State Experiences Linking Medicaid Data With Birth Certificates and Other Data Sources

Correspondence to: Zachary Predmore, PhD, RAND Corporation, 20 Park Plaza, Suite 920, Boston, MA 02116. E-mail: gro.dnar@romderpz.

This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. http://creativecommons.org/licenses/by-nc-nd/4.0/

Background:

Many states link Medicaid claims with birth certificates or other data, often to inform programs and policies aimed at improving maternal and child health (MCH).

Objectives:

To develop an up-to-date understanding of the extent of the use of linked Medicaid claims for MCH research by state.

Research Design:

We completed a structured literature review, developed an inventory of linkage efforts, and facilitated semistructured discussions with representatives from 9 states with established Medicaid claims data linkages to understand the technical details of linkages, experiences creating and maintaining linkages, and barriers or facilitators to establishing linkages.

Results:

We identified 45 peer-reviewed journal articles representing 22 states that used linked Medicaid data to study MCH and 33 states and territories that publicly report on Medicaid data linkages for a total of 39 states with any in-scope linkage. Discussions revealed that linkages often arose from the desire to answer a specific question or evaluate a program but then expanded to other use cases and that most states enable external researchers to access data for analysis. Respondents provided a few examples of where linked birth certificate data were used for health outcomes research.

Conclusion:

Additional resources including technical assistance for identifying best practices along with interagency collaboration could overcome barriers and facilitate a coordinated and consolidated approach across states.

Maternal mortality rates in the United States have increased steadily from 2000 to 20201,2 and are over twice that of other high-income countries.3 Significant racial and ethnic disparities in maternal and infant health exist with Black, Native American, and Hispanic populations bearing a disproportionate burden of maternal and infant mortality.4,5 Medicaid is the payer for almost half of all births in the United States, including more than half of all births by Black and Hispanic women,6 and as such is a critical factor in policies needed to improve maternal and infant health. Maternal and infant Medicaid claims and enrollment data are tracked by state Medicaid agencies and contain detailed longitudinal data on diagnoses and health care utilization that are not captured on birth certificates. Births are registered by state vital records offices and reported to the National Center for Health Statistics at the Centers for Disease Control and Prevention.7 Information on birth certificates varies by state but they typically contain information on key maternal and infant health characteristics, many of which are not captured in claims (eg, parent’s educational attainment) or are captured inconsistently (eg, low birth weight).

Medicaid data linked with birth certificates or other sources of data are a potentially rich data resource for research on the outcomes and effectiveness of public health programs, different models of clinical care, and policy interventions. Efforts to link Medicaid claims to birth certificates for the purposes of maternal and child health (MCH) research have been ongoing for more than 20 years.8,9 In addition, once Medicaid data are linked to birth certificates, Medicaid data can then be linked to other sources of data that are already regularly linked to birth certificate data. For example, birth certificates are used to create the sample for the Pregnancy Risk Assessment Monitoring System—a large Centers for Disease Control and Prevention and health department survey on maternal and infant health administered in 46 states, the District of Columbia, and 2 U.S. territories,10 allowing for the Pregnancy Risk Assessment Monitoring System to be linked to Medicaid data. Other data that some states may regularly link with live birth certificate data include death certificates and hospital discharges.

Despite the benefits of linking Medicaid data with birth certificates and other data, efforts to link these data at the state level have been fragmented. Across these linkage efforts, there is substantial variation in the data sources linked to Medicaid data, linking methods, fields used to create linkages, match rates, and research and policy applications of the linked data.

The specific aim of this study is to comprehensively describe efforts to link Medicaid claims to birth certificates and other data and provide examples of how these linked data can be used and expanded upon to advance public health for researchers and policymakers. This aim is achieved through a structured literature review and semistructured discussions with representatives of states that link these data.

