Description

This project investigates the alarming trends in drug overdose deaths across the United States, utilizing a comprehensive dataset covering various demographic variables. The data encompasses crucial factors such as gender, race, drug type, and age, allowing for a detailed exploration of the epidemic's impact on different population segments.

The analysis focuses on trends over time, identifying key factors associated with higher death rates, and examining the effectiveness of public health policies implemented to combat the crisis. By parsing and cleaning the data, the team ensured its accuracy and reliability, setting the stage for insightful analysis and visualization.

My team and I are honored to have received a Special Mention in the Best Data Visualization category at the 5th DubsTech Datathon, the University of Washington's data science hackathon. Competing with talented data enthusiasts, this recognition highlights our dedication to excellence in data analysis and visualization.

Objective / Success Metrics

The primary goal of this project was to understand the underlying patterns and drivers of drug overdose deaths in the U.S. Specifically, the objectives included:

• Trend Analysis: To identify significant changes in drug overdose death rates over time, broken down by drug type, sex, age, and race.
• Factor Association: To determine which factors are most strongly associated with higher death rates.
• Policy Impact: To assess the impact of major public health policies and interventions on overdose death rates.

These objectives aimed to provide actionable insights for public health officials and policymakers to better address the ongoing drug overdose epidemic.

Approach

1. Data Preprocessing (Python):

   During the data preprocessing phase, we systematically organized the dataset to ensure it was well-structured for analysis. Key actions included:

   • Parsing Labels: Extracted demographic information from the 'STUB_LABEL' into new columns ('SEX', 'RACE', 'ETHNICITY', 'AGE').
   • Column Removal: Dropped the 'FLAG' column, which identifies the presence of null values in the 'ESTIMATE' column, as it was irrelevant for our analysis.
   • Null Value Treatment: Removed rows with null values to ensure data completeness, reducing the dataset from 6,228 to 1,050 rows. This step eliminated duplicate entries with different subcategories, resulting in a clean and reliable dataset for accurate analysis.

2. Data Analysis and Visualization (Tableau):

   With the refined dataset, Tableau was utilized to create visualizations that uncover trends and patterns in drug overdose deaths. The visualizations provided a clear narrative of the epidemic's progression and highlighted critical insights for public health strategies.

Results - Trend Analysis

Results - Significant Factors

OLS Regression Model Summary:

Our regression analysis examines the relationship between various factors and drug overdose death rates. The dependent variable is 'ESTIMATE' (drug overdose death rates), and the independent variables include 'PANEL', 'SEX', 'ETHNICITY', and 'RACE'.

• R-squared: 0.474 - Approximately 47.4% of the variance in drug overdose death rates is explained by the model.
• Adjusted R-squared: 0.465 - Adjusted for the number of predictors.
• F-statistic: 52.56 - Indicates the model's overall significance.
• Prob (F-statistic): 1.81e-60 - Strong evidence that the independent variables collectively influence the dependent variable.

The model is statistically significant, demonstrating that the independent variables have a significant relationship with the drug overdose death rates.

Significant Factors:

1. Gender:
   • Female: Coef: -0.3594, P-Value: 0.014 (98.6% significance level). Being female is associated with lower death rates.
   • Male: Coef: 1.6966, P-Value: 0.000 (99% significance level). Being male is associated with higher death rates.

   The negative coefficient for females suggests that being female is associated with a decrease in drug overdose deaths. This aligns with data showing males generally have higher overdose death rates due to various factors including drug use patterns, healthcare utilization, and biological differences.

2. Race:
   • American Indian or Alaska Native: Coef: 2.3205, P-Value: 0.000 (99% significance level). Strong positive association with death rates.
   • White: Coef: 1.9881, P-Value: 0.000 (99% significance level). Strong positive association with death rates.
   • Asian or Pacific Islander: Coef: -3.0819, P-Value: 0.000 (99% significance level). Strong negative association with death rates.

