Neonatal Abstinence Syndrome Vulnerability Index for Pennsylvania Counties
Open Access
- Author:
- Haas, Nikaela
- Area of Honors:
- Industrial Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Soundar Kumara, Thesis Supervisor
Andris Freivalds, Thesis Honors Advisor
Kamesh Madduri, Thesis Supervisor - Keywords:
- Healthcare
NAS - Abstract:
- Neonatal abstinence syndrome (NAS) has been a growing concern in Pennsylvania. NAS occurs when a mother uses opioids during pregnancy, resulting in the infant having a dependence on the drug [1]. NAS stays are costly compared to the costs of newborn infants without NAS [2] and understanding the features of a county that contribute to NAS will allow for improved early intervention rather than treating the symptoms after birth. The objective of this study is to understand which features contribute to high NAS rates in Pennsylvania by looking at data from previous years. The features will then be used to create a NAS Vulnerability Index (NVI) for each county in Pennsylvania. This NVI can be used by researchers and other policy makers to better allocate limited resources to counties most affected by NAS. By collecting data on a variety of factors including demographics, socioeconomic and health related factors, an XGBoost Regressor based Machine Learning model was fit and the impact of each factor on the output was calculated using SHAP values to reveal the three most impactful features on NAS rates were 1) the percentage of women who smoked during pregnancy, 2) the rate of opioid use disorders among pregnant women, and 3) the estimated opioid dispensing rate to pregnant women. These three features were then used to calculate NVI for each county in Pennsylvania. The top five counties with the highest NVI were Venango, Elk, Fayette, Clearfield, and Lawrence. Their NVIs were 0.973, 0.964, 0.944, 0.918, and 0.899, respectively. Counties in the southeast region of Pennsylvania on an average had lower NVIs than other regions of the state. The NVI provides an easily explainable metric that can be used by government officials, researchers, or other policymakers. By examining the specific features that increase NAS rates in a county, efforts can be better targeted towards mitigating these risks. Future work should expand to consider other data and apply the methodologies developed in this thesis to the states outside of Pennsylvania to see how NVI trends in other areas of the US.