What data from 205 million private health insurance claims reveals about America’s opioid crisis
In 2012, there were roughly a quarter of a million privately insured patients diagnosed with opioid use disorder in Amino’s insurance claims database (241,000 to be exact). By 2016, there were 1.4 million. That’s a 6-fold increase in opioid use disorder diagnoses in just four years.
These numbers tell an all-too familiar story. The opioid epidemic currently sweeping the United States has crippled communities and killed thousands, and it has quickly become the worst drug crisis in American history.
The causes and solutions to this crisis are complicated, and often deeply intertwined. Sometimes it's political—a debate over medication-assisted treatment. Sometimes it's economic—a crisis hitting hardest in rural areas. And sometimes it's medical—a botched surgery followed by an addictive prescription.
These issues have played out hundreds of thousands of times in households across the country. And they've played out in doctor's offices and hospitals, too.
Amino’s database of commercial health insurance claims gives us unique insight into those interactions. To better understand the crisis, we analyzed 205 million private health insurance claims involving patients diagnosed with “opioid use disorder”—a recently updated classification that considers severity of addiction and removes the distinction between “abuse” and “dependence.”
You can read our full methodology here—but first, a few big takeaways:
1.4 million privately insured patients were diagnosed with opioid use disorder in 2016—6 times more than in 2012.
Kentucky is home to 9 of the top 10 counties nationwide for doctors treating a high volume of patients for opioid use disorder.
Patients diagnosed with opioid use disorder are much more frequently diagnosed with a host of other general health issues (like hepatitis C and chronic pain), behavioral and mental health issues (like alcoholism and depression), and back and spinal issues (like failed back syndrome and herniated discs).
America’s opioid crisis, by the numbers
In 2005, the CDC reported that 14,918 Americans died of drug overdoses related to opioids. In 2015, 33,091 opioid overdose deaths were recorded—more than double over 10 years. In fact, the New York Times recently estimated that overdose deaths involving all types of drugs likely exceeded 59,000 in 2016—the largest annual jump in overdose deaths ever recorded (there were 52,404 in 2015).
We analyzed 205 million commercial health insurance claims involving patients diagnosed with opioid use disorder and found that diagnoses jumped from 241,000 patients in 2012 to 1.4 million in 2016—a 6-fold increase over just four years.
It’s important to note that these are privately insured patients, which shows just how pervasive this crisis has become. Opioid use isn’t just a problem for Medicaid—many of these 1.4 million patients are on health insurance plans sponsored by their employers.
Alarmingly, the CDC warns that the opioid crisis is likely underestimated—even their own estimates are just the tip of the iceberg (see top chart on Figure 1 above). They noted that drug deaths may be listed as unexplained or “lack [the] correct ICD-10 codes to be reported by statewide opioid death surveillance,” as they found was the case in Minnesota.
But there’s another reason for organizations to be underestimating the problem—particularly among the commercially insured population. Patients may be hesitant to get treated for addiction because of the stigma associated with opioid use disorder, especially if an employer is sponsoring their health insurance plan. Therefore, many patients who are affected by opioid use disorder are not likely to be accounted for in analyses of health insurance claims.
According to Dr. Anna Lembke, a Stanford University psychiatrist and opioid addiction expert, opioid-related diagnoses are highly stigmatizing. “Patients don’t want to carry them on their charts, and doctors don’t want to stigmatize their patients,” said Lembke. “But they will go ahead and chart it if there’s utility in it. And the utility is you can’t get buprenorphine, methadone maintenance, or naltrexone paid for by a third-party payer unless it’s diagnosed.”
Medication-assisted treatment (MAT) is the use of medications (like buprenorphine, methadone maintenance, and naltrexone) in conjunction with counseling and behavioral therapies to treat opioid use disorders and prevent opioid overdose. MAT is widely accepted as improving treatment outcomes for opioid use disorder—but each medication works differently. Naltrexone (Revia, Vivitrol) works by blocking the effects of opioids, especially in cases of overdose. Methadone and buprenorphine are opioids with unique properties that make them safe and effective for treating opioid use disorder. They work by occupying receptors in the brain, tricking them into thinking they’re receiving an opioid without producing the “high” caused by such drugs.
