Reading a provider’s public record
The data year describes when the reported activity happened. A release date describes when the publisher made a file available; a load date describes our copy. These dates are not interchangeable. The dataset status below is the place to check our recorded coverage; missing load information does not establish that a source is current.
NPI registration is not years in practice
The enumeration date is when an NPI was assigned, as defined in the CMS NPI rule. Time since that date is not clinical experience or proof of continuous practice. Specialty labels follow the taxonomy selection described below.
Part B: Original Medicare services
The CMS Part B dataset covers Original Medicare fee-for-service claims, generally excluding Medicare Advantage and excluding Medicaid and private insurance. For 2020 and 2021, CMS paid for COVID-19 vaccinations of Medicare Advantage enrollees through Original Medicare, so counts for those years can include those enrollees. Service counts are not patient counts or a quality score.
Part D: prescriptions and refills
The CMS Part D methodology describes filled prescriptions, including refills, for people in stand-alone drug plans and Medicare Advantage drug plans. Prescribing volume is not a quality score or a recommendation to take a drug.
Missing and small counts
We do not display patient or beneficiary counts below 11 or reconstruct suppressed values. CMS suppresses Part B data representing fewer than 11 beneficiaries. In its Part D detailed file, CMS also omits prescriber-drug rows with fewer than 11 claims and suppresses small beneficiary counts. Missing values are not zero. Service, prescription and Open Payments record counts use the published data; we do not apply a blanket below-11 rule to those counts.
Open Payments: financial relationships
These records describe payments and transfers of value reported by companies. CMS explains that a reported relationship is not necessarily improper. A payment is not proof of wrongdoing or a measure of care quality; its program year and payment category provide context.
How our figures are calculated
Active and deactivated NPIs
An NPI counts as active if NPPES has never deactivated it, or has reactivated it on or after the deactivation date. A deactivated NPI keeps its page, marked as deactivated, but it is not counted and is not included in directory lists. NPPES does not publish the reason for a deactivation.
Specialty of a provider
Each individual provider is listed under one specialty: the taxonomy code marked as primary in NPPES, otherwise the first one listed, skipping codes that NUCC defines for organizations only. Specialty names come from the NUCC code set.
Provider counts
Counts on the home, state, city and specialty pages are active NPIs grouped by the practice address in NPPES. Different spellings of the same city name are combined. Organizations (hospitals, clinics, pharmacies, labs, suppliers) are counted separately from individual practitioners.
Population and clinicians per 100,000 residents
Active individual (Type 1) NPI records with a practice address in the area, divided by the number of people who live there, times 100,000. A specialty rate counts only the clinicians listed under that specialty. Population is the U.S. Census Bureau’s American Community Survey (ACS) 5-year estimate, table B01003 (Total Population), for 2020-2024. A 5-year estimate describes the whole period, not the population today.
The U.S. rate is the same calculation for the 50 states and the District of Columbia. Puerto Rico has its own rate and is compared with the U.S. rate, but is not part of it. For a specialty, the U.S. rate covers every NUCC code listed under that specialty.
City counts use the city name in each clinician’s practice address, which is a postal name. Population comes from the Census place with the same name (for some cities a town or township, a New York City borough or a Puerto Rico municipio). The postal city and the Census area can cover different ground, so a city rate is an estimate. A city without a matching Census area has counts but no rate.
ACS estimates are published by the U.S. Census Bureau. As a work of the U.S. federal government they are in the public domain in the United States (17 U.S.C. 105); every page that uses them names the Census Bureau and the table as the source. Each estimate has a margin of error, which we do not show; for a small place it can be large compared with the estimate.
A rate is compared with the U.S. only when the area has at least 10 active clinicians in the group, a city has at least 10,000 residents, and the U.S. group has at least 100 clinicians. A higher rate does not mean shorter waits or better care, and a lower rate does not by itself mean a shortage.
How a rate is compared with the U.S. rate
The comparison uses the ratio of the area’s rate to the U.S. rate before rounding. A rate less than 10% above or below the U.S. rate is described as close to it. From 10% up to 25% above or below, the page gives the difference in percent. At 25% or more above, it gives the rate as a multiple of the U.S. rate (for example, “about 1.5 times”); at 25% or more below, as a percentage of the U.S. rate. On a city page for one specialty, the city rate is also set against the state rate when the state can be compared: within 10%, it is called close to the state rate; otherwise the state rate is given.
The summary on a state or city page names up to three specialties with at least 1.25 times the U.S. rate and up to three with 75% of the U.S. rate or less. A specialty is named among those with fewer clinicians only if its U.S. rate is at least 5 per 100,000 residents. When no specialty qualifies and at least three specialties can be compared, all between 75% and 1.25 times the U.S. rate, the summary says that their rates are close to the U.S. rate.
