Are Medicare Prescriptions Fueling Purple Drank? An Iodine Data Story

An Iodine data analysis shows that drugs intended for American seniors under Medicare may actually end up in party cocktails.

Turns out you can learn a lot about Medicare prescriptions — and its unexpected link to hip hop — if you just follow the data.

The Centers for Medicare and Medicaid, better known as CMS, is the checkbook that pays for nearly $600 billion in Medicare spending in the U.S. Every pill that is prescribed to any of the country’s 36 million seniors or disabled people covered under Medicare goes through CMS, by way of Medicare Part D. That creates a trove of data, data that was released, following a Freedom of Information Act request, to Pro Publica last year.

The data reveal some fascinating patterns among prescribers and practitioners — how some providers are signing off on thousands of suspect prescriptions, how billions of dollars are wasted on brand-name prescriptions, and other insights.

What’s more, Pro Publica has opened the data to exploration by other parties, such as our data team at Iodine. Recently, we churned through the data to see what else it could tell us. And quite unexpectedly, we soon found ourselves parsing Three 6 Mafia lyrics and TMZ reports on Lil’ Wayne.

Our analysis indicates that even as abuse of prescription medication becomes front-page news, the problem extends far beyond pain killers and Adderall. The research raises suspicions that there is an illicit trade even in innocuous drugs like cough medication — and that Medicare might be a significant source of the supply, via so-called drug diversion, where prescribing doctors or pharmacists arrange to sell medications on the black market, or patients re-sell their own prescriptions.

According to our research, there is not only a significant drug diversion problem going on with Medicare prescriptions (a problem that has been well accounted in a recent Pro Publica report published in the Pacific Standard), but it is likely that drugs prescribed to American seniors under Medicare are actually ending up in party cocktails – particularly promethazine, a drug that is prescribed to reduce nausea and motion sickness. Promethazine is often combined with codeine in a prescription cough syrup.

Promethazine with codeine is a key ingredient in the illicit brew known as Lean, Sizzurp, or Purple Drank. Mixed with Sprite soda and Jolly Rancher candies and served in styrofoam cups, the concoction is a popular party drink and a frequent reference in hip hop lyrics. The drink is especially popular in the South, having originated in the Houston area and spread to Atlanta and other southern areas. It often leads to abuse, as in Lil’ Wayne’s case.

Lil’ Wayne poses with a bottle of promethazine with codeine

(The fact that Justin Bieber has also recently been connected to Purple Drank suggests that the drink’s popularity may indeed have maxed out).

The trade in promethazine as a component of Purple Drank has been well documented for the past decade. In 2011, the Drug Enforcement Agency arrested four Houston men, accusing them of running a $10 million ring in contraband promethazine syrup.  But our analysis appears to be the first indication that Medicare prescriptions could be a significant source for at least some of the illicit trade.

We began with a download of the CMS data. Our team then normalized the data so it represented prescriptions for a drug per 100,000 prescriptions in each state. Since the data is formatted by state, we were looking for data with strong geographical patterns.

One drug had an especially strong geographic signal: promethazine. In 10 states, promethazine is prescribed at five times the national average. The distribution is largely regional: the states with higher prescriptions are all clustered in the south, with Texas, Tennessee, North Carolina, Georgia and Kentucky ranking the highest. Texas has the highest, with a stunning 229,000 prescriptions written for the drug — 10 times the number of prescriptions written in 24 states.

Screen Shot 2014-03-11 at 3.43.41 PM
The states with highest claims and cost of promethazine prescriptions, from Pro Publica.

In itself, this doesn’t tell us much: many drugs have regional variations, such as cyproheptadine and hydrocodone. So our data team wanted to find ways to bring more meaning to this dataset. What else could we learn about this data and the pattern it was revealing?

The first step was to understand that promethazine is a highly abused drug, usually in its cough syrup preparation, particularly in the hip hop community. Several prominent rappers have died of overdoses of the mix, and last year Lil’ Wayne was admitted to the hospital in connection with drinking Lean.

The next step was to tap Google Correlate. Correlate is like Google Trends in reverse – you give it real world activity data with a pattern, and it gives you search terms which have that pattern across time or place. (For details, see the Google Correlate Whitepaper or Comic Book!) In this case, we uploaded the normalized regional pattern of promethazine prescriptions.

The top search result from this pattern was “gg 225”, the imprint for the pill form of promethazine, with a Pearson correlation coefficient of 0.877 (perfect correlation is 1.0).

correlate-cms-imprintGoogle Correlate maps that show a strong correlation between states with high volume of
promethazine prescriptions and web searches for both “promethazine” and “gg 225”. Full results

This was both surprising and impressive to us: Correlate searched over millions of candidate search terms, and the top result was the imprint for the pill. This was also surprising as other drug activity from this CMS dataset did not surface the respective imprints. When doing a search on google for “gg 225”, the related searches at the bottom of the search results page give you a sense of the user intent around this search.

Screen Shot 2014-03-13 at 10.44.27 AM

Then we went further and looked at how Google searches for “promethazine” correlate with searches for “drank”. The data shows a .71 regional correlation – a strong signal indeed (by comparison, the correlation between “promethazine” and “nausea” – the drug’s primary approved indication – is .79, and follows the same highly regional pattern).

Screen Shot 2014-03-11 at 4.18.53 PM


To be sure, our research is simply suggestive, a chain of correlations from CMS data to search data to patterns of drug abuse. It is important to make clear that we did not search for nor did we find any direct evidence that the Medicare promethazine prescriptions behind this odd pattern are ending up in styrofoam cups. What we have is a significant regional anomaly in prescriptions and a known pattern of abuse that overlaps with the same regional variation. It’s worth noting here that Actavis, one of the major manufacturers of the flavored purple syrup that went into Purple Drank, appears to have recently stopped making the product (news that made pharamceutical blogger The Angry Pharmacist extremely happy — see his NSFW rant).

But there’s one more obvious question to ask: Is there a legitimate use for promethazine for seniors, a plausible reason for such high prescription volume among the elderly in the deep South? Here another data set comes in handy: The list of “potentially inappropriate medications for older adults” issued by the American Geriatric Society, the professional association for healthcare pratitioners of geriatric medicine.

Promethazine appears in the first group of drugs listed, under a one-word recommendation: “Avoid.”

One thought on “Are Medicare Prescriptions Fueling Purple Drank? An Iodine Data Story”

  1. The “purple drank” is news to me. As a child of an 87 y.o. mother, I’m concerned that she may have Promethazine proscribed for her at some point. I appreciate knowing it’s on the “Avoid” list.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s