Video Abstract

Video Abstract

Close modal
BACKGROUND AND OBJECTIVES:

Most transgender individuals assigned female at birth use chest binding (ie, wearing a tight garment to flatten chest tissue for the purpose of gender expression), often beginning in adolescence, to explore their gender identity. Although binding is often critical for mental health, negative physical side effects, ranging from chronic pain to rib fractures, are common. Time to first onset of symptoms is unknown.

METHODS:

A community-engaged, online, cross-sectional survey (“The Binding Health Project”) enrolled 1800 assigned female at birth or intersex individuals who had ever used chest binding. Lifetime prevalence of 27 pain, musculoskeletal, neurologic, gastrointestinal, generalized, respiratory, and skin or soft tissue symptoms related to binding was assessed. Nonparametric likelihood estimation methods were used to estimate survival curves.

RESULTS:

More than one-half (56%) of participants had begun binding by age 21, and 30% had begun by age 18. In 18 of 27 symptoms, the majority of people who go on to experience the event will do so within the first binding-year, but several skin-related and rare but serious outcomes (eg, rib fracture) took longer to occur. Pain presents rapidly but continues to rise in intensity over time, peaking at >5 years of binding.

CONCLUSIONS:

Although many symptoms emerge quickly, others can take years to develop. Individuals and their clinicians can use this information to make informed decisions on how to structure binding practices and top surgery timing while meeting goals related to gender expression and mental health. Access to puberty blockers may delay initiation of binding, preventing binding-related symptoms in youth.

What’s Known on This Subject:

Most transmasculine individuals use chest binding for gender expression at some point and report significant benefits for mental health, but ∼97% report experiencing negative physical symptoms. Timing to first onset of symptoms is not known.

What This Study Adds:

For most symptoms, the majority of people who will experience the outcome experience it within the first year of binding. Some skin-related and musculoskeletal symptoms take longer to emerge. Pain levels increase over the first 5 years of binding.

Chest binding is the practice of compressing chest tissue for gender expression to create a flatter appearance of the chest by using methods ranging from commercial binders to homemade products such as sports bras, elastic bandages, or compressive athletic wear. Chest binding is often adopted as a daily practice by transmasculine individuals (individuals assigned female sex at birth who identify on a spectrum of masculinity or as a nonbinary gender) as a means to decrease feelings of gender dysphoria related to their chest.1,2  Gender dysphoria (ie, psychological distress that a person feels because of a difference between their internal gender and their sex assigned at birth) can be debilitating for some, making binding an important intervention when surgery is either unwanted or inaccessible or in youth too young to qualify for the procedure. However, people who bind often report physical discomfort with binding.3  In a study of 1800 transmasculine adults, 97.2% of study participants reported some negative symptoms, with the most common being back pain (53.8%), overheating (53.5%), chest pain (48.8%), and shortness of breath (46.6%). Rarer symptoms included skin infections (5.3%) and rib fractures (2.8%).1  Eight of 10 individuals believed that it was important to discuss these symptoms with a health care provider, but only 15% had actually done so, likely because of lack of access to a provider with whom they felt safe discussing issues related to binding and gender.4 

Despite the nearly universal negative physical impact from binding, the benefits of binding often outweigh the disadvantages. These benefits include improved mood, reduced anxiety, reduced depression and suicidality related to gender dysphoria, and improved overall emotional wellbeing.1  Moreover, binding allows some to go out in public safely or maintain employment when not “out” as transgender. These benefits make binding essential for safety, mental health, and quality of life for many individuals, despite negative symptoms.3,5  Because stopping binding is often not an option in these instances, more research is needed to understand when negative symptoms emerge in response to binding, so transgender individuals can make informed choices and reduce risk to the extent possible. In previous research, it has been noted that 24 of 28 binding-related symptoms were significantly associated with length of cumulative exposure to binding in bivariate analyses.1  However, information on the timing and onset of symptoms remains unexplored. This information is important for patients and clinicians to understand the potential time line of symptoms and may also inform decisions about whether and when to use puberty blockers in youth as a means to delay onset of binding or pursue top surgery (double mastectomy), if desired, as a way to obviate the need for further binding.

