The Birth Certificate and Medicaid Data Match Project: Initial Findings in Infant Mortality

1
The Birth Certificate and Medicaid Data Match
Project Initial Findings in Infant Mortality

• John Oswald, PhD
• Director, Minnesota Center for Health Statistics
• June 7, 2006
• NAPHSIS Annual Meeting

2
Background

• The Minnesota IMR in 2002 was 5.3 per 1000 live
  births
• One of the lowest infant mortality rates in the
  nation
• The low rate masks great disparities in infant
  mortality within the state

3
Background

• The Minnesota Department of Human Services (DHS)
  and the Minnesota Department of Health (MDH)
  entered into a data sharing agreement to link
  Minnesota Medicaid data with birth certificate
  data
• This linking allows for comparisons between the
  Medicaid and non-Medicaid populations
• 1997-2001 linked data are analyzed for this
  presentation

4
Background

• The Medicaid population is comprised of both
  Medical Assistance and MinnesotaCare enrollees
• The non-Medicaid population includes those
  covered by private health insurance,
  self-insured, or uninsured

5
Linkage Methods

• Iterative matching procedure that included
  mothers name, mothers date of birth, and
  fathers last name.
• State of residence of the birth mother was used
  rather than state of occurrence of the birth

6
Linkage Methods

• overall match rate has been 93.1
• No major differences between Medicaid women who
  matched and those who did not except
• Hispanic women
• Border communities with North Dakota

7
Linkage Methods

• Future methodological issues in data linkage and
  probabilistic techniques
• Agenda for future research
• Gyllstrom ME, Oswald JW, et al. Linking Birth
  Certificates with Medicaid Data to Enhance
  Population Health Assessment Methedological
  Issues Addressed. Journal of Public Health
  Management and Practice, July 2002, 38-44.

8
Other Methods

• Race categories are based upon the race of the
  mother as reported on the birth certificate and
  include White, Black/African American, Asian, and
  Native American
• Hispanic ethnicity is also based upon ethnicity
  of mother as reported on the birth certificate
  and includes anyone indicating Hispanic/Latina
  descent regardless of race

9
Other Methods

• Age of mother is obtained from the birth
  certificate
• Initiation of prenatal care is also taken from
  the birth certificate

10
Methods
11
Results
Medicaid Births by Maternal Age Minnesota,
1997-2001
Non-Medicaid Births by Maternal Age Minnesota,
1997-2001
12
Results

• Minnesota as a whole had an IMR of 5.7 per 1,000
  for calendar years 1997-2001
• Medicaid had an IMR of 7.4 per 1,000
• 7.5 per 1,000 for Medical Assistance
• 6.2 per 1,000 for MinnesotaCare
• Non-Medicaid had an IMR of 5.2 per 1,000

13
Results
14
Results
15
Results
16
Results
17
Discussion

• Whites had significantly lower IMR than all other
  race categories, and White non-Medicaid had
  significantly lower IMR than White Medicaid
• Finding for teenaged mothers supported in results
  from other states
• Those who receive early prenatal care have lower
  infant mortality in both the Medicaid and
  non-Medicaid populations

18
Discussion

• Unintentional injury deaths accounted for a high
  proportion of deaths in the Medicaid population
• 1 in 3 was related to co-sleeping with adults
• Inadequate sleeping accommodations
• SIDS also much higher in the Medicaid populations
  (25.5 vs. 7.3)

19
Limitations

• Small number of events
• Lack of clear definition of Non-Medicaid
  population
• Inherent challenges of birth and death records
• Unknown enrollment date in Medicaid

20
Further Research

• Further investigation into the teen mother
  finding
• Formal analysis of these data to examine the
  interplay of race and Medicaid status
• Ideally, a comparison of Medicaid and those with
  private insurance