Buprenorphine is relatively new to the market (released in 2002) and, unlike methadone, primary care physicians in the U.S. can prescribe it and patients can receive it at any pharmacy (not just specialized clinics, like in the case of methadone). Because of this, buprenorphine offers the unique opportunity to improve access to treatment for opioid use disorder across the country.
However, the opioid crisis has not swept the nation evenly. There are certain regions that have been inundated with high volumes of patients struggling with opioid use disorder. Below, we take a closer look—county by county.
Where do doctors see the most patients for opioid use disorder?
To identify where the opioid crisis is worst, we looked at every doctor in the U.S. in a clinically relevant specialty who diagnosed patients with opioid use disorder in 2016. Then, we focused our analysis on the doctors who saw the highest volume of opioid use disorder patients and mapped their practicing locations, producing a density figure measured in doctors per 100,000 residents.
The result paints a picture that should look very familiar to anyone who has studied the opioid crisis in America—in areas where the crisis is at its worst (such as New Mexico, Kentucky, Appalachia, and Florida), we see the highest concentrations of doctors treating high volumes of patients for opioid use disorder.
In the map above (Figure 2), a few areas stand out—particularly Kentucky, home to 9 of the top 10 counties nationwide for doctors treating a high volume of opioid use disorder patients:
Top 10 counties by density of physicians that treated a high volume of patients for opioid use disorder in 2016
|County||Physicians per 100,000 residents|
|Breathitt County, Kentucky||51.5|
|Franklin County, Georgia||45.2|
|Lee County, Kentucky||42.1|
|Wolfe County, Kentucky||41.3|
|Perry County, Kentucky||39.2|
|Boyle County, Kentucky||34.0|
|Floyd County, Kentucky||33.6|
|Menifee County, Kentucky||31.5|
|Bourbon County, Kentucky||30.0|
|Rowan County, Kentucky||29.7|
New Mexico also stands out (see Figure 2). This is no surprise—the state has one of the nation’s highest drug overdose rates and recently passed a law requiring all of its officers to carry overdose kits. Albuquerque, N.M. has been added to a growing list of cities receiving federal aid for its opioid and heroin crisis.
Counties in Florida are similarly affected. The state is home to Palm Beach County, sometimes referred to as the “Recovery Capital of America.” Local officials estimate that the city of Delray Beach, Fla. (with only 67,000 residents) has more than 800 treatment facilities.
Health insurance claims reveal where doctors see the most patients for opioid use disorder. Tweet
Experts agree that access to doctors who can treat opioid use disorder is crucial, but barriers to treatment remain. The Substance Abuse and Mental Health Services Administration (SAMHSA) reported in 2015 that “half the people who could use treatment are not able to get it, in large part because they can’t afford it or can’t find providers.”
Even if a patient does have access to a doctor who can help them, they may not have access to the medication-assisted treatment (MAT) that many consider a crucial component for recovery.
“I definitely think that we need more access to treatment with buprenorphine,” said Lembke. “There [are] wait times and incredible inconvenience for people to get buprenorphine, even when insurance has agreed to pay for it. But buprenorphine won’t solve the whole problem.”
Any potential solution to the crisis is complicated by the fact that opioid use disorder often comes hand in hand with a slew of other medical issues.
Opioid use disorder is linked to many serious health issues
To understand the complex relationship between opioid use disorder and other health conditions, we used Amino’s database to compare medical diagnoses between patients diagnosed with and patients not diagnosed with opioid use disorder over a 3-year period. This allowed us to see three major categories of conditions that were diagnosed more frequently among patients who were also diagnosed with opioid use disorder (versus those who were not diagnosed with opioid use disorder). (See our methodology below for more information.)