The data are derived from a community-engaged, anonymous, online survey that was circulated on social media platforms and was distributed to a variety of regional, national, and international lesbian, gay, bisexual, transgender, or queer community organizations. The survey was also posted on web-based forums that serve transgender and gender-diverse communities. The final sample included 1800 individuals who were >18 years old and who had bound at the time when the survey was completed or in the past. There was no incentive for completion. Participants were young (58% were 18–24), primarily from the United States or Canada (82%), and most commonly used commercial binders (87%) as a binding method.1  More than one-half (56%) had begun binding by age 21, and 30% had begun by age 18.

To explore how symptoms emerged over time, a standardized measure of total cumulative exposure to binding was developed called “binding-years.” Binding-years were calculated by multiplying the average number of hours per day spent binding (intensity), the average days per week spent binding (frequency), and the number of years spent binding (duration) and by scaling this number to be equivalent to the number of years spent binding daily for 10 hours per day.1  This scaling was chosen because 7 days per week and 10 hours per day were the median and modal intensity and frequency of binding (ie, indicative of typical binding). A standardized measure was used rather than just unstandardized years spent binding to account for different intensities and frequencies of binding among participants, similar to the use of pack-years to measure total exposure to tobacco among smokers. In our previous work, standardized binding-years were associated with likelihood of experiencing 24 of 28 negative health outcomes, whereas length of time binding unstandardized for frequency or intensity was only associated with 13 of 28 outcomes, indicating the importance of using standardized years for predicting health outcomes.

Frequency, intensity, and duration of binding were assessed with the questions, “How many days on average do/did you bind a week?” “How many hours on average did you bind on those days?” and “How long have you been binding/bound in the past (ie, weeks, months, years)?” These measures were revealed to be reliable and valid compared to a gold standard retrospective assessment in which a trained interviewer conducted a one-hour interview eliciting a detailed life history calendar of binding. Correlation between standardized binding-years calculated from the life history calendar assessment and standardized binding-years calculated from a survey assessment administered 6 to 9 months later was 0.88 in a sample of 16 individuals with diverse binding histories (data not shown). As a sensitivity analysis to explore the impact of recall bias on the validity of this measure, we restricted the sample to participants who began binding <2 calendar years ago, a period in which recall is easier. Results were essentially unchanged (data not shown).

Participants were asked “Have you experienced any of the following health problems and attribute them to binding? Please select all that apply,” which was followed by a list of 27 symptoms (1 symptom of the original 28 was dropped because of issues estimating the survival curve). The data obtained represent current status data, a form of interval-censored survival data (a patient with a cumulative exposure of T years who experienced the symptom must have first experienced it sometime in the time range [0, T], whereas if they did not experience the symptom, then they are right-censored at T years of follow-up). The proportion of patients experiencing an event after different numbers of standardized binding-years and a bootstrap 95% confidence interval were estimated via nonparametric maximum likelihood estimation (NPMLE) via the R interval package. Because of data sparsity, the NPMLE was only plotted to a maximum of 10 standardized binding-years. For virtually all the 27 outcomes, there did not appear to be a substantial increase in prevalence after 10 standardized binding-years, making a 10-year cutoff reasonable. The NPMLE contains several intervals called Turnbull intervals that have an associated upper and lower bound for the survival function; any monotone survival function connecting the upper and lower bound has the same likelihood, so the NPMLE does not identify a single unique survival function estimate.6  Turnbull intervals are indicated via boxes when plotting the NPMLE.

To describe the speed at which each symptom manifested for the first time, the R interval package was used to compute estimates, and 95% bootstrap confidence intervals for the proportion of people who will have experienced the event by 10 binding-years who already experienced the event by 1 binding-year (ie, 1-year prevalence divided by 10-year prevalence), as well as the number of binding-years before the prevalence of the outcome reached 50% of its 10-year prevalence. We also estimate the conditional probability of experiencing the symptom by 10 binding-years given not experiencing the event by 1 binding-year.

