Data

Downloads

All of the statistics presented on this website can be downloaded in Formatted Excel or Unformatted Excel files. If you use this data to generate new results (such as visualizations, tables, figures or statistics), we ask that you cite the Immigrant Doctors Project as the source of the data. The data is released under a CC0 license to permit the widest possible reuse.

Source and Representativeness

The data underlying our results come from Doximity, an online networking site for doctors. Doximity assembles the data from a variety of sources, including the American Board of Medical Specialties, specialty societies, state licensing boards, and collaborating hospitals and medical schools. Doximity data has been verified to be highly reliable and has been used in research published in leading peer-reviewed journals, such as the Journal of the American Medical Association (e.g. Jena et al 2015).

The Doximity data cover 1,005,419 physicians, representing virtually all practicing physicians in the United States; every doctor assigned a National Provider Identifier by the Centers for Medicare and Medicaid Services is included. We restrict our analysis to the subset of observations in the Doximity data with both current zip code and country of medical school information, leaving us with a final sample of 827,522 doctors.

Classifying Doctors from Affected Countries

Doctors are classified as immigrants from an affected country based on the country in which they went to medical school.

Although not a perfect measure of citizenship of the affected countries, we believe this to be a useful proxy. We discuss in more detail why this definition is likely to provide a conservative estimate of the total number of affected doctors in the FAQ.

Location Data

Doximity contains data on the address of a doctor’s practice. We group doctors based on the commuting zone (CZ) of their practice address. Commuting zones are groups of adjacent counties that have close economic ties; for instance, 7 counties in the eastern part of Massachusetts make up the Boston commuting zone.

Rounding and Binning

As part of the conditions laid out by Doximity for use of the data, we cannot publish statistics that could identify individuals or exact counts of doctors. For this reason, all of the statistics that we report have been rounded or categorized into ranges.

Statistics Presented on the Map

Concentration of Doctors

Each commuting zone (CZ) is put into one of five equal-sized groups (quintiles) based on the fraction of doctors in the CZ who are from affected countries. The CZs with larger fractions of doctors from affected countries are colored in darker shades of blue on the map.

Number of Doctors from Affected Countries

Hovering over a CZ on the map displays a range for the number of doctors from affected countries in that CZ. Each CZ is put into one of nine groups based on the number of affected doctors in that CZ, and the range of numbers for each group is shown.

Appointments Provided

Hovering over a CZ on the map also displays an estimate of the number of appointments provided to patients each year by doctors from the affected countries. We arrive at this number by multiplying the number of doctors by 2,000 appointments per year. This estimate is based on research by Hannah Neprash (2016, Table 2), who finds that an average doctor serves just over 40 appointments per week (40 appointments/week x 50 weeks worked/year = 2,000 appointments/year). Estimates from other sources, such as the 2016 Survey of America's Physicians, generate similar estimates for the number of appointments provided by doctors.

Doctor Shortages

We characterize commuting zones as having a shortage of doctors if the population of the CZ is more than 3,500 times the number of doctors with an internal medicine specialty in the CZ. This definition corresponds closely with one of the main criteria used in the federal Health Professional Shortage Area (HPSA) designation: whether the ratio of population to primary care providers exceeds 3,500. Population information is obtained from the 2015 American Community Survey. We obtain similar results if we instead classify CZs based on whether they are in the bottom fifth (quintile) of CZs by the ratio of doctors to population.