- General health issues like hepatitis C and chronic pain
- Behavioral and mental health issues like alcoholism and depression
- Back and spinal issues like “failed back syndrome” and herniated discs
General health issues
Patients diagnosed with opioid use disorder are frequently diagnosed with common diseases found among intravenous drug users, like hepatitis C (9.1x more frequent) and HIV (2.6x more frequent).
However, we saw lots of co-occurring diagnoses related to pain, as well. Chronic pain was 5.8 times more frequently diagnosed among these patients, as well as osteomyelitis (2.3x), fibromyalgia (2.1x), and myalgia (2x).
We cannot make causal inferences with this data, but there are certainly documented instances where patients are prescribed opioids for a painful medical condition and only then become addicted.
Behavioral and mental health issues
Behavioral health issues like alcoholism and binge drinking were 8.4x and 5x more frequently diagnosed among patients who were also diagnosed with opioid use disorder, while mental health issues like suicidal ideation and post traumatic stress disorder were 6.9x and 4.2x more frequently diagnosed.
Back and spinal issues
“Failed back syndrome” (or post-laminectomy syndrome) represents a group of chronic pain conditions that follow back surgeries, usually high-risk procedures that are conducted to relieve lower back pain. Patients diagnosed with opioid use disorder were 7.2x more frequently diagnosed with failed back syndrome, 3.3x more frequently diagnosed with inflammatory back conditions, and 2.9x more frequently documented as having a pre-existing arthrodesis (a surgical immobilization of a joint by fusion of the adjacent bones—in this case, most likely spinal fusion).
"Perhaps failed back syndrome is a risk factor for developing an opioid use disorder." —Dr. Anna Lembke
“It’s known that people with co-occurring behavioral and mental health issues are at high risk for addiction even when prescribed opioids for a bonafide prescription medical use,” said Lembke. “What I thought was really interesting was the correlation with failed back syndrome. Perhaps failed back syndrome is a risk factor for developing an opioid use disorder—and that could be part of the reason why this community experiences such chronicity and lack of improvement. This is a subgroup that’s especially vulnerable to opioid misuse.”
Amino’s perspective on America’s opioid crisis
Health insurance claims can tell us a lot about the opioid crisis—where it’s spreading, how doctors are combating it, and what kind of medical issues patients face. However, while data can inform good solutions, too often it fails to help those on the front lines.
Amino’s aim has always been to connect people to the best care possible—no matter what they need. If you or someone you know is struggling with addiction, you can use Amino below to find a doctor who has treated a high volume of patients diagnosed with opioid use disorder.
There are also many other resources available to you—here are just a few:
Substance Abuse and Mental Health Services Administration: SAMHSA is a government organization that offers a 24-hour, free, confidential helpline that provides referrals to local treatment facilities. Call 1-800-662-4357 or visit their website to find a treatment center or locate a buprenorphine prescriber online.
National Council on Alcoholism and Drug Dependence: NCADD provides information, help and guidance for people struggling with drug use disorder and their families. They have a network of more than 90 affiliates that offer addiction support and recovery services across the country. Call their 24-hour affiliate referral line at 1-800-622-2255 or visit their website for more information.
National Institute on Drug Abuse: Part of the NIH, the National Institute on Drug Abuse provides comprehensive information on the opioid crisis online. Visit their website to read more about opioid abuse and find clinical trials in your area.
If you are experiencing an emotional crisis or thinking of suicide, call the National Suicide Prevention Hotline at 1-800-273-8255. You are not alone.
