Birth Origins Limitations

Robert C. Bowman, M.D.

The birth origins data in the AMA Masterfile is listed in city, state, and country columns for 900,000 physicians. For allopathic graduates of medical schools in the United States over 97% of the physicians can be categorized although it took many months and required significant effort to understand the data and the geography involved. Over the decades there have been omitted data categories, handwriting interpretation errors, abbreviations, keystroke errors, and spelling errors that have been replicated. There are a variety of data formats. Sometimes the city is listed, but not the nation of origin. Other times the county is listed. This mean extra time searching until the author became more familiar with counties, cities, and other geographic listings. The improved searches and mapping noted in multiple internet sites, most recently Wikipedia and Google, also greatly aided searches and coding. Over the past 3 years the author worked to make the origins consistent.

There were a number of areas that limited coding.

The Masterfile data is less complete for certain Osteopathic schools, those outside of the United States, and certain populations such as graduates of schools in Puerto Rico and graduates of traditional black schools.
Coding for cities such as New York city can be problematic. The city column includes New York City, but at a smaller percentage. New York can also represent New York county, not a problem. But New York or NY can also mean born in New York state which can mean any location in the state. There are so many “New York” listings, that this becomes a problem in interpretation.
There also may be a tendency for students born in suburbs of a major city to list the major city as a birth origin. For coding in metro areas this is not a problem. However for calculations of admissions ratios by city or county this can be a problem. For example inner city locations may have very poor admissions ratios, but these are masked by outlying suburbs. To compare these ratios, multicounty regions are a better consideration and have been used in reports and articles.
Location categories sometimes did not include city, state, or country of origin. Where these were unique, they cold be filled in. If they were not unique, the data was listed as birth state only or missing data.
When a major city is on the border of a county line or extends into two or three counties, again this introduces the problem of county origins. By considering city level, population densities, income levels, or geographic divisions such as core metro or urban, the data is not a problem.
Puerto Rico also has different city and census organization and higher poverty levels and sometimes this data cannot be included since the codings are different. Distances between zip codes also are problematic as also for Hawaii and Alaska.
Those born in military bases involved a great variety. The bases included those that have been closed and those in other nations. Setting up a military birth category helped resolve some of the difficulties in coding.
Many small towns presented a challenge. Some were no longer listed on maps or in census information. Other towns had changed their names. Multiple internet searches resulted in solutions to most of these. Historical data was helpful for small towns, military bases, and towns with name changes.

Birth origins also are different than where students grew up or attended high school. Birth origins involving instate birth are similar, but different than students who are determined by a state or medical school to be instate. Born in a state and attending a medical school in a state implies longer connections with a state that appears to be confirmed in studies of retention.

Major Medical Centers - tables regarding physician distribution and definitions

Physician Workforce Studies

Birth Origins Articles