Medicaid data yields details on prescribing habits

by Christina Jewett

One psychiatrist in Chicago was prescribing a risky psychiatric medication to more patients than all the doctors in Texas, based on Medicaid prescribing data. That was a fact I gleaned while reporting an article for ProPublica that appeared in the Chicago Tribune. 

The sentence is simple and evocative. And I never could have written it if I hadn't navigated the world of Medicaid data. Doctors who bill the health insurer to the needy leave a mighty paper trial, and states vary in their ability and willingness to reporters follow it. For the valiant, though, the rewards are great. You can learn about prescribing, services rendered to patients and payments denied to doctors.

The data could shed further light on the work of a doctor who comes into the spotlight for questionable practices. Or it could provide a baseline for asking whether doctors who get payments from pharmaceutical companies – which are increasingly being made public – are unduly influenced by their sponsor. The data could also be a roadmap to identifying doctors whose practices are far outside the norm, if a norm is established.

I stumbled into this goldmine of data after getting tips that a Chicago psychiatrist, Michael Reinstein, was seeing an outsized number of patients. The tipsters also said that a surprising number of patients were prescribed a medication that carries five of the FDA’s “black box” warnings. What's more, he was being characterized in a product liability case as an opinion leader working for pharmaceutical companies and making dubious claims about one antipsychotic drug.

Getting prescribing data

My first goal was to find out his prescribing track record and glean some basis of comparison to determine whether he stood out from the norm.

I started in Illinois. I asked a spokeswoman for the state Medicaid agency to arrange an interview with an expert on their prescribing data. I soon learned that the agency is quite proud of a comprehensive and flexible data system. Technicians can query the system and spit out reports with information such as how many of which prescriptions a particular physician is handing out.

I soon sent a public records request for a spreadsheet that would rank the state's doctors in terms of how many antipsychotic drugs they prescribe each year. I defined the time frame I was seeking and what medications I sought data for, including generic and name-brand formulations of atypical antipsychotic medications. I also specified that for each prescribing doctor, I sought the number of prescriptions requested and numbers of patients prescribed to.

The data that came back was clean and clear. Illinois officials provided it in a simple spreadsheet broken down by doctor, year, medication and number of patients and prescriptions. The data revealed that the doctor of interest was the top prescriber of the drug I was most curious about.

What about elsewhere?

What I was missing, though, was context. So I made a similar request to several large states, including Florida, Texas, Pennsylvania, New York, California and North Carolina. States varied in their speed in responding, and Pennsylvania refused to supply information at all. But states that provided information quickly made it clear that Reinstein's prescribing habits were indeed unusual.

During the process of seeking data from states outside of Illinois, it smoothed and sped things along when I attached the Illinois spreadsheet in e-mails to other states. That way they could see what I was seeking, including content and format. It helped ensure that I would get information from other states that was comparable to what I had in hand. And, a student of human nature, I suspect it goaded their instinct not to be outdone by those folks in the other state.

As I waited for that data to start coming, I decided to call up the folks in Illinois again to see what other data I might request – and have a reasonable chance of receiving – that would help paint a picture of a physician’s practices. It turns out there was more. I learned that when doctors bill Medicaid, each service corresponds with a number (referred to as “ICD codes”) to simplify billing.

Reinstein's workload

I requested – and got – a year-by-year summary of all services rendered by Reinstein. This proved valuable to the story. Reinstein’s work was almost always billed as a brief visit with medication management. In a state audit, he described how long he tends to spend with each patient. He also wrote that about half of his patient load was Medicaid patients. Using that information, we were able to do simple math to see that his workload would take up far more than a 12-hour workday.

The state also tracks how many prescriptions and services it refuses to fund for each doctor. And for each refusal, state workers enter a code describing why the service or prescription was not funded. I saw that Reinstein was not paid on multiple instances because he prescribed to a patient who, state officials deemed, was dead.

With data in hand, the next step was finding experts and research that could better explain what it means. The Centers for Disease Control conduct a Health Interview Survey that establishes some norms in how doctors work. And research established that the drug I was most interested in, clozapine, is best reserved for patients with schizophrenia (about 1 in 1,000 people) who do not respond to drugs with fewer side effects (about 1 in 4). Obviously, this drug is meant only for the sickest patients.

Here’s an excerpt from a sidebar story published Nov. 10, 2009, in the Chicago Tribune by my esteemed colleague Sam Roe and I:

No psychiatrist in Illinois – or Texas, Florida and California, for that matter – has come close to Dr. Michael Reinstein in prescribing the antipsychotic drug clozapine to public aid patients, Medicaid records show.

At the request of ProPublica and the Tribune, Columbia University researcher Dr. Mark Olfson reviewed Reinstein's prescribing numbers. In 2005, the year Reinstein wrote the most clozapine prescriptions, Olfson said the number was 70 times greater than what would be expected of even a busy psychiatrist.

"A concern that arises when you have someone seeing an inordinate number of patients is: Do they have time to care for people?" said Olfson, who specializes in psychiatric practices.

Tracking pharmaceutical company payments

A second-day story focused on Reinstein’s work for AstraZeneca in promoting and studying the medication Seroquel. The data came in handy again, allowing us to compile a graphic showing AstraZeneca’s payments to Reinstein (gleaned from a lawsuit) on the same timeline as his annual rate of prescribing Seroquel. It showed that his prescribing generally rose and fell with company payments for speeches.

All told, though, this data has its limits. Most doctors see very few Medicaid patients, so when you seek data on a particular physician, you only see a fraction of his or her work.

Or alternately, some doctors who work at clinics serving mostly needy patients see an outsized number of patients, so it’s tricky to get a sense of proportion when viewing the data. Medicare, which renders care to those older than 65, refused to give me data linked to individual doctors unless I had a waiver signed by that doctor. And private insurance companies are unlikely to release such information.

Alas, I’ve talked to several reporters about working with this data since these stories were published. I’m now with California Watch, a project of the Center for Investigative Reporting. If you have questions, feel free to e-mail me at cjewett@californiawatch.org.