METHODS

We took a 3-pronged approach to study current data linkage practices of Medicaid claims to birth certificates and other data sources across U.S. states and territories. First, we conducted a structured literature review, followed by an inventory of state and territory linkage efforts, and lastly, we held semistructured discussions with representatives from 9 states identified as having ongoing data linkage processes. Organization‘s Institutional Review Board reviewed and approved the study protocol (2022-N0054). All participants received a consent document with study information in advance of the discussion and provided oral consent. A technical expert panel of federal government employees with expertise in MCH data linkages provided guidance on the design of the literature review, inventory of state efforts, and discussion guide.

Literature Review

We conducted a structured literature review of recent (February 28, 2017–March 1, 2022) peer-reviewed articles to identify published research that linked Medicaid or all-payer claims database (APCD) claims (including Medicaid claims) at the individual level with data from birth certificates or other sources of data. Using a predetermined set of databases and search terms (Supplemental Appendix 1, Supplemental Digital Content 1, http://links.lww.com/MLR/C618), we conducted a title and abstract search. All articles from the search were then screened by 2 researchers based on predetermined inclusion criteria (Supplemental Appendix 2, Supplemental Digital Content 1, http://links.lww.com/MLR/C618). The researchers then abstracted articles using predetermined codes relating to state, years of data linked, types of data linked, and details of linkages (Supplemental Appendix 3, Supplemental Digital Content 1, http://links.lww.com/MLR/C618). When we identified studies using secondary analyses of previously linked data, we included information from the original study if cited in the article.

State and Territory Inventory

We developed an Inventory of state and territory linkage efforts by conducting targeted searches for each state and territory of each state’s Medicaid agency website, vital records and statistics website, health department website, MCH Title V Block Grant Fiscal Year 2022 Application/2020 Annual Report,11 and APCD website, where applicable.

Semistructured Discussions

We organized and facilitated semistructured discussions with representatives from 9 states that had more than 50,000 live births in 202012 and perform ongoing data linkages of Medicaid claims with other data for the purposes of improving maternal health. The states were purposefully sampled to include states of varying population sizes (split into the larger and smaller half of eligible states sorted by the number of births) and from different census regions; sampling was informed by the literature review and inventory of state efforts (details on state selection in Supplemental Appendix 4, Supplemental Digital Content 1, http://links.lww.com/MLR/C618). Representatives included state agency leads, data analysts, academics, and public health practitioners who analyze the data and/or conduct linkages. We developed a semistructured discussion guide that included questions about the origins of the linkage, the technical specifications, uses of the linked data for research, policy, or practice, and any barriers or facilitators to creating or maintaining the linkage (Supplemental Appendix 5, Supplemental Digital Content 1, http://links.lww.com/MLR/C618). Semistructured discussions were attended by at least 2 members of the study team, conducted virtually, typically lasted 1 hour, and were audio-recorded and transcribed. Transcripts were coded by a single coder and reviewed and edited by a second coder in Dedoose version 9.0.5413 to identify common themes using a conventional content analysis approach.14 There were no major areas of disagreement between the 2 coders.

RESULTS

Literature Review

The initial search identified 687 articles. Articles were excluded for no U.S. data (N = 145), no in-scope study type (N = 70), not a maternal/child health topic (N = 60), no in-scope data linkage used (N = 349), and not a state-level linkage (N = 18) (Supplemental Appendix 6, Supplemental Digital Content 1, http://links.lww.com/MLR/C618). A total of 45 studies were included that used data from 22 different states, including 4 studies that used data from multiple states, one of which used data from 3 unspecified states.15 The findings are presented in Table ​ Table1. 1 . States linked Medicaid data to a variety of types of other data, with some studies involving linkages to multiple other types of data. Birth certificates were the most commonly linked type of data in 75.6% of studies followed by social services data (15.6% of studies) and death certificates (13.3% of studies). Of the studies, 37.8% linked claims with more than one category of data. Just over half of the studies provided details on whether the linkages were deterministic, probabilistic, or both, and specified the types of variables used to link. Less than half of the studies provided match rates or validation analyses. References for all articles included in the literature review are provided in Supplemental Appendix 7 (Supplemental Digital Content 1, http://links.lww.com/MLR/C618).

