How data analytics combats the opioid epidemic

Ask your friends and neighbors what the leading cause of accidental death is in America and some will likely answer guns while others guess car accidents. Both groups, however, would be incorrect.  

BielinskiAccording to the CDC, opioid abuse far outstrips them both. In 2014, the latest year for which there are figures, drug overdoses accounted for more than 47,000 deaths, versus less than 33,000 each for cars and guns. In fact, while motor vehicle-related deaths have declined over the last 10 years and gun-related deaths have increased slightly, deaths due to opioid abuse have risen sharply-and show no signs of leveling off.

While some of that number can be attributed to the use of heroin and other illicit drugs, the reality is that much of the increase is coming from the abuse of legally prescribed opioids. Spending on these medications alone among Medicare Part D beneficiaries increased 156% from 2006 to 2014-a period when deaths due to accidental overdose increased by roughly 40%. While the word “epidemic” is often overused, in this case it is an apt description of the crisis brought on by opioid abuse in America.

Yet as concerning as this is, finding the 1% to 5% of members, prescribers, and pharmacies who are abusing the system in the midst of the hundreds of millions of legitimate pharmacy claims processed each month is no small task. Especially when those attempting to discover it are relying on the manual processes and vast amounts of data.  

The solution is next-generation analytics that rely on multiple data points-more than humans can process at one time-to surface purchasing and prescribing patterns with a high probability of abuse. Following is how they are addressing this epidemic.

Identifying drug-seeking behavior among members

Analytics that use color-coded dashboards can assign scores based on risk factors so those performing the analysis can focus on the most likely cases of fraud, waste, and abuse (FWA) while minimizing false positives. Scores are based on pre-set thresholds such as members who are seeing more than 10 physicians or filling prescriptions at more than 10 pharmacies. The thresholds can be set based on payer or pharmacy benefit manager (PBM) preferences, or industry benchmarks.

Where it gets challenging is being able to discern legitimate reasons for these patterns, such as an oncology patient who is receiving multiple prescriptions from several different specialists. Next-generation analytics help by bringing in additional data, such as showing the locations of prescribers, pharmacies, and the patient’s home on a map (geospatial analytics). Clustering in one location is likely to be normal, while filling several prescriptions at locations far away from the patient’s home can be a strong indicator of FWA. Overlay this with consumer information, such as income, population, socioeconomic status, etc., and the picture becomes more clear. 

By automating the process through the intelligent application of analytics, payers and PBMs can focus their efforts on the most likely perpetrators of FWA, reducing costs while ensuring not to alienate members in good standing.

Internal server error