A Quantitative Analysis of Impacts of the Opioid Epidemic

West Virginia, 2006-2014

IPEC 4396 / 04.30.2020 / [Lydia Nelson, Kathleen McNatt, Ty Williams, Kristin Kolb, Avirut Mehta]

Literature Review

What is known?
Conflicting information?
Remains to be learnt?

Previous Consensus

Opioids, even those taken as prescribed, are highly addictive and dangerous, having resulted in a loss of about 5.5 million QALY between 1990 and 2017 in the United States (Massey, 2017).

Links exist between major mood disorders and opioid addiction at the biochemical level (Peciña, 2019).

Current opioid abuse interventions fail to account for effects on mental health, e.g. through counseling and psychotherapy (Phillips, 2017).

Conflicting Information

Does the correlation between mood disorders and opioid use at the biochemical level extend to trends at a societal scale, measurable by demographic factors? (Halbert, 2017)

Do other socioeconomic factors, like population trends and income, play a role in determining such relationships? (Campbell, 2010)

Can interventions targeting potential underlying socioeconomic factors be effective at resolving the issue of the opioid epidemic? (Keane, 2016)

Areas to Investigate

Quantifying correlations between opioid epidemics and mood disorders will allow for an understanding of whether known biochemical mechanisms scale to societal trends.

Analyzing epidemic intensity against socioeconomic trends will provide evidence for or against the effect of certain underlying factors in driving the epidemic.

Resulting conclusions can help drive decisions about appropriate interventions for those suffering from opioid addiction.

Our Question

What observable trends and correlations are present between...

...at the county level in West Virginia between 2006 and 2014?

intensity of the opioid crisis
population
income
unemployment rates
suicide rates

Rationale

Approximately 1.7 million US citizens suffer from opioid abuse issues, with tens of thousands dying a year from overdoses.

This problem has been exacerbated by abuse of prescribed opioids.

Mental health and socioeconomic factors surrounding opioid abuse are yet to be fully understood.

Primary Factors to Investigate

Opioid distribution and suicide rates form the backbone of our investigation.

We use quantities of distributed opioids as a proxy measurement for intensity of the opioid crisis within a given county.

We use suicide rates as a proxy for the prevalence of mental health issues and mood disorders within a region.

Secondary Factors to Investigate

We additionally analyze population trends as an indicator of underlying region-wide status changes (e.g. drops in population may signify decreasing quality of life within a region over time).

Per capita income and unemployment serve as additional indicators for the socioeconomic health of the counties we analyze.

Area of Investigation

We chose to investigate West Virginia given its position in the center of the broader nationwide opioid crisis - WV is the single most affected state.

Given data availability and time constraints, we chose to perform a county level analysis across all factors, measuring each value as its mean or sum over a year, for every year between 2006 and 2014.

Hypothesis

Given previous literature, we expect to find relatively strong correlations between opioid distribution and suicide rates, as well as unemployment and opioid distribution.

We may find weaker correlations between population trends and opioids, as well as per capita income and opioid distribution or sucide rates.

Collected Data

Analysis & Results

(graphs and charts are interactive)

Avg. % Change in Population

(2006-2014)

Increase in population in a small handful of counties, with a decrease in many others, particularly Southern counties,

Avg. % Change in Unemployment

(2006-2014)

General year over year increase in unemployment not accounting for external factors (e.g. recession).

Avg. % Change in Income/Yr.

(2006-2014)

Overall slow and steady increase in per capita income per year in every county, not accounting for inflation.

Avg. % Change in Qt. Opiods/Yr.

(2006-2014)

Overall increase in quantity of opioids sold per year in many counties, with a small decline in a handful of counties.

Avg. % Change in Suicides/Yr.

(2006-2014)

Overall increase in suicides per year per capita in almost every county, with a stronger trend in northern counties.

On average, population remains relatively constant from year to year.

Unemployment trends are not linear and likely driven by other external factors.

On average, per capita income increase steadily from year to year.