TABLE 1

Summary of Linkage Processes in Literature Review (N = 45)

Bold indicates the number of studies in which data from multiple of the “other datasets” listed above are linked.

EHR indicates electronic health record; MCH, maternal and child health.

State and Territory Inventory

Of the 59 jurisdictions with Medicaid programs reviewed, 33 jurisdictions had in-scope data linkage efforts (excluding those identified in the peer-reviewed literature), of which 28 (47.4%) mentioned a Medicaid-to-birth record linkage in the Title V MCH report and 5 jurisdictions mentioned an APCD (inclusive of Medicaid claims) to vital record linkage (Table ​ (Table2). 2 ). Among those that mentioned an APCD, 2 jurisdictions had no in-scope linkages described in the Title V MCH report. We noted that 14 articles from the peer-reviewed literature were from 6 states that did not have publicly available information on current, in-scope state-level data linkage efforts. Some of these articles used older data, whereas others may have linked data solely for a specific research project. In total, 39 jurisdictions had in-scope data linkage efforts when considering both the peer-reviewed literature and the targeted reviews of state agency websites. Some of the Title V MCH reports discussed challenges related to data linkages and infrastructure including staff turnover, lack of expertise needed to develop and maintain data infrastructure, interagency access to data (even within the same state), and resource diversion to the COVID-19 pandemic response. A list of information found for each jurisdiction is provided in Supplemental Appendix 8 (Supplemental Digital Content 1, http://links.lww.com/MLR/C618).

TABLE 2

Characteristics of Linked Data and Details on Linkage Process

Bold indicates there were 33 jurisdictions in which we are able to abstract some details of the linkage.

APCD indicates all-payer claims database; MCH, maternal and child health.

Group Discussions

We conducted discussions with representatives from 9 states with ongoing linkages to Medicaid claims; included states along with data features and example use cases are presented in Table ​ Table3. 3 . The number of discussion participants from each state varied from 2 to 5. Most states conducted data linkages within a specific agency in their state department of health, but a few outsourced linkages to academic or private partners. Most of the linkages were done using probabilistic approaches. Also, most linkages were done annually, though there were a few that did more frequent linkages. Variables used for the linkage commonly included mother or child name, mother or child date of birth, and mother or child Social Security numbers. Address or other geographic variables like the county of residence were used less frequently. Some states included the payer on the birth certificate, which was helpful when present; one state included the Medicaid ID number on the birth certificate, which facilitated matches. We also included several states that linked Medicaid and birth certificate data through larger efforts to build data warehouses or health information exchanges.

TABLE 3

Data Features and Use Cases

APCD indicates all-payer claims database; COVID, coronavirus disease; MCH, maternal and child health; WIC, women, infants, and children.

The linked data are used for a variety of purposes, but most states used data primarily to assess MCH outcomes, often around specific programs or interventions they implemented. Several states used the linked data to target programs or quality improvement initiatives regionally, and 2 states reported using this linked data for individual patient-level outreach (Louisiana provides data to Medicaid managed care plans so they can reach out to mothers with a previous preterm delivery,16 Maryland has done outreach to remind people to complete coronavirus disease vaccination series17).

Generally, the closer that the Medicaid agency and the agency responsible for vital records were in the state government, the easier the data of each agency was to access for linkage. For example, in some states where these agencies were very close organizationally (within the same department or cabinet), staff at each agency were able to access data from the other agency without any additional data use agreements (DUAs). However, in other states where the agencies may not be housed within the same department, DUAs were needed to share data between agencies.

Almost every state allowed external researchers to access linked data, usually by submitting a proposal that needed to be approved by each agency providing data. Some states also required review and approval by the state’s Institutional Review Board before data could be shared. The frequency of these requests varied; some states reported rarely or almost never receiving them, whereas others reported getting monthly requests for linked data.

