Wittman, Samuel
(2021)
Spatial and temporal patterns of overdose in Pennsylvania, 2012-2017: analysis with spatiotemporal scan and Getis-Ord statistics.
Master's Thesis, University of Pittsburgh.
(Unpublished)
This is the latest version of this item.
Abstract
The purpose of this thesis was to investigate spatial and temporal patterns of fatal overdoses at the census tract level in the U.S state of Pennsylvania (PA), from 2012 to 2017. During these years, fatal overdose reached historically high rates in PA and became a key public health concern in the state. This thesis identified and visualized fatal overdose clusters and hot/cold spots at a more granular level than previous analyses. This information has important implications for understanding where and how fatal overdoses “spread”, and for identifying critical regions for overdose surveillance and intervention.
Individual-level overdose mortality data including decedent age, race/ethnicity, sex, county, and census tract were compiled by the PA Department of Health. SaTScan was used to analyze this data with space-time scan statistics, using a discrete Poisson model. Some SaTScan analyses were stratified by age or sex. The most likely high-rate or low-rate clusters were identified and tested for significance using Monte Carlo methods. ArcGIS was used to analyze the data for hot spots or cold spots using Getis-Ord Gi* statistics. Gi* statistics were generated for each census tract during each quarter of this period, and an animation was generated to show patterns during the entire period.
SaTScan analysis revealed a statistically significant increase in fatal overdose throughout the entire state during 2016-2017. After adjusting for this trend, SaTScan analysis further showed large, statistically significant high-rate clusters in southwestern PA and Philadelphia, as well as smaller significant high-rate clusters in Pottstown, York, Allentown, and Hazleton. Stratified SaTScan analyses revealed significant clusters in Philadelphia and southwestern PA for all age groups and both sexes; however, there were important demographic-specific differences in these regions and others across the state. In contrast, Getis-Ord analysis was not useful for examining the significance of specific overdose clusters. Rather, this analysis revealed a fine-grained depiction of relative rates and local patterns of overdose in cities and regions throughout the state. Getis-Ord analysis showed that hot spots were sporadic and inconsistent throughout much of the state (including high-risk southwestern PA), while there were a few highly regular hot spots in neighborhoods of northeastern Philadelphia.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
19 January 2021 |
| Date Type: |
Publication |
| Defense Date: |
11 November 2020 |
| Approval Date: |
19 January 2021 |
| Submission Date: |
2 December 2020 |
| Access Restriction: |
2 year -- Restrict access to University of Pittsburgh for a period of 2 years. |
| Number of Pages: |
48 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
School of Public Health > Biostatistics |
| Degree: |
MS - Master of Science |
| Thesis Type: |
Master's Thesis |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
overdose, Getis-Ord, SaTScan, Pennsylvania, PA |
| Additional Information: |
Thesis includes supplementary animations |
| Date Deposited: |
19 Jan 2021 20:52 |
| Last Modified: |
19 Jan 2023 06:15 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/40060 |
Available Versions of this Item
-
Spatial and temporal patterns of overdose in Pennsylvania, 2012-2017: analysis with spatiotemporal scan and Getis-Ord statistics. (deposited 19 Jan 2021 20:52)
[Currently Displayed]
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[img]](https://d-scholarship.pitt.edu/style/images/fileicons/video.png)
![[feed]](/images/feed-icon-32x32.png){kind=icon}