Distilling down the work that my colleagues and I do as engineers is simple: we see something that can be improved, and we build a solution. Earlier this year, there was one problem that stood out above the rest for me: our platform was only allowing new patients to select “Male” or “Female” when signing up for our product.
As a company, Grand Rounds supports a diverse population of patients and works to provide them with the best and most supportive care. That being said, an estimated 0.3-0.6% of American adults identify as transgender, which translates to about 1.4 million people. At Grand Rounds, that means 18,000 (or more) of our covered lives may identify as transgender, without even considering those who describe themselves as gender-nonconforming.
Consider this in the context of the 2008 National Transgender Discrimination Survey, which found that “respondents in [the] study seeking health care were denied equal treatment in doctor’s offices and hospitals (24%), emergency rooms (13%), mental health clinics (11%), by EMTs (5%) and in drug treatment programs (3%)…19% of [the] sample reported being refused care altogether, due to their gender identity or expression, with even higher numbers among people of color.”
The next point to consider is that many transgender folks are not even out to their health care providers. In the same survey, only “twenty-eight percent (28%) of respondents said they were out to all their medical providers. Eighteen percent (18%) said they were out to most, 33% said some or a few, and 21% were out to none.” Part of this disparity may be because, according to the National LGBT Health Education Center, “Most clinicians do not discuss sexual orientation or gender identity (SO/GI) with patients routinely, and most health centers have not developed systems to collect structured SO/GI data.”
Transgender and gender-nonconforming folks have barriers to accessing health care, sometimes due to something as simple as not being able to share their gender on an intake form. A change to our intake process was necessary to align with Grand Rounds’ mission to democratize access to health care for all.
With so much work on our plates for all of our patients, we had taken some time putting this change on our roadmap. The issue was what I like to call the “Goldilocks Project Size”— it was still too small for the quarterly road mapping of larger projects, but had grown with our product over the course of the years to be too big for a quick bug fix. However, it was “just right” for…a hackathon! We recently hosted an internal hackathon for two days. These are different from public hackathons in that they generally focus on hacking on an existing product in an existing company. They’re a great space to foster creativity and also a valuable opportunity to boost camaraderie across teams. Goldilocks-size projects can be explored, and boundaries can be pushed, with the hope that at least some of the projects will eventually make it to production.
As soon as I heard about the hackathon, I knew what I wanted to work on. Over the two-day event, I collaborated with four other folks across engineering, data science, and product to design a solution and hack together an implementation to prove that this project was worth scheduling. The first thing we decided was that none of us was an expert in the subject, so for the first two hours, we pored over existing studies and research about improving gender selections on intake forms. We decided on the “one question” approach (versus two questions, which some other intake forms use) and set out to make it happen.
The implementation needed to focus on a balance between two needs: 1) patient needs, which are varied and nuanced, especially when it comes to gender; and 2) data mapping, which tends to be much more rigid and discrete.
The solution was to add a few more options to the selection list, but not too many. First, we wanted to explicitly include transgender folks, so we added Transgender Female and Transgender Male to the list. Then, we wanted to add an option for anyone who described themselves differently, so we added an option to self-describe, which then brings up a field to enter their chosen gender descriptor. Finally, we wanted to be sure that someone who would prefer to share their gender directly with their health care providers could do so, so we added an option to choose not to disclose.
We also needed to consider the various audiences for this information—how will we ensure that a patient is in control of who has access to the information about their gender? To prevent this information from reaching inappropriate audiences, we designed a mapping function to mask this information when sending reports to our employer-customers. Trans status is removed, and the other options are grouped together and discluded until a certain threshold is met.
After demoing at the hackathon, I was approached by several people who asked how soon this project could be implemented in production. Armed with positive reinforcement from all these folks, I was able to push upwards and put together a team to implement these changes in the actual product.
Finally, in the middle of June, the project had been prioritized, completed on an aggressive timeline, and announced to the public.
A single question has been preventing some people from sharing their full self with their doctor—a single question that could mean the difference between receiving appropriate care and not. If just one person finds solace in the fact that they can share their whole story with their doctor, just because we thought to ask, then we have succeeded.