Gender Identity and Access to Primary Health Care

2025-03-01

2026-04-30

The primary outcome variables include:
Whether or not an appointment request receives any response (via email or phone) from a primary care provider
Whether or not a request receive an offer of an appointment (via email or phone) from a primary care provider
Whether or not a request receives a negative response (practice is full, not taking new patients)
Whether or not a request receives an ambiguous response (neither a definitive offer for an appointment nor a rejection) -- for ambiguous responses possibilities include being asked for follow-up information or being asked to complete a form or call a different number without a specific offer of an appointment.

To assemble a nationally representative set of primary care providers, we first generated a random sample (with replacement) of all U.S. counties, with the probability of selection proportional to the county’s population, as measured by the 2020 Census. Within each selected county, we ran standardized Google searches combining the county and state name with the terms “family medicine,” “primary care,” or “internal medicine,” along with “contact” and “@” to find providers with either publicly accessible email addresses and/or patient portal contact forms. Each time a county was selected, we randomly selected two eligible providers per county, resulting in a final sample of 992 providers across 49 states plus the District of Columbia. The field period ran from November 2025 through March 2026. Of the 992 providers, 74 percent were contacted via a website patient portal or contact form and 26 percent via direct email. Each provider received exactly one message from one treatment arm, a between-subjects design with no repeated contact.
To test for potential discrimination according to gender identity, the treatment varied a single element of an otherwise identical appointment request: the relationship between the patient’s legal name and chosen name. Each provider randomly received one of four standardized messages. The exact message sent to each provider is as follows:

“Hello, I am looking to schedule a new patient primary care appointment. My legal name is [LEGAL NAME], but I go by [NICKNAME]. You can reach me at [PHONE NUMBER] or [EMAIL]. Thank you.”

To disentangle any effects of being transgender from general effects of going by a nickname, all individuals indicate that they have a chosen name that is different from their legal name. Prospective patients with a legal/chosen name that is gender-congruent (Stephen/Steve or Margaret/Margo) signal as potentially cisgender. Those with a name pair that is gender-incongruent (Stephen/Stephanie, Margaret/Mark) signal themselves as potentially transgender. The message is identical across all four arms except for the legal and chosen names.

Randomization done in office by a computer

United States county

Yes

500 counties

1,000

250 control, 250 treatment arm 1, 250 treatment arm 2, 250 treatment arm 3