Utilizing Mississippi 2007-2011 linked birth and death data to explain district level variances in infant mortality rates

Background: The highest infant mortality rate (IMR) in the US is consistently found in Mississippi. In 2011, Mississippi had an IMR of 9.4 infant deaths per 1,000 live births compared to the national rate of 6.05 infant deaths per live births (MSDH, 2011; Hoyer & Zu, 2012). However, there are discrepancies among Public Health Districts (PHD). During 2011, rates ranged from 8.2 infant deaths per 1,000 live births in PHD IX to 14.0 in PHD VII (MSDH, 2011). IMR is considered a key indicator of the health of a population. District-level variances in IMR suggest that where one lives in Mississippi indeed affects health. Identifying factors that result in these variances could inform community-level planning to address population health needs and subsequently, reduce infant deaths.

Methods: We performed secondary analysis of Mississippi’s 2007-2011 linked birth and death certificate data (MSDH, 2007-2011). The sample included 205,955 live-born singleton infant births. The dependent variable, infant death, is defined as an infant that died during the first year of life (< 365 days). Independent variables included descriptive infant characteristics such as birth weight, gestational age, and abnormal complications of the newborn, and maternal characteristics such as age, race, education, smoking, previous number of live births, and public health district of residence. Multiple logistic regression analysis was used to identify variables having an effect on the varying rates of infant mortality at the district level.

Results: Differences can be partially attributed to several mediating infant factors including birth weight, clinical gestational age, abnormal conditions of the newborn, and maternal factors including tobacco use, marital status, education, race, complications during labor and delivery, obstetric procedures and the previous number of live births, and interactions among these variables (R²=0.29, P<0.0001).

Discussion: While the logistic regression was statistically significant, the model only accounts for approximately 29% of the variability around infant mortality rates between the districts (R²=0.29). More research is needed to further explore the relationship between public health district and infant mortality.  

Implications for Public Health: Identifying factors that result in district-level IMR variances could inform efforts to reduce infant deaths.