Discussion participants reported challenges with linking data, including several related to the nature of the datasets being linked as presented in Table ​ Table4. 4 . There were some issues, around which services were billed to a mother versus a child’s Medicaid plan, as sometimes the baby’s claims ended up on the mother’s plan. Sometimes the baby’s name on the birth certificate was “Baby Boy” or “Baby Girl,” impairing the ability to match claims data. Twins were reported to be hard to match, as were babies born out of state. Missing data were also cited as a common concern. Despite these concerns, match rates were almost always over 90% and every discussion participant believed these linked datasets were of high enough quality to conduct research.

TABLE 4

Discussion Themes and Illustrative Quotes

DUA indicates data use agreement; MCH, maternal and child health; VS, vital statistics.

Staff turnover was another challenge, as establishing and maintaining linkages often required staff with specialized technical knowledge. Facilitators of linkage efforts included having strong leaders who were able to drive multiyear projects across multiple departments, and strong relationships between staff working in these departments; in many states, the key people at Medicaid and Vital Records agencies had worked together for years and connected regularly for these linkages and other projects.

DISCUSSION

In this study, we detailed state and territory efforts to link Medicaid claims with birth certificate data and other data sources and to use these data to improve population health. Our review of the academic and grey literature identified linkage efforts between Medicaid claims and birth certificates or other data sources in more than half of jurisdictions. Birth certificates were the type of data most commonly linked to Medicaid claims followed by social services data and death certificate data. These findings were supported by our discussions with state officials, though multiple states also noted recent efforts to link immunization data. This was a new dataset not noted in the peer-reviewed literature but may represent a new key source of data that could be valuable in the state’s responses to the COVID-19 pandemic and other public health issues. Through both reviews of state agency websites and discussions with state representatives, we noted several recent efforts to develop more centralized data repositories linking APCD data with birth certificates and other sources of data or creating health information exchanges across state agencies. Some of these states use central data hubs using Master Patient Indexes to facilitate quick-turnaround analysis projects using linked data (including California, Maryland, Michigan, and New York). Some states performed linkages within the state health department or Medicaid agency whereas others subcontracted the linkage to a local university or government contractor.

Common challenges noted included obtaining access to data across agencies or departments within a state and finding time, resources, and experienced staff to develop and maintain data infrastructure. Facilitators to data linkage efforts including strong relationships between staff working in different agencies and institutional buy-in to the importance of building and maintaining data linkage capacity.

Limitations

We note some limitations that should be considered when interpreting our findings. First, although we restricted our search to studies used in the past 5 years, due to lags in analyzing and publishing data some of the data used in the studies were much older and may not be generalizable to data linkages today, which are rapidly evolving in scope and sophistication. Second, states may not consistently report on linkage efforts or may report conflicting information. Although we examined diverse sources including published literature, state websites, and state reports, we may not have captured every major state linkage effort. Supplemental Appendix 8 (Supplemental Digital Content 1, http://links.lww.com/MLR/C618) provides additional information on the few instances of conflicting information that we noted. We also did not systematically explore the quality of the data linkages in this study, so researchers interested in using these linked data may need to be mindful of potential problems with missingness or other data quality issues. Finally, we note that our group discussions were limited to larger states with ongoing data linkage efforts and with representatives who had availability to speak with us. This allowed us to get information on methods related to large-scale linkages, but limited generalizability to other states and territories. We also did not systematically explore the fine-grained technical details of the linkages. Future studies could include technical document reviews to rigorously study the methods used and technical details of these linkages.

Policy Implications

The results of this study demonstrate that the large-scale linkage of Medicaid data with birth certificates and other data sources is feasible and has many potential applications to research, policy, and practice. Our literature review found that linked data have been used extensively by researchers to evaluate the impact of policies such as Medicaid expansion18–23 and social services24,25 on mothers and infants. Group discussions with states who link these data also identified key use cases involving policy analysis such as evaluating the maternal and infant health outcomes after postpartum Medicaid expansion or receipt of supplemental nutrition programs. States, either alone or in partnership with academic institutions, can analyze linked data to inform important policy decisions. Another use case for these data highlighted by states was identifying populations or individuals that need to be targeted with additional resources. For example, states might use linked data to direct additional resources to counties with high maternal morbidity and mortality rates. States even gave examples of using linked data to target appropriate programs for individuals with specific risk factors.

