Tracking the Opioid Epidemic with New Indicators
With the opioid crisis continuing to take a heavy toll on communities across the country, our data team got to work and added new datasets to supplement the drug overdose data we already have, to help illuminate different aspects of the issue. We’ll have a number of posts over the next couple weeks dedicated to the opioid epidemic.
The new data includes detailed overdose information for both opioids and all narcotics, emergency department visits, hospitalizations, and prescription rates. We’ll go into each of these new datasets below.
Drug Overdose Deaths
The first dataset to look at is one that isn’t new to PolicyMap, drug overdose deaths. At a county-by-county level, this data from the CDC provides a look at overall drug fatalities, from 2000 to 2016. Though this data doesn’t focus on opioids exclusively, it combines death certificate data with statistical modeling, giving the most detailed look at the whole country.
Although statistical modeling means that the values are estimates rather than exact counts, it has the benefit of reducing the presence of data anomalies and missing data. Also, because of the use of statistical modeling, the data from CDC is provided in ranges, rather than precise numbers.
Opioid Overdose Deaths
Narrowing the data to opioids and narcotics (which includes opioids and cocaine, and their derivatives), we get more precise numbers, based on death certificate data collected by the CDC. Opioid overdoses includes overdoses by heroin, fentanyl and other synthetic opioids, oxycodone and other natural or semi-synthetic opioids, and methadone.
Because of inconsistencies in the ways different jurisdictions fill out and amend death certificates, some areas don’t include the specific opioid leading to the overdose. Opioid overdoses where no specific drug is specified are not included in the data’s count of opioid overdoses, leading to what may appear to be unexpectedly small values. However, these overdoses may be included in the count of unspecified narcotics overdoses, so we’ve also made available data on overdoses due to unclassified narcotics and unclassified drugs. If you’re looking at an individual county, you should make sure to check these indicators as well, as, in some jurisdictions, a majority of overdoses might not be included in the more specific data. (We’ll have more on this in a future blog post.)
This detailed overdose death data is most useful where the epidemic is most pervasive. Due to CDC suppressions, and the aforementioned death certificate issues, many counties do not have data for these indicators.
The drugs at the center of the epidemic include both legally prescribed medications like oxycodone, and illegal drugs such as heroin. Often, those who have been prescribed legal medications become dependent, and are forced to begin purchasing those drugs illegally without a prescription, or to turn to heroin or other opioids. Though only a fraction of opioid users fall into addiction, looking at the prescription rate can show which areas may be more likely to face high levels of addiction.
This data from the CDC, which goes from 2006 to 2016, represents the rate of prescriptions written per 100 people. It’s important to keep in mind that many individuals may receive multiple prescriptions in a year, so this does not represent the percent of people receiving opioid prescriptions. Well over 20% of counties in the country have rates of over 100 prescriptions per 100 people. This helps show the depth of the issue, where opioids may be being prescribed less judiciously.
A change in time indicator is also available in the legend to show year-to-year trends in prescriptions. This is particularly useful, as it shows how the epidemic has hit different areas of the country at different times, and that the solution to the problem has started earlier in some areas than others.
Emergency Department Visits and Hospitalizations
Data on emergency department visits and hospitalizations related to opioid use is available, broken down by patient age, patient income level, and the rural/urban spectrum of a patient’s zip code of origin.
This data, which comes from the Healthcare Cost and Utilization Project (HCUP), is available at the state level, from 2005 to 2015. Note that some years have data for more states than others.
Over the next couple weeks, we’ll take a closer look into trends seen in the data, and things to keep in mind while looking at these datasets.