   Being American Indian or Alaska Native is associated with an increase in drug overdose deaths, reflecting higher rates of substance use disorders and limited access to healthcare. Being Asian or Pacific Islander is associated with a decrease in drug overdose deaths, likely due to lower rates of substance use and strong cultural norms. Being White is associated with an increase in drug overdose deaths, attributed to the opioid epidemic and socioeconomic challenges.

Health Policy Impact

Prescription Drug Monitoring Programs (PDMP)

| Introduction of PDMPs Policy in Early 2000s

• Purpose: PDMPs are state-run programs aimed at monitoring the prescribing and dispensing of controlled prescription drugs to patients. These programs seek to identify and prevent drug abuse and diversion at the prescriber, pharmacy, and patient levels.
• Initial Impact: The introduction of PDMPs in the early 2000s aimed to control prescription drug use. However, the initial impact was not very pronounced.

| 2002 - Harold Rogers Prescription Drug Monitoring Program

• Purpose: In 2002, the U.S. federal government launched the Harold Rogers Prescription Drug Monitoring Program to support states in developing and implementing PDMPs.
• Support: This program provided federal funding and technical assistance to states, enhancing their ability to track and control prescription drug use.
• Impact: Following the introduction of the Harold Rogers Program, the overall trend in drug overdose deaths became less steep, indicating some improvement and a moderating effect of these policies.

Comprehensive Addiction and Recovery Act (CARA)

| 2013: Chinese Suppliers Turn to Fentanyl Production

• Event: Chinese companies began exporting large quantities of fentanyl to the U.S.
• Impact: Fentanyl, due to its high potency and low cost, replaced many illegal opioids.
• Trend: Rise in fentanyl-related overdose deaths.

| 2014-2015: Surge in Fentanyl-Related Deaths

• Event: Significant increase in fentanyl overdose deaths reported by states.
• Impact: Fentanyl's high potency and low cost rapidly replaced other illegal opioids.
• Trend: Steep increase in fentanyl overdose deaths.

| 2016: Passage of the Comprehensive Addiction and Recovery Act (CARA)

• Event: CARA was signed into law to address the opioid crisis.
• Impact: Expanded naloxone access and improved treatment services.
• Trend: Continued rise in fentanyl-related overdose deaths.

| 2017: Fentanyl Scheduled as a Controlled Substance

• Event: Fentanyl and its analogs were classified as controlled substances.
• Impact: Aimed to control manufacture and distribution.
• Trend: Slowed rise in fentanyl overdose deaths, moderating the increase.

CDC Opioid Prescribing Guidelines

| Introduction of CDC Opioid Prescribing Guidelines (2016)

• Event: Introduced in 2016 to reduce inappropriate opioid prescribing.
• Impact: Mixed effects observed in overdose death rates.

| Impact on Different Opioid Categories

• Heroin (Purple Line): Slight decline in overdose deaths, suggesting a reduction in heroin-related fatalities.
• Methadone (Green Line): Stable and low overdose death rates, minimal impact from guidelines.
• Natural and Semisynthetic Opioids (Light Blue Line): Slight decrease in overdose deaths, indicating a positive impact of the guidelines.
• Other Synthetic Opioids (Yellow Line): Sharp increase in overdose deaths, highlighting a shift to more dangerous synthetic alternatives like fentanyl.

| Shift to Synthetic Opioids

• Observation: Significant rise in deaths involving synthetic opioids (excluding methadone) post-2016.
• Impact: While guidelines reduced prescription opioid misuse, they inadvertently contributed to the shift towards potent and illicit synthetic opioids.

| Overall Effectiveness

• Positive Effects: Reduction in deaths from prescription opioids and heroin.
• Negative Effects: Continued rise in overall drug overdose deaths due to synthetic opioid-related fatalities.

| Conclusion

• Positive Impact: Reduction in prescription opioid and heroin deaths.
• Negative Impact: Increased fatalities due to synthetic opioids.
• Need for Comprehensive Strategies: Regulation of prescription practices and addressing the availability/misuse of synthetic opioids are essential.