User:Smerry1124/sandbox

The Medicare Provider Utilization and Payment Data: Part D Prescriber Public Use File (Part D Prescriber PUF) is a set of several data files that provides payment and usage data for healthcare providers reimbursed by Medicare Part D, the prescription drug component of Medicare. These files are provided to the public for free by the Centers for Medicare and Medicaid Services (CMS).

The Part D Prescriber PUF is publicly available for calendar years 2013 and 2014. Within the Part D Prescriber PUF, the detail file provides information about the drugs prescribed under Part D by each provider. Summary files are also provided that aggregate the data for each provider, each drug, each state, and the nation.

History

The Part D Prescriber PUF is one of eight suites of data files looking at payment and utilization that CMS provides for free, public use. The first of these suites of data files to be released was CMS - The Physician and Other Supplier PUF, which was originally released in April of 2014, following an executive order instituting an “open data policy.” This executive order stated that “the default state of new and modernized Government information resources shall be open and machine readable,” which may have been influential in the release of the Utilization and Payment Data.

The release of these data files was met with concern over its usefulness and interpretability by some groups and with appreciation of expanded transparency by other groups. Several studies used this PUF to examine variations in cost by specialty across geographies and a NY Times article used the dataset to show differences in the use of Lucentis and Avastin, two drugs that were reported to have similar clinical efficacy but different Medicare reimbursement rates.

The Part D Prescriber PUF is generally released two years following the end of the calendar year in which the services occurred. The datasets were first released in April of 2015. This file included prescriptions from calendar year 2013. The calendar year 2014 file was released in 2016. On August 8, 2016, the 2014 provider summary was updated to also include information that summarizes prescriptions for various drug categories: opioids, antibiotics, anti-psychotics, and drugs that are high-risk for elderly populations.

Data Availability

All data are available, for free, from the CMS website. The data include sets of files for calendar years 2013 and 2014. The data are provided in five zip files:

• PartD_Prescriber_PUF_NPI_DRUG – each line in this file shows prescription information organized by provider National Provider Identifier (NPI) and drug (i.e., unique combination of NPI, drug name, and generic name). That is, for each provider (represented by his/her NPI), cost and prescribing information is provided for each drug that they wrote for patients enrolled in Medicare Part D. This file includes provider information (name, address, specialty), drug name, and drug utilization (days supply, cost). This file is also referred to as the ‘detail’ file.
• PartD_Prescriber_PUF_NPI – each line in this file summarizes prescription information for a single provider (NPI). This includes additional information about each provider beyond what is found in the detail file, including entity type and Medicare enrollment. Drug use is described for various cohorts: low-income-subsidy status of patients, brand vs. generic, opioids, antibiotics, and high-risk drugs.
• PartD_Prescriber_PUF_Drug_St – each line in this file summarizes prescription information in a state for each drug (i.e., unique combination of state, drug name, and generic name). This file includes the same utilization information as the detail file.
• PartD_Prescriber_PUF_Drug_Ntl – each line in this file summarizes total, national, prescription information for a drug (i.e., unique combination of drug name and generic name). This file includes the same utilization information as the detail file.

The zipped files range in size from over 450 MB to under 1MB. The state and national summaries are provided in Excel workbooks, while the detail and prescriber files are provided in tab delimited formats and will require specialized software to open properly. The prescriber file is also viewable through an online interface that allows the user to filter and manipulate the file.

Publications

A 2017 article published by the Vermont Biz magazine detailed the opioid epidemic in Vermont and the Burlington Police Department's use the Part D Prescriber PUF. The BPD used the data to characterize the scope of Vermont's prescribing rates relative to other New England states. This information drove their development of new initiatives targeted at countering opioid abuse.

A 2016 article published online in BMC Infectious Diseases described the geographic variation in costs and rates of antibiotic prescriptions among Part D enrollees. The authors found that antibiotic use among Medicare patients was highest in the south and that Fluoroquinolones were the most commonly-used class of antibiotic on a national level.

A 2016 article from the Hawaii Journal of Public Health explored the relationship between the use of antibiotics and rates of Methicillin-resistant Staphylococcus aureus (MRSA). State level rates of prescriptions for methicillin-sensitive Staphylococcus aureus (MSSA) found in the Part D PUF were paired with rates of MRSA. This was used to determine whether outpatient use of antibiotics was associated with MRSA infections. The study found that there was a positive, statistically significant correlation between rates of antibiotic prescriptions and the prevalence of MRSA.

CMS created an opioid mapping tool based off these datasets. This is an interactive tool that allows the user to view data about opioid prescriptions at the state, county, and zip code level. The release of this tool has been noted by several media organizations and may lead to additional avenues of investigation.

Methodology

The Part D Prescriber PUF was created using the Chronic Conditions Data Warehouse. The data in this warehouse have completed the claims processing procedure to resolve adjustments and resubmittals (known as the final-action process). All Prescription Drug Event (PDE) records that were submitted by Medicare Advantage Prescription Drug Plans (MAPD) and standalone Prescription Drug Plans (PDP) (MAPD plans combine Part D coverage with a Medicare Advantage plan, whereas a PDP only offers Part D coverage) before the submission cutoff were included in the file; the cutoff is June 30th of the year following the end of the calendar year. That is, the calendar year 2013 file will include data submitted through June 30, 2014. The Part D Prescriber PUF contains records for all providers who submitted at least one prescription to a Part D plan. PDE records with National Drug Codes (NDCs) indicating over-the-counter usage (coverage status code = “O”) were excluded.

An NPI was used to identify the provider on each PDE record. NPIs were then linked to the National Plan and Provider Enumeration System (NPPES). NPPES includes provider-generated information including name, address, and specialty and this information is merged with the Part D Prescriber PUF. NPPES data can be updated by the provider at any time, and the NPPES information included on the Part D Provider PUF is extracted from the NPPES database at the end of the following calendar year. That is, the 2013 Part D Provider PUF uses NPPES extracted at the end of 2014.

Drug information is taken from First Databank’s MedKnowledge database and is linked to the Part D Provider PUF using NDC. Beginning in the 2014 file, PDE records with NDCs that do not link to this database are excluded.

Limitations

The Part D Prescriber PUF only includes records for Part D enrollees, so this file may not be representative of all prescriptions created by the provider.

Only drugs covered by Medicare Part D will be included in this file, although some Part D plans may cover additional drugs through supplemental coverage.

Medicare pays Part D plans a capitated, monthly amount for each enrollee. At the end of the year, a reconciliation process takes place to account for issues like unexpectedly high utilizers and for drug manufacturer rebates. It is therefore challenging, if not impossible, to assign a cost to a specific PDE. These datasets calculate cost as the sum of the ingredient cost of the medication, dispensing fees, sales tax, and any applicable administration fees.

The NPI may not be sufficient for recognizing individual physicians. Some physicians may issue prescriptions using both their individual NPI and an organizational NPI. Other physicians may only issue prescriptions using an organizational NPI.

For privacy reasons, any summaries with 1-11 PDEs are censored. For example, if a physician only wrote 8 prescriptions for opioids, the opioid count for this provider would be censored.