Medicare assignment and telehealth shares
On state and city pages, the Medicare assignment share is the part of the area’s active clinicians in the group who are listed in the CMS Doctors and Clinicians file and whose record says they accept Medicare assignment. Clinicians listed as “may accept assignment” are not counted as accepting it. The telehealth share is the part of the same clinicians whose record carries the CMS telehealth indicator. Clinicians who are not in the file are not part of either share; when fewer than half of the active clinicians are in the file, the page says so. Shares are shown when at least 10 clinicians of the group are in the file. A city share is compared with the state share, and a state share with the U.S. share, when that group also has at least 10 clinicians in the file; a difference of less than 5 percentage points is described as similar.
A doctor’s profile shows the same two indicators for the clinicians listed under the doctor’s specialty with a practice address in the same city, but out of all active clinicians in that group: a clinician without a Doctors and Clinicians record counts as not accepting assignment and not having the indicator. These shares are shown only when the city group has at least 11 clinicians. Because the groups are counted differently, a profile share and a city page share for the same specialty can differ.
Medicare assignment applies to Original Medicare only. The telehealth indicator marks clinicians whose recent Medicare claims include telehealth services; it does not mean that every visit can be done online.
Comparisons with the same specialty in the city and the state
A doctor’s profile compares “Medicare Part B services”, “Medicare-allowed amount”, “Paid by Medicare”, “Medicare Part D prescriptions” and “Part D drug cost” with other clinicians listed under the same specialty: one group in the same city and one in the same state. A group includes only active individual (Type 1) NPIs with a practice address there; organizations, deactivated NPIs and profiles hidden at the provider’s request are left out. A clinician without a value for a measure is left out of that measure only.
Each measure uses the latest data year of its CMS dataset, the same year for everyone in the group, and each comparison line shows that year and the number of clinicians with data. If the provider’s latest CMS data are from an earlier year, the provider is not compared, and the profile says which year the comparisons use.
A line says “More” (“A higher” for dollar amounts) than X% of the group when X% of the other clinicians in it have a strictly lower value, and “Fewer” (“A lower”) than X% when X% have a strictly higher value. Clinicians with the same value as the provider count on neither side. X is rounded down to a multiple of 5 and is at most 95. A line says more or fewer only when X is at least 60; otherwise it says the provider is close to the middle of the group. The group’s median and middle half (25th to 75th percentile) are shown with each line, and the small scale next to it shows only what the sentence says.
A group is compared only when it has at least 11 clinicians with data, the provider included; the city and the state group are checked separately. When only the city group is too small, the profile shows the state comparison and says so. These comparisons describe the amount of Medicare billing and prescribing, not quality of care.
Medicare patients on a doctor’s profile
The number of Medicare patients is the total beneficiary count (the CMS field Tot_Benes) in the Medicare Physician & Other Practitioners - by Provider dataset for the provider’s latest Part B data year: the different people with Original Medicare the provider billed Part B services for, each counted once. It is not the sum of the Patients column of the services table, where CMS counts a person once for every service code they received. If CMS does not publish the count for the provider, or that year of the dataset is not loaded, the profile shows no patient figure rather than an estimate. Patient counts are not compared with other clinicians.
Years since graduation
The current calendar year minus the graduation year listed for the provider in the Doctors and Clinicians file. It is not the number of years in practice.
Submitted charges, allowed amounts and Medicare payments
The services table on a provider page shows, for each service code, the average submitted charge and the average Medicare allowed amount as CMS publishes them. The yearly Medicare-allowed amount and Medicare payment on the profile are estimates from the published averages: for each service code, the average amount times the number of services, added up for the year, which is why they are shown as “about” a dollar amount. Providers set their own submitted charges, while Medicare pays according to its fee schedule; the difference is not a discount or a saving for patients.
Yearly totals of Medicare and Open Payments data
Other totals for a year on a provider page, such as services, prescriptions, drug cost and industry payments, are sums of the rows CMS publishes for that provider and year. Rows that CMS redacts or suppresses for privacy are not in the data, so the totals can be lower than the provider’s actual activity. Dollar amounts are shown as published, not adjusted for inflation.
Changes across years
Trend text uses up to five years with published values for the same measure. A missing year is not treated as zero. The endpoints describe recorded billing or prescribing, not improvement or decline in care. Changes in coverage, reporting and privacy suppression can affect the series; see the Part B and Part D source limitations.
What we do not calculate
We do not calculate quality scores, rankings or ratings of providers from these datasets. Star ratings on a profile come only from moderated patient reviews.