On a sweltering weekend in August last summer, first responders in Cincinnati, Ohio, a bustling mid-sized American city on the Ohio River, were summoned 30 times to assist citizens who had overdosed on heroin and needed medical attention. Seventy-eight more overdoses and at least three deaths were reported during a 48-hour period on the first two days of the following week. By the end of the day on Thursday, 174 people in total had been treated; 3 had died.
Lab testing found that the batch of heroin responsible for the staggering number was cut with carfentanil, a tranquilizer for large animals, like elephants, that is 10,000 times as strong as morphine.
In order to get a better handle on this and future crises, the City of Cincinnati began analyzing emergency medical services (EMS) response data to identify trends and geographic "hotspots," helping public safety identify key areas for strategically deploying personnel and medical resources. Said Chief Data Officer Brandon E. Crowley:
Before we can respond effectively we need to know just what we're dealing with. We need data, presented in an analytical way that helps us target our limited resources. So we built an application for it. Deployed to the web with a mobile-friendly interface, the "Heroin Overdoses in Cincinnati" page is an integral part of our city's interactive dashboard. Informally, it goes by another name: the Heroin Tracker.
The Heroin Tracker is a subset of EMS data which captures responses to reported heroin overdose incidents. The dataset helps the city dispatch roving medics and increase public safety and public health response in hotspots while it also advises on trends. Automatically updated nightly from the 911 call and public safety dispatch database, the Heroin Tracker displays overdoses by day and time, neighborhood and ambulance transport (if any), along with total overdose rates across a 13-month span. Writes Crowley:
The data is also precise enough for practical action. Every day, dozens of public-safety and health-department officials, along with Talbert House (a fantastic nonprofit offering substance-abuse treatment and other services) check the Heroin Tracker. It helps them plan more targeted patrols in neighborhoods with the most serious overdose hotspots and anticipate overdose spikes based on previous trends.
The EMS dashboard is part of a larger effort by the city to embrace data gathering and sharing as part of becoming a Smart City. Since 2014, the city has been focused internally, to uncover the wealth of data at its disposal and how to use it. It has now launched of a wide range of interactive public dashboards featuring real-time city data. These 15 new dashboards contain various datasets, and are now accessible 24/7 on the city’s new CincyInsights webpage, as well as through the city’s website and open data portal. Anyone is able to interact with, and easily analyze mapped data using filters such as neighborhood location, date, activity type and more.
The dashboards take existing city data already found on the Open Data Cincinnati portal and translates the content into user-friendly visualizations. Each is organized according to the Administration’s five strategic priorities: Safer Streets, Growing Economy, Thriving and Healthy Neighborhoods, Innovative Government and Fiscal
Dashboards range from heroin overdoses to in-progress road projects. You can even check the location of snow plows in real time. Other interactive datasets include: crime, emergency response, potholes, registered vendors and blight reduction.
Data visualizations have become a key weapon in the struggle to keep both policymakers and the public informed in real-time as the overdose crisis continues to worsen. The National Association of Counties released an interactive map last year that shows the staggering amount of overdoses nationwide and tracks opiate providers and filled prescriptions. The map, created by Esri, allows users to hone in on a specific county and compare numbers against national averages.
Monitoring prescription abuse
While Cincinnati has not yet implemented predictive modeling with its heroin incident dataset, analysts for the State of Ohio have begun using Prescription Drug Monitoring Program (PDMP) data, which Ohio launched in 2010, to predict opioid risk patterns throughout the state. The state combines data on known overdose deaths with prescription data from the Ohio Automated Rx Reporting System (OARRS), which tracks the dispensing of controlled prescription drugs to patients Ohio, to build predictive models as early detection for overdose.
PDMPs are state databases that gather information from pharmacies on dispensed controlled substance prescriptions. All states except Missouri have implemented such programs within the past 20 years, allowing registered users--typically physicians and dentists, as well as pharmacists, law enforcement agencies and medical licensure boards-- to identify patients at high risk for obtaining multiple prescriptions from multiple sources. They also spot providers who may be engaged in aberrant prescribing.
