An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Adil Haider, MD, MPH, FACS, Eric Schneider, PhD, Jeremiah Schuur, MD, MHS, Lisa Kodadek, MD, Laura Vail, MS, Claire Snyder, MHS, PhD, Susan Peterson, MD, Danielle German, PhD, MPH, Anju Ranjit, MBBS, MPH, Rachel Adler, ScD, RD, Omar Harfouch, MD, MPH, Danielle Pelaez, and Brandyn Lau, MPH, CPH.
Author Information and AffiliationsLack of data on sexual orientation and gender identity (SO/GI) is a major challenge to understanding and addressing sexual and gender minority (SGM) health disparities. National organizations recommend routine collection of SO/GI information in electronic health records to help identify and address these health disparities. In the United States, there are nearly 130 million emergency department (ED) visits every year. Despite the importance of SO/GI for providing high-quality, patient-centered care and the opportunity to collect a high volume of SO/GI information for SGM disparities research, routine collection of SO/GI in the ED setting is rare, and optimal patient-centered approaches remain unclear.
The overall aim of this sequential mixed-methods study was to develop and evaluate the comparative effectiveness of patient-centered approaches for collecting SO/GI in the ED setting. The specific aims were to (1) gather qualitative input on perceived facilitators, barriers, and preferred approaches for collecting SO/GI in the ED setting; (2) develop and prioritize patient-centered approaches for collecting SO/GI; and (3) evaluate the comparative effectiveness of 2 patient-centered approaches, identified in aim 2, for collecting SO/GI in the ED setting.
The Emergency Department Query for Patient-Centered Approaches to Sexual Orientation and Gender Identity Study used a multiphase mixed-methods design. Phase 1 consisted of qualitative and quantitative data collection. This included in-depth interviews with a purposive sample of patients and ED providers from 1 major city in the Mid-Atlantic United States and a nationally representative phone survey of patients, ED physicians, and nurses, to identify facilitators, barriers, and preferred methods to collect SO/GI in the ED. Phase 2 used data from phase 1 in modified Delphi rounds that resulted in stakeholder advisory board (SAB) members identifying 2 methods of SO/GI collection to implement in a trial. In phase 3, we sequentially implemented and evaluated 2 methods of SO/GI collection—verbal collection by nurses (mode 1) and nonverbal self-report during registration (mode 2)—in 4 hospital EDs in the Northeast and Mid-Atlantic United States. We assessed ED patient satisfaction and comfort using average modified Communication Climate Assessment Toolkit Questionnaire (CCAT) scores as advised by our SAB. We invited all eligible patients to complete outcome surveys. We calculated primary and secondary outcome results using analysis of variance (ANOVA) or chi-square tests between modes for each patient match group. We used multivariable ordered logistic regression to examine whether intervention mode was associated with modified CCAT scores after controlling for potential confounding variables: age, race, illness severity, and hospital site.
In phase 1, a total of 53 patients and 38 physicians, nurses, and registrars participated in qualitative interviews. Interviews revealed that, although patients believed SO/GI was important in all clinical circumstances, clinicians thought SO/GI was rarely important. The online survey confirmed these findings on the national level. In total, 1617 patients (244 lesbian, 289 gay, 179 bisexual, 101 transgender, and 804 heterosexual and cisgender) and 429 providers (209 physicians, 220 nurses) completed the national survey. Among patients, only 10.1% refused to provide SO, whereas 77.8% of clinicians believed patients would refuse to provide SO. Similarly, only 7.4% of patients refused to provide GI, whereas 73.4% of clinicians believed patients would refuse to provide GI.
In phase 2, SAB members used findings from phase 1 to identify 2 methods to collect SO/GI information and test in an interventional study: verbal collection by nurses during patient assessment (mode 1) vs nonverbal self-report during patient registration (mode 2). SAB members highlighted the importance of staff education and training, provided important direction for trial implementation, and approved the outcome measures used in phase 3 below.
In phase 3, CCAT scores (on a scale of 0-100) during mode 1 were 89.5 (SD, 20.5; 95% CI, 85.7-93.3), 91.8 (SD, 18.9; 95% CI, 88.3-95.3), and 92.7 (SD, 15.9; 95% CI, 89.8-95.7) for SGM, non-SGM, and Blank