On average, suicide rates remain pretty steady within counties from year to year.

No correlation of statistical

significance between

opioid distribution and suicide rates

r² = 0.000825

No correlation of statistical

significance between

unemployment rates and suicide rates

r² = 0.010

Conclusions

We reject our hypothesis, having observed no trends of strong statistical significance between opioid distribution, suicide rates, population trends, income, and unemployment.

Although these trends may exist and are even likely to exist given current literature, limitations within our analysis of the data - discussed further later - prevent us from observing these patterns within our compiled dataset.

Limitations

A primary limitation of our study was the use of suicide rates as a proxy for prevalence of mood disorders and mental health issues. Particularly in rural counties, low absolute numbers of suicides prevent this metric from serving as an effective measure of the overall prevalence of mood disorders such as depression.

We note additional limitations in our ability to perform analysis on other factors, such as uninsured rates or hospital stays, given both sparse data availability and time constraints.

Future Investigation

To create a more robust understanding of how the opioid crisis may have driven mental health disorders, future investigations should focus on compiling and analyzing data directly pertaining to mood disorder prevalence.

Future investigations should additionally include other factors mentioned in previous literature, such as uninsured rates, quality of education, and healthcare utilization.

Supplement: Methods Overview

Data was imported from sources listed in references and analyzed using Python's numpy and pandas packages in a Jupyter environment with Google Colaboratory.

Data visualizations were created with plotly.py and seaborn/matplotlib, using GeoJSON files provided along with plotly.py.

This presentation was created using reveal.js and slides.com, and then hosted on GitHub Pages.

Supplement: Data

Sanitized datasets used to generate our visualizations are available at www.github.com/avirut/fortyfive.

These include individual HTML files containing each visualization used in this presentation separately.

This presentation is hosted at www.avirut.me/fortyfive as well as www.slides.com/avirut/fortyfive.

References - Data

Opioids Dataset (DEA ARCOS, WashPo)

- https://www.washingtonpost.com/wp-stat/dea-pain-pill- database/summary/arcos-wv-statewide-itemized.csv.gz

Unemployment Dataset (Bureau of Labor Statistics)

- https://www.bls.gov/lau/#cntyaa

Per Capita Income Dataset (Bureau of Economic Analysis)

- https://apps.bea.gov/regional/downloadzip.cfm

Suicide Rates & Population Dataset (CDC Wonder)

- https://wonder.cdc.gov/ucd-icd10.html

References - Literature

Opioid Overdose Outbreak - West Virginia, August 2016

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5657780/

Trends in Opioid Use, Harms, and Treatment

http://www.ncbi.nlm.nih.gov/books/NBK458661

Endogenous Opioid System Dysregulation in Depression: Implications for New Therapeutic Approaches

https://www.nature.com/articles/s41380-018-0117-2

Mahali, Saghar Chahar, et al. "Associations of negative cognitions, emotional regulation, and depression symptoms across four continents: International support for the cognitive model of depression." BMC psychiatry 20.1 (2020): 18.

References - Literature (cont.)

Halbert, B. T., Davis, R. B., & Wee, C. C. (2016). Disproportionate longer-term opioid use among U.S. adults with mood disorders. Pain, 157(11), 2452–2457. doi: 10.1097/j.pain.0000000000000650

Campbell, C. I., Weisner, C., Leresche, L., Ray, G. T., Saunders, K., Sullivan, M. D., … Korff, M. V. (2010). Age and Gender Trends in Long-Term Opioid Analgesic Use for Noncancer Pain. American Journal of Public Health, 100(12), 2541–2547. doi: 10.2105/ajph.2009.180646

Keane, H. (2016). Facing addiction in America: The Surgeon Generals Report on Alcohol, Drugs, and Health U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES, OFFICE

Muntaner, Carles, and Elisabeth Barnett. "Depressive symptoms in rural West Virginia: labor market and health services correlates." Journal of health care for the poor and underserved 11.3 (2000): 284-300.