We identified patients diagnosed with opioid use disorder (as defined in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition) by looking for claims in our database with any of the following International Classification of Diseases, Ninth Revision (ICD-9) codes, either as the primary diagnosis or the secondary diagnosis, between 2012 and 2016.
|ICD-9 code||Code description|
|30400||Opioid type dependence, unspecified|
|30401||Opioid type dependence, continuous|
|30402||Opioid type dependence, episodic|
|30403||Opioid type dependence, in remission|
|30470||Combinations of opioid type drug with any other drug dependence, unspecified|
|30471||Combinations of opioid type drug with any other drug dependence, continuous|
|30472||Combinations of opioid type drug with any other drug dependence, episodic|
|30473||Combinations of opioid type drug with any other drug dependence, in remission|
|30550||Opioid abuse, unspecified|
|30551||Opioid abuse, continuous|
|30553||Opioid abuse, in remission|
Statistics for drug overdose deaths involving opioids were collected from the CDC WONDER database using the methodology outlined in Rudd RA, Seth P, David F, Scholl L. Increases in Drug and Opioid-Involved Overdose Deaths — United States, 2010–2015. Morbidity and Mortality Weekly Report, 2016.
To determine counties with “high-volume doctors,” we consulted two sources of information—Amino’s database of private insurance claims and the American Community Survey’s five-year population estimates for 2015.
For every doctor in our database with a clinically relevant specialty (internists, family practitioners, gynecologists, psychiatrists, general surgeons, cardiologists, pain medicine doctors, and physiatrists), we counted every patient they diagnosed with opioid use disorder (as defined using the ICD-9 codes above) in 2016. We defined “high volume doctors” as those who were in the 95th percentile for patient diagnoses in 2016—in other words, these doctors saw more patients than 95% of similar doctors nationwide. The cutoff for the 95th percentile was determined to be 31 patient diagnoses.
We then determined all of the practicing locations for these doctors by consulting Amino’s doctor database. We summed the doctors by county, and divided the sum by the county’s population to produce the density figures seen in Figures 2 and 3, represented as high volume doctors per 100,000 residents.
To determine relationships between opioid use disorder and other conditions, we first defined a “condition” as a clinically relevant grouping of ICD-9 codes, with a common name given as a label. Where applicable, we mapped ICD-10 codes to ICD-9 codes to keep these groupings consistent.
Next, we collected every health insurance claim in our database that occurred between January 1, 2014 and December 31, 2016. We computed the relative frequencies of conditions observed among patients diagnosed with opioid use disorder and compared them to the relative frequencies of conditions observed among patients who were not diagnosed with opioid use disorder. Next, we created a relative frequency ratio by dividing the relative frequencies of conditions observed among patients diagnosed with opioid use disorder by the relative frequencies of conditions observed among patients not diagnosed with opioid use disorder.
Example: Opioid use disorder and hepatitis C. Consider the following example of determining the relationship between opioid use disorder and hepatitis C. First, we computed the relative frequency of hepatitis C among patients diagnosed with opioid use disorder in our database. To do this, we counted the number of patients diagnosed with hepatitis C and opioid use disorder and divided that number by the total number of patients diagnosed with opioid use disorder during the timeframe defined above.
Next, we computed the relative frequency of hepatitis C among patients who were not diagnosed with opioid use disorder by repeating the same process, this time excluding patients diagnosed with opioid use disorder. Then, we divided the relative frequency of hepatitis C among patients diagnosed with opioid use disorder by the relative frequency of hepatitis C among patients without opioid use disorder to produce the relative frequency ratio. A ratio greater than 1 suggests that hepatitis C is more frequently diagnosed among patients with opioid use disorder vs. those without. In fact, we found that hepatitis C is 9.1 times more frequently diagnosed among patients with opioid use disorder.
We acknowledge that our data and analyses conducted on it have several limitations, including:
We cannot make causal inferences with this data. For example, we cannot say that opioid use disorder leads to hepatitis C (or any of the conditions listed in Figure 4).
We are only observing patients who sought care for opioid use disorder. Furthermore, we are observing these patients using commercial health insurance claims. This creates an inherent bias. We acknowledge that patients who seek care for opioid use disorder may inherently be more likely to receive a diagnosis for the conditions listed in Figure 4. We acknowledge that the lack of an observation of opioid use disorder in our data is not a guaranteed signal that the patient lacks the condition entirely.
Research and copyediting by Olivia Marcus