The ultimate goal of PDMPs is to track patients' drug histories to prevent “doctor shopping” (going from provider to provider obtaining different prescriptions), identify people with substance abuse problems and refer them to treatment. Unfortunately, their usefulness varies widely and the rules are inconsistent. For example, 29 states require doctors to use them, while 20 make doctors' usage of the databases voluntary. In 2012, fewer than 50 percent of prescribers were using them. That seems to be changing. There have been more than 100 bills introduced this year regarding PDMPs, according to Heather Gray, legislative director at the National Alliance for Model State Drug Laws.
One major study found that if every state maintained strict standards for its database--such as requiring reporting and monitoring more drugs--there would be 600 fewer opioid overdose deaths per year. Investigators analyzed 10 years of data from the National Ambulatory Medical Care Survey, which collects information on patients, office visits and clinicians, such as reasons for particular visits and related diagnoses. Their analysis revealed a more than 30 percent drop in the rate of prescribing Schedule II opioids (such as Oxycontin and Percocet--the most addictive class of painkillers—in two-dozen states with such monitoring programs. Yuhua Bao, associate professor of health-care policy at Weill Cornell Medical College and one of the study’s researchers, says:
The reduction is encouraging because it reflects providers’ growing awareness of opioid misuse and abuse, and also suggests that they are moving away from prescribing drugs that have the highest risks of abuse and dependency.
The key to success for PDMPs is mandatory use. Ohio requires doctors to access OARRS, and the state’s Board of Pharmacy requires pharmacists to use it before dispensing medication.
The seeds of the drug overdose epidemic began in the late 1990s when 15 states loosened restrictions on synthetic narcotic painkillers and doctors began writing more prescriptions for drugs like oxycodone, hydrocodone, and methadone. It is probably not a coincidence that Perdue Pharma released oxycontin in 1996, doubled its sale force, and two years later oxycontin accounted for 80 percent of the company’s profits. This kind of “success” gets noticed and other pharmaceutical makers began pushing their painkiller products. For a detailed look at how Big Pharma was complicit in the manmade opioid crisis, see Christopher Bowe’s excellent Fixing Pharma’s Incentives Problem in the Wake of the U.S. Opioid Crisis in the Harvard Business Review.
Since 1999, the amount of prescription opioids sold in the U.S. nearly quadrupled, yet there has not been an overall change in the amount of pain that Americans report. In West Virginia, where I grew up, drug wholesalers dumped 780 million hydrocodone (the generic name for Vicodin) and oxycodone (the generic name for OxyContin) pills between 2007 and 2012. That's 433 pills per resident and I have to assume not everybody uses it so some people are getting extras. By the time, the Feds started cracking down a substantial portion of the population was already addicted.
Street dealers saw an opportunity and moved in with much cheaper heroin ($10 a fix instead $80 for an oxycodone to be smashed up and injected). The result is that West Virginia now leads the nation in overdoses. Parents are literally dropping dead at their daughters' softball games. The National Institutes of Health estimate that prescribed opiates were the gateway to addiction for close to three-quarters of new heroin users.
In this public health disaster, knowledge really is power. Big data is now producing positive results in three key ways:
- Allowing policymakers and concerned citizens to track and visualize the who, what, where and when of the problem in real-time and plan appropriate responses to save lives;
- Monitoring abuse of prescriptions by patients, doctors and pharmacies’ and taking immediate corrective action. As Ohio has demonstrated, mandatory compliance by medical providers is essential to success;
- Providing evidence of what treatments work and what don’t. Big Data studies by organizations like Pew and the National Institute on Drug Abuse have found that Medication-assisted treatment (MAT)—a combination of psychosocial therapy and the FDA-approved medications methadone, buprenorphine, or naltrexone—is the most effective intervention to treat opioid use disorder (OUD) and is more effective than either behavioral interventions or medication alone. The hang up on this approach is that many Americans wrongfully believe it is “just substituting one drug for another” and object on moral or religious grounds.
Undoing the damage done to individuals and society by both legal and illegal opioid abuse in the first two decades of the 21t century will take years and cost billions of dollars. Congress made a good start last year by passing the 21st Century Cures Act, which received wide bipartisan support in Congress and was signed into law by President Obama. The legislation invests billions in medical research, and will give $1 billion to the states to help in the fight against opioid addiction and improve access to recovery programs.
Several states are now suing the major drug companies involved in fueling the opioid epidemic. You can expect that data analysis will play a huge role in assessing liability in those cases as they move forward in the months and years ahead.
This whole mess is a sordid chapter in the history of American capitalism and, as usual, no one at the upper levels is going to jail.