Tuesday, June 23, 2015

Garbage in, garbage out: The Medicare Part D prescriber database

by Christian Sinclair

A colleague of mine recently brought to my attention his listing of all the medications he prescribed in 2013 on the Medicare Part D prescriber database. What's that you say? You have not heard of this tool to find out what medications your doctor prescribes?

At the end of April 2015, Medicare (CMS) released a database of cumulative prescriptions by provider from the calendar year 2013. Sites like ProPublica took the raw data and setup a clean interface that allows you to sort the data by state, drug prescribed and provider, among other nifty features.  (Similar to what happened with Physician Medicare billing data in 2012.)

The reason for releasing all this data per CMS:
This new dataset provides key information to consumers, providers, researchers, and other stakeholders to help drive transformation of the health care delivery system. This data enables a wide range of analyses on the type of prescription drugs prescribed in the Medicare Part D program, and on prescription drug utilization and spending generally.
No worries, patient information is not shared, but what does this really tell us about physicians, and in particular physicians who practice in hospice and palliative care settings?

Admittedly, some of the national and state level data is interesting.  Knowing the average number of drugs prescribed by Family Medicine (74.9) is higher than Internal Medicine (65.7) could be used for some fun ribbing about who knows the most about what. Seeing that Nexium ($2.5B), Advair ($2.3B) and Crestor ($2.2B) are the three most costly medications in terms of total expense helps demonstrate the power of advertising.

But what is the purpose of looking up an individual doctor and their prescribing patterns?  Does this help the public figure out quality? No. Could this be used to find outliers? Sure, but why does everyone's data need to be published to find the outliers? The Forbes 100 list doesn't need to publish the net worth of every person to figure out who the outliers are. Please understand, I am fine with finding physicians consistently prescribing unsafely, but that can be done with CMS and state boards working together and doing investigations.

The real problem with this data is that it is woefully incomplete and without context to answer the questions some people are asking it. Looking at one individual physician's data, one finds it difficult to understand.  Let's take a look at some examples for our field of Hospice and Palliative Medicine (HPM). Even more important since we tend to prescribe a lot of opioids and benzodiazepines for complex symptom control. Oh why did I cross that out? Because the reason and appropriateness of a prescription cannot be gleaned from this database. Unfortunately, someone may look at your data and consider you a 'pill mill' doc without that context. That's not right.

Some problems with the database:

  • It lists only one specialty for each provider. So if you are a family medicine doctor who does a lot of hospice or palliative care, no one will be able to tell with this database because it may likely list you as just a family medicine doctor.
  • Specialties are listed multiple times. Looking at the National List of specialties, 'Hospice and Palliative Medicine' is listed 9 times (which links to 94, 94, 16, 10, 4, 4, 3, 2, and 1 for # of physicians) totaling to 228 HPM physicians in the USA. That cannot be right. in 2013, there were over 6,000 board certified HPM docs and this database only found 228 of them? 
  • Lots of physicians are missing. I looked up 10 doctors I commonly work with in outpatient and inpatient settings and found listings for only 4 of them. How can you compare someone to their peers if you are missing such a large number? Looking for dirt on me? Too bad, I'm not in the database, because I was in a non-clinical role in 2013.
  • Scripts attributed to NPs and PAs? Guess who gets the credit for those scripts? I nmany states the supervising physician does. So if you supervise the practice of 2 NPs then you may have 3x the prescribing rate as your peers. But you can't tell that from this database.
  • Hospice Medical Directors prescribe under the Medicare Hospice Benefit. So all the comfort kits, opioids, benzodiazpines prescribed appropriately for dying patients on hospice will not be included in this database of Part D prescriptions.

So at least tell your physician friends to check out their own name (modern physician equivalent of 'Googling' yourself), so you know what people are seeing about you. If you don't like what you see you may not have much recourse, but the next time you start hearing about physician level data being published you may want to have a voice in the matter.

Overall, I find this database to not be very helpful except looking for national and state trends and with future database releases looking at trends over time. Maybe it will help us understand how REMS may change the available number of prescribers of certain opioids. Recent publications trying to shame doctors who prescribe large amount of certain medications are disheartening to read on many levels, because when it comes to this database, the numbers are not the whole story. Garbage in, garbage out.

Christian Sinclair, MD, FAAHPM is editor of Pallimed, and practices palliative medicine at the University of Kansas Medical Center. He is a fan of data, statistics and science, in addition to appreciating the value of starting with good data and informed context.

Image credit: Garbage In, Garbage Out by Austin Kleon via Fickr and Creative Commons License

Pallimed | Blogger Template adapted from Mash2 by Bloggermint