Trans People Remain Largely Invisible in Health Data

Dr. Tonia Poteat, who teaches epidemiology at the Johns Hopkins Bloomberg School of Public Health, delivered the CROI’s first-ever plenary session on transgender people.

Dr. Tonia Poteat, who teaches epidemiology at the Johns Hopkins Bloomberg School of Public Health, delivered the CROI’s first-ever plenary session on transgender people.

The annual Conference on Retroviruses and Opportunistic Infections is the top HIV science and medical gathering in the US, so it’s not surprising that the federal Centers for Disease Control and Prevention (CDC) chose this year’s CROI as the place to unveil new data estimating the lifetime risk of HIV based on a person’s race and ethnicity, age, sex, and state of residence.

The analysis — which estimates that about one of every two black gay men and other black men who have sex with men and one in four Latino MSMs will acquire HIV — garnered a lot of attention from the mainstream media at the February conference in Boston.

But transgender people went unmentioned.

However, in December, the CDC released data confirming that transgender women had the highest rates of HIV in a sample of more than five million HIV testing events carried out in 2012-2013.

CDC verifies high HIV rates, but experts agree standard collection methods fail

The non-recognition of transgender people is a common practice in research, according Dr. Scout, director of HealthLink (formerly the Network for LGBT Health Equity) at CenterLink, the national association of LGBT community centers. In particular, population-level data like the lifetime risk numbers depend on having a “denominator” — an estimate of the total size of a particular population, in this case transgender people. Without that number, it’s impossible to create risk calculations parallel to other populations in the CDC’s report.

“There is no population denominator for trans folk,” Scout explained. “But the feds create denominators, and they chose not to create a trans one by never testing and adding trans surveillance questions.”

Scout added that the federal government says it is “urging” trans data collection by states in the Behavioral Risk Factor Surveillance System (BRFSS), meaning that each state must individually decide to adopt the questions.

Tonia Poteat, PhD, assistant professor at the Department of Epidemiology at the Johns Hopkins Bloomberg School of Public Health, delivered the first-ever plenary session that focused on transgender people at CROI 2016. In a follow-up interview, she said it is unclear why the federal government cannot mandate inclusion of gender identity questions in the BRFSS.

“I think it would be very useful to look at many other surveys that already ask ‘sensitive’ questions, like the National Survey of Family Growth, which asks some pretty detailed sexual behavior questions,” she noted. “I can’t imagine it would be too sensitive to include gender identity. It would also be very helpful, from a population size estimate perspective, to include gender identity on the national census.”

“We’re starting to do a better job of collecting transgender information in surveillance data so we will be able to have diagnosis numbers, but the problem is we don’t have the population denominator as part of our calculation,” confirmed Kristen Hess, PhD, a CDC official who presented the lifetime risk data at CROI 2016.

Indeed, the CDC includes a two-step process for gender in the National HIV Surveillance System, which reports statistics on HIV diagnoses, in which respondents report not just their sex assigned at birth but also current gender identity. As reported by journalist Liz Highleyman in, using the two-step method in HIV testing forms resulted in a 69 percent increase in identifying transgender participants over just asking one question about gender.

The inclusion of the two-step question allowed CDC to produce the analysis verifying high rates of HIV-positive testing results among transgender women. However, integration of this method has been uneven.

“Parts of CDC have been really progressive in using the two-step method, and there’s other parts [of CDC] that have not even heard of the two-step method,” Poteat said. “At the [National Institutes of Health], it’s the same thing. Different institutes are still differing in how they approach the questions and if they feel it’s relevant.”

In CDC’s online fact sheet “HIV Among Transgender People,” the agency notes that “identifying transgender people within current data systems can be challenging.” While the agency goes on to recommend the two-step method, the fact sheet’s authors appear to focus primarily on transgender people themselves as the reason for why accurate accounting poses challenges — rather than on other factors cited by advocates, such as explicit or implicit assumptions made by those filling out the forms or the inherent limitations in using just one question to accurately assess a person’s gender identity and experience in a brief surveillance form.

“Some transgender people may not identify as transgender due to fear of discrimination or previous negative experiences,” the fact sheet reads. “Since some people in this community do not self-identify as transgender, relying solely upon gender to identify transgender people is not enough. Gender expression may fluctuate for some transgender people due to issues such as perceived safety or reluctance to identify as transgender in certain situations.”

Comprehensive change needs to start at the top, Poteat said, noting that Dr. Francis S. Collins, director of the National Institutes of Health, held an LGBT meeting more than a year ago to set an agenda for research on sexual and gender minorities. While Poteat is seeing some follow-through on how data is collected, she said, there’s much more to be done.

“The best case scenario would be a combination of specific studies that recruit sizeable numbers of trans people to address pressing health concerns in the community — HIV, mental health, etc. — as well as broad inclusion and visibility of trans people in national — or local — population-based data collection efforts,” she said.

In the meantime, how should researchers and public health entities like the CDC characterize the lack of transgender-specific data in their published or presented analyses, rather than leaving trans people unmentioned?

At a press conference following the conference presentation of the lifetime risk data, the CDC’s Hess confirmed that the analysis did not account for transgender people and that her slides and presentation had not noted this nor included it as one of the limitations of the study.

“We can do a better job of that in the paper to make it clear,” she acknowledged.

As of late March, the CDC had not yet published the lifetime risk assessment. However, Scout, who serves on a research advisory panel to NIH, suggested what the agency may say in its article: “We are working to develop measures and add them to federal surveys that allow us to identify and quantify the risks for trans people. All available literature shows risks for trans women, especially trans women of color may be higher than any other single population.”

Poteat stressed that ideally public health and science will reach the point where it’s no longer acceptable to publish data that doesn’t disaggregate findings by gender identity.

“However, until then, it’s most appropriate for researchers to either clarify — as a limitation — that they did not gather or report transgender status in their study,” she suggested. “In fact, if their data collection didn’t distinguish sex at birth from current gender identity, it’s unclear if their ‘sex’ data reflects birth-assigned sex, anatomical sex, current gender identity, or a mix of those things depending on how the respondent and/or data collector understood the question.”

Poteat added, “It’s also futile to report that the sample included transgender people but not to provide disaggregated data for that population. I’ve seen this occur many times when MSMs and trans women participate in the same study. Even if the sample was small, the data can be provided with all of the caveats provided for data with small sample sizes. This may also move researchers toward recruiting significant numbers of trans people into their studies so that they can have sample sizes of trans people large enough to make statistically significant conclusions.”