We were interested in describing trajectories of the survival curves and their hazard functions (ie, the instantaneous risk that those who have not experienced the outcome will experience the outcome), with the idea that certain trajectories would be consistent with different hypothesized mechanisms for how a symptom develops. We theorized that survival curves could follow 1 of 3 trajectories. Curves with a high initial hazard function that continually decreases would be indicative of an individual risk mechanism, wherein individuals who are prone to an outcome because of their bodies or binding practice would quickly experience the outcome for the first time, and the risk of newly developing the outcome decreases over time. Curves with a constant hazard would be consistent with a random chance mechanism, wherein there is a constant hazard of experiencing the outcome at any point in one’s binding career. Curves with an increasing hazard would be consistent with a cumulative damage hypothesis, wherein risk of the outcome is low to start, but it becomes more and more likely over time, perhaps because of tissue damage over long-term exposure to binding.

To estimate the hazard function, Weibull curves were fitted to each of the survival curves. The Weibull distribution is a commonly used parametric survival model for modeling the hazard function; its shape parameter k cleanly indicates the shape of the hazard function (k <1 indicates decreasing hazard, k = 1 indicates constant hazard, and k >1 indicates increasing hazard).7  Diagnostic plots indicated that it provides a close fit to the survival curves in this study (data not shown). Maximum likelihood estimates and confidence intervals for the Weibull shape parameter for each of the 27 outcomes were estimated by using the R icenReg package.

Participants were asked to rate the severity of 5 types of pain they experienced on a scale of 1 to 10 with the question “Does binding cause you [pain symptom]? If so, please rank the pain on a scale of 1 (no pain) to 10 (worst pain)” for symptoms such as shoulder pain, back pain, rib or chest pain, breast pain, and neck pain. The sample was split into 10 quantiles on the basis of standardized binding-years of participants (roughly 180 participants per quantile). Standardized binding-years were skewed strongly right in this sample, resulting in more closely spaced quantiles among people with low exposure to binding and aggregation across a wider range of binding-years in quantiles with greater exposure to binding. Mean pain rating was averaged among participants in each quantile and plotted against average binding-years of participants in that quantile.

Estimated survival curves for each of 27 symptoms are revealed in Fig 1. In 18 of 27 symptom outcomes, 50% of 10-year prevalence was achieved in <1 binding-year (Table 1). In 9 of those 18 symptoms, 50% of ten-year prevalence was achieved within the first binding-month. The proportion of people who will have experienced the event by 10 binding-years who already experienced the event by 1 binding-year ranged from 20.5% (skin infections) to 97.6% (chest pain). The probability of experiencing a given symptom by 10 binding-years, given not experiencing the symptom by 1 binding-year, ranged from 2.6% (chest pain) to 44.0% (bad posture). For all symptoms, the estimated hazard function revealed decreases through time with a Weibull shape parameter k <1 (all P < .05, with the exception of weakness [P = .053]), which is most consistent with an individual risk hypothesis.

Pain intensity rises over time (Fig 2). Breast pain and shoulder pain were rated as most intense. The majority of quantiles reported an average breast and shoulder pain rating between 4 and 5, whereas the majority of the quantiles reported pain ratings of between 3 and 4 for back pain, ∼3 for neck pain, and between 2 and 3 for rib or chest pain.

