department of Medicaid. Subsequently, the strategy of targeting high volume academic centers had to be abandoned based on this data. Instead, the Ohio team partnered with community organizations and health systems to develop a prenatal care connection. Using a 1-800 number, women on Medicaid can quickly and easily arrange prenatal visits at multiple sites of service. By addressing provider access, this solution provides additional time to potentially address some social determinants of health issues.
Getting Access to Data
Applegate suggests that those interested in tracking Medicaid data call the state’s Medicaid medical director. For states where no such position exists, the Medicaid director should be able to direct the data request to the best person.
Washington’s success with Medicaid data is, in part, due to the state’s investment in connecting databases to create richer information to explore. Cawthon notes that another factor is Washington’s dedication to developing in-house analytic capacity.
But that shouldn’t discourage those in other states, with less available funding, from trying to use such data. Selk suggests taking some time to find the right contact at a state office.
“Ideally, you want to find someone who shares your passion for maternal child health and specific interests within the field,” advises Selk. “It takes less time and effort if they’ve considered the issue before, plus they’ll have more drive and capacity.”
Once you find the data, one of the greatest challenges with analyzing it is the frequency at which Medicaid data is updated. Both Medicaid claims and enrollment data may be updated daily as providers submit claims for services and new clients enroll in Medicaid. The Medicaid claims for recent time periods can change as claim denials, adjustments and resubmitted claims are processed.
Traditionally, researchers look at 12 months of data at the end of each year. But, warns Applegate, that data involves new clients as well as those who entered the system before an initiative took root, so it isn’t ideal for looking at whether a particular approach has been helpful.
Other approaches are analyzing data on a “rolling year,” measuring back 12 months from the current quarter. Applegate says this method is quite stable, but again isn’t great at isolating the impact of new programs. Her preferred method looks at just the last three months of data, which is ideal for tracking only new patients.
Cawthon takes a similar approach. “To avoid the data that’s still in flux, we let claim data “mature” for six months before including it in the analysis.”
Another useful strategy to evaluate the impact of changes is to define intervention and comparison groups based on the timing of a key outcome, such as delivery, says Cawthon. All the mothers who delivered their babies before the intervention might serve as a comparison group for the mothers whose babies were delivered after the intervention was in place.
One of the major points of focus within the NICHQ and U.S. Maternal and Child Health Bureau-led Collaborative Improvement and Innovation Network to Reduce Infant Mortality (IM CoIIN) initiative is to improve state and national data capacity including use of real time data, even provisional, so that state and program leaders can respond in a more timely way to make improvements and changes that benefits infants, families and communities.
“There’s so much untapped potential. We really haven’t seen all the various ways this data can be applied,” says Selk. “States are already collecting this data; it is just waiting to be used in many cases. The work of these early adopters is crucial to help other states model their own efforts to learn from their data.”
Learn more about NICHQ's infant health work.
Learn more about the IM CoIIN initiative.