Since I am overly paranoid about SIDS claiming my daughter, I have spent considerable time playing with the Centers for Disease Control Wonder database on linked infant birth and death records. Not only is Wonder a free, public source of primary data, but the high level of trust I have in the U.S. system of birth and death certificates makes me confident about the quality of the data.
After my non-scientific analysis of my daughter’s risk of SIDS, I wanted to understand who is at risk for SIDS death in the United States. The CDC purposefully masks characteristics within the database to prevent identification of individual cases. As such, I believe I am only able to analyze independent risks of one characteristic at a time rather than any interactions between multiple characteristics.
Phrased differently, I can analyze whether having a Black/African-American mother makes a newborn more likely to die of SIDS than having a White mother. However, I don’t think I can analyze whether the increased risk of SIDS death with a poorly-educated mother is equally large for Black/African-American mothers versus White mothers. The 3×3 table below shows the deaths from SIDS per 100,000 live births for mothers education by mother race. For example, 71.89 babies died of SIDS per 100,000 live births to White mothers with less than high school education.
Analyzing Interactions in the CDC Wonder Database
Deaths Per 100,000 Live Births in 2007
| Mother Education by Mother Race |
White |
Black / African-American |
| Less Than High School |
71.89 |
161.33 |
| Bachelor’s Degree or Higher |
12.49 |
14.04 |
Raw Data
| Mother Education + Mother Race |
Deaths |
Live Births |
| Less Than High School + White |
286 |
397,842 |
| Less Than High School + Black / African-American |
104 |
64,465 |
| Bachelor’s Degree or Higher + White |
48 |
384,437 |
| Bachelor’s Degree or Higher + Black / African-American |
4 |
28,482 |
I believe I can use the Kaplan-Meier estimator with no right censoring to test whether the cells in any specific cell are significantly different than the cells in any other cell. This allows me to use a simple online calculator rather than opening a statistical package and allows me to record the 95 percent confidence intervals for simultaneous comparison of all cells.
95 Percent Confidence Intervals For Death Rate From SIDS
|
|
Lower Bound |
Upper Bound |
| Less Than High School + White |
63.56 |
80.22 |
| Less Than High School + Black / African-American |
130.32 |
192.33 |
| Bachelor’s Degree or Higher + White |
8.95 |
16.02 |
| Bachelor’s Degree or Higher + Black / African-American |
0.28 |
27.81 |
The lower bound of the 95 percent confidence interval for Less Than High School + Black / African-American is higher than the upper bound of all other groups. Thus I can say with 95 percent confidence that being born to a mother with less than high school education who is Black / African-American makes a baby more likely to die of SIDS. Similarly, the two groups in green (both Bachelor’s Degree or Higher groups) have an upper bound that is lower than the lower bound of both Less Than High School groups. The interval for Bachelor’s Degree or Higher + White overlaps the interval for Bachelor’s Degree or Higher + Black & African-American. So while I can’t say that there is any difference in mother’s race once a mother obtains a Bachelor’s Degree or Higher, I can say that SIDS risk penalty for being poorly educated is exacerbated for mothers who are Black / African-American.