These use cases could justify the additional effort to support the development, expansion, and maintenance of data linkages. Many states have demonstrated the value of performing these linkages to decision-makers and secured dedicated funding and staff to perform linkages. Data linkages do not necessarily have to be performed in-house, however, and some states have found success by partnering with academic institutions or contracting data linkages. States can also support data linkages in ways that do not necessarily require significantly increased funding. For example, states may be able to integrate linkages of Medicaid and birth certificate data into existing APCD or health information exchange efforts. States can also facilitate data sharing by simplifying the DUA process within the state by creating a universal form that can be used by all agencies. However, these are state-based efforts and the differences among states make it harder to aggregate linked data for national-level research and programmatic and policy-making decisions. As suggested by respondents, state data linkages could benefit from federal support to achieve a consensus on the best processes for linkage methods.

CONCLUSIONS

In this 3-pronged study consisting of a literature review, inventory of state efforts, and group discussions, we found many applications for using Medicaid data linked with birth certificates and other data sources and highlight several barriers and potential facilitators to build and maintain these linkages. Researchers may consider using linked data from more than one state when conducting policy evaluations, particularly in the areas of maternal and infant health. In addition, discussion participants suggested that the state collaboratives or networks could facilitate the use of linked data to inform policymaking and that technical expertise could identify best practices for data linkage strategies.

Supplementary Material

Footnotes

This research was funded by the Office of the Assistant Secretary for Planning and Evaluation (ASPE) (contract #HHSP233201500038I) and carried out within the Payment, Cost, and Coverage Program in RAND Health Care.

The authors declare no conflict of interest.

Supplemental Digital Content is available for this article. Direct URL citations are provided in the HTML and PDF versions of this article on the journal’s website, www.lww-medicalcare.com.

REFERENCES

1. Hoyert DL. Maternal mortality rates in the United States, 2020. National Center for Health Statistics. 2022. Accessed March 31, 2023. https://www.cdc.gov/nchs/data/hestat/maternal-mortality/2020/maternal-mortality-rates-2020.htm.

2. MacDorman MF, Declercq E, Cabral H, et al.. Is the United States maternal mortality rate increasing? Disentangling trends from measurement issues short title: US maternal mortality trends . Obstet Gynecol . 2016; 128 :447. [PMC free article] [PubMed] [Google Scholar]

3. Tikkanen R, Gunja MZ, FitzGerald M, et al.. Maternal mortality and maternity care in the United States compared to 10 other developed countries . Commonwealth Fund . 2020; 10 . 10.26099/411v-9255. [CrossRef] [Google Scholar]

4. Bryant AS, Worjoloh A, Caughey AB, et al.. Racial/ethnic disparities in obstetric outcomes and care: prevalence and determinants . Am J Obstet Gynecol . 2010; 202 :335–343. [PMC free article] [PubMed] [Google Scholar]

5. Centers for Disease Control and Prevention.Infant Mortality. Published 2022. Accessed September 9, 2022. https://www.cdc.gov/reproductivehealth/maternalinfanthealth/infantmortality.htm

6. Martin JA, Hamilton BE, Osterman MJ. Births in the United States, 2021. MCHS Data Brief. No. 442. 2022. Accessed March 31, 2023. https://www.cdc.gov/nchs/data/databriefs/db442.pdf.

7. National Center for Health Statistics. Vital Statistics Online Data Portal. Published 2022. Updated June 8, 2022. Accessed August 8, 2022 https://www.cdc.gov/nchs/data_access/vitalstatsonline.htm. [Google Scholar]

8. Gyllstrom ME, Jensen JL, Vaughan JN, et al.. Linking birth certificates with Medicaid data to enhance population health assessment: methodological issues addressed . J Public Health Manag Pract . 2002; 8 :38–44. [PubMed] [Google Scholar]

9. Buescher PA. Method of linking Medicaid records to birth certificates may affect infant outcome statistics . Am J Public Health . 1999; 89 :564–566. [PMC free article] [PubMed] [Google Scholar]

10. Centers for Disease Control and Prevention. Pregnancy Risk Assessment Monitoring System. Published 2021. Accessed April 11, 2022. https://www.cdc.gov/prams/index.htm.