These results allow individuals who bind to understand risk of first onset over time for 27 binding-related symptoms. Most symptoms (18 of 27) emerged quickly: The majority of people who will experience these symptoms experience them for the first time immediately (ie, within their first binding-month) in the case of 9 of these symptoms, or within their first binding-year in the case of the remaining 9 symptoms. Only 9 symptoms took longer than 1 binding-year to reach 50% of 10-year prevalence: rib fracture, rib or spine changes, muscle wasting, respiratory infection, breast changes, scarring, swelling, skin changes, and skin infections. These symptoms tended to be lower prevalence, but potentially more serious, and fell primarily in the categories of musculoskeletal symptoms and skin or soft tissue symptoms. Skin damage and changes to skin elasticity can compromise top surgery outcomes,8  so the longer time to onset for many skin and soft tissue symptoms is encouraging for individuals for whom binding is an interim measure before top surgery can be obtained. Three of these symptoms took ∼4 or more binding-years to reach >50% of 10-year prevalence: rib fractures, skin infections, and respiratory infections. Long-term binders should remember that some of these outcomes may present later in their binding practice. Although these data can only describe when individuals experienced a symptom for the first time and cannot speak to current symptoms, it seems likely that individuals who are later in their binding career may have a different and potentially more serious symptom profile than individuals who began binding more recently.

Understanding 10-year prevalence is a helpful addition to our previously reported results on overall prevalence in the sample.1  For instance, the prevalence of self-reported rib fracture in the overall sample was 2.8%, but the 10-year prevalence was 6.6%. Individuals who are binding for a longer duration may view 6.6% (1 in 15 long-term binders) as the more relevant prevalence figure on the basis of their own long-term binding practice. We also present conditional probabilities of experiencing the event long-term given no experience of the event in the first binding-year; patients and their clinicians may consult these numbers to understand risk of experiencing the symptom given successfully avoiding the symptom within the first binding-year.

Pain outcomes were some of the quickest to onset, and the intensity of pain experienced continued to increase over longer durations of binding. All pain outcomes peaked in intensity after ≥5 standardized binding-years. Breast pain and shoulder pain had higher average pain ratings than back pain, neck pain and rib or chest pain. Average levels of reported pain correspond to moderate chronic pain for breast and shoulder pain, as well as back pain in long-term binders, and mild chronic pain for neck and rib or chest pain.9  Individuals who have preexisting conditions (eg, shoulder injuries) that may exacerbate pain in these regions may want to take particular precautions when binding.

These data are most consistent with an individual risk model of symptoms from binding, wherein the hazard of the outcome decreases over longer binding periods. This means that the instantaneous risk of developing each outcome is highest in the earliest days of binding. As time elapses, those who have not yet developed the symptom become less and less likely to develop it for the first time, although it is still possible. This decreasing hazard pattern was true even for symptoms that took longer to reach their long-term prevalence (eg, rib fractures), although the hazard did not decrease as dramatically as with back pain or many other symptoms. It may be reassuring to individuals who bind that emergence of symptoms does not necessarily seem to be driven primarily by long-term cumulative damage to their bodies. However, this pattern of decreasing hazard could equally be explained by individuals initially using a binding method that is more prone to cause damage before switching to a binding method that is safer (perhaps because of symptom emergence). We also do not have data on whether long-term binding is associated with more severe or recurrent symptoms, only their first emergence. It is clear from these data that pain levels do continue to increase over time. More clinical research is needed on the physiologic impact of binding and mechanisms driving symptom emergence.

In previous work, researchers found that binding fewer days per week was associated with lower risk of experiencing most of these 28 symptoms.1  Taking off days each week from binding may therefore help reduce risk but is not always feasible for those who rely on binding for its therapeutic effects.3  Clinicians should continue assessing patients who bind for new symptoms over time and work with the patient to manage and treat their symptoms while supporting their goals for binding and/or top surgery.4  These new data allow health care providers to set expectations with patients with respect to the possible time line of negative side effects after helping patients calculate how their intensity and frequency of binding translates to the standardized binding-years measure (10 hours per day of daily binding) used in this analysis. Given how quickly symptoms emerge and their prevalence, these findings also support access to puberty blockers in interested youth to delay onset of binding and its corresponding symptoms. Puberty blockers may therefore be crucial not just for mental health of transmasculine youth,10  but also for their physical health.