11. Health Resources and Services Administration. State Application/Annual Report. Published 2022. Accessed August 8, 2022 https://mchb.tvisdata.hrsa.gov/Home/StateApplicationOrAnnualReport.

12. National Center for Health Statistics. Births, births rates, deaths, and death rates by state and territory. Published 2021. Accessed July 13, 2022. https://www.cdc.gov/nchs/fastats/state-and-territorial-data.htm.

13. Oregon Health Authority. APAC Data Requests Fact Sheet. Published 2022. Accessed April 12, 2022. https://www.oregon.gov/oha/HPA/ANALYTICS/APAC%20Page%20Docs/APAC-Data-Request-Fact-Sheet.pdf.

14. Hsieh H-F, Shannon SE. Three approaches to qualitative content analysis . Qual Health Res . 2005; 15 :1277–1288. [PubMed] [Google Scholar]

15. Clemans-Cope L, Lynch V, Howell E, et al.. Pregnant women with opioid use disorder and their infants in three state Medicaid programs in 2013–2016 . Drug Alcohol Depend . 2019; 195 :156–163. [PubMed] [Google Scholar]

16. Mehta P. Prematurity Prevention & Louisiana Medicaid: Progress to date and a path forward. Published 2018. Updated May 17, 2018. Accessed August 8, 2022 https://ldh.la.gov/assets/docs/MQI/MQIMeetings/PrematurityImprovementMedicaidQualitymeetingFINAL051718.pdf.

17. CRISP. HIE Tools for Vaccinations and COVID-19 Response Efforts. Published 2021. Updated February 22, 2021. Accessed August 8, 2022. https://health.maryland.gov/mmcp/Documents/MMAC/2021/MMAC%20Presentation%20-%20HIE%20Vaccination%20and%20COVID%20Tools.pdf .

18. Gordon SH, Hoagland A, Admon LK, et al.. Extended postpartum Medicaid eligibility is associated with improved continuity of coverage in the postpartum year: study examines stability of health insurance enrollment in Colorado for people who retain Medicaid coverage for the entire postpartum year . Health Aff . 2022; 41 :69–78. [PubMed] [Google Scholar]

19. Dunlop AL, Joski P, Strahan AE, et al.. Postpartum Medicaid coverage and contraceptive use before and after Ohio’s Medicaid expansion under the Affordable Care Act . Womens Health Issues . 2020; 30 :426–435. [PubMed] [Google Scholar]

20. Harvey SM, Gibbs S, Oakley L, et al.. Medicaid expansion and neonatal outcomes in Oregon . J Eval Clin Pract . 2021; 27 :1096–1103. [PubMed] [Google Scholar]

21. Harvey SM, Oakley LP, Gibbs SE, et al.. Impact of Medicaid expansion in Oregon on access to prenatal care . Prev Med . 2021; 143 :106360. [PubMed] [Google Scholar]

22. Hawkins SS, Horvath K, Noble A, et al.. and Medicaid expansion increased breast pump claims and breastfeeding for women with public and private insurance . Womens Health Issues . 2022; 32 :114–121. [PubMed] [Google Scholar]

23. Steenland MW, Wilson IB, Matteson KA, et al.. Association of Medicaid expansion in arkansas with postpartum coverage, outpatient care, and racial disparities . JAMA Health Forum . 2021; 2 :e214167. [PMC free article] [PubMed] [Google Scholar]

24. Clark RE, Weinreb L, Flahive JM, et al.. Health care utilization and expenditures of homeless family members before and after emergency housing . Am J Public Health . 2018; 108 :808–814. [PMC free article] [PubMed] [Google Scholar]

25. Bersak T, Sonchak L. The impact of WIC on infant immunizations and health care utilization . Health Serv Res . 2018; 53 :2952–2969. [PMC free article] [PubMed] [Google Scholar]