This study was limited by its cross-sectional nature and lifetime (ever or never) assessment of self-reported symptoms. An ideal data set would measure symptom prevalence longitudinally to observe exact event times and minimize recall bias. Because of survival biases in our cross-sectional sample, our estimated 10-year prevalence is likely lower than the average prevalence that would be experienced if everyone who started binding continued binding for at least 10 binding-years (ie, people who reach a high number of binding-years are more likely to be the individuals who face relatively fewer symptoms and so find binding more tolerable long-term). We collected data on lifetime prevalence rather than recent prevalence, so we do not have incidence data about or period-specific prevalence of current symptoms at different total binding-years. These results do not reveal which symptoms are long-lasting versus transient and which symptoms occur once and which are recurring; furthermore, except for pain, we have no information on severity of symptoms or how much they impacted quality of life. In these graphs, we do not control for or stratify by relevant features such as the type of binder used or the individual’s chest size. The greatest limitation of this data set, however, may be the lack of positive outcome data. Future research revealing time to onset of reduced dysphoria and improved mental health as a result of binding would illustrate not just the risks of binding, but also its critical benefits for many transgender individuals.

We thank Alexandra Corbet for her contributions to the study design and data collection and Avani Samandur and Emily Sabia for contributing to the literature review. We also thank our participants, who freely shared their information and enabled this work.

Dr Peitzmeier conceptualized the research question and analysis and wrote the majority of the manuscript; Dr Silberholz helped shape the data analysis, conducted the data analysis, and wrote portions of the manuscript; Drs Gardner and Weinand and Ms Acevedo conceptualized and designed the study, designed the data collection instruments, conducted data collection, and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: No external funding.

NPMLE

nonparametric maximum likelihood estimation

1
Peitzmeier
S
,
Gardner
I
,
Weinand
J
,
Corbet
A
,
Acevedo
K
.
Health impact of chest binding among transgender adults: a community-engaged, cross-sectional study
.
Cult Health Sex
.
2017
;
19
(
1
):
64
75
2
Jones
T
,
de Bolger
ADP
,
Dune
T
,
Lykins
A
,
Hawkes
G
.
Female-to-Male (FtM) Transgender People’s Experiences in Australia: A National Study
.
Cham, Switzerland
:
Springer
;
2015
3
Lee
A
,
Simpson
P
,
Haire
B
.
The binding practices of transgender and gender-diverse adults in Sydney, Australia
.
Cult Health Sex
.
2019
;
21
(
9
):
969
984
4
Jarrett
BA
,
Corbet
AL
,
Gardner
IH
,
Weinand
JD
,
Peitzmeier
SM
.
Chest binding and care seeking among transmasculine adults: a cross-sectional study
.
Transgend Health
.
2018
;
3
(
1
):
170
178
5
Julian
JM
,
Salvetti
B
,
Held
JI
,
Murray
PM
,
Lara-Rojas
L
,
Olson-Kennedy
J
.
The impact of chest binding in transgender and gender diverse youth and young adults [published online ahead of print October 26, 2020]
.
J Adolesc Health
.
doi:10.1016/j.jadohealth.2020.09.029
6
Turnbull
BW
.
The empirical distribution function with arbitrarily grouped, censored and truncated data
.
J R Stat Soc B
.
1976
;
38
(
3
):
290
295
7
Machin
D
,
Cheung
YB
,
Parmar
MKB
.
Survival Analysis: A Practical Approach
, 2nd ed.
Chichester, United Kingdom
:
John Wiley & Sons
;
2006
8
Wilson
SC
,
Morrison
SD
,
Anzai
L
, et al
.
Masculinizing top surgery: a systematic review of techniques and outcomes
.
Ann Plast Surg
.
2018
;
80
(
6
):
679
683
9
Boonstra
AM
,
Schiphorst Preuper
HR
,
Balk
GA
,
Stewart
RE
.
Cut-off points for mild, moderate, and severe pain on the visual analogue scale for pain in patients with chronic musculoskeletal pain
.
Pain
.
2014
;
155
(
12
):
2545
2550
10
Turban
JL
,
King
D
,
Carswell
JM
,
Keuroghlian
AS
.
Pubertal suppression for transgender youth and risk of suicidal ideation
.
Pediatrics
.
2020
;
145
(
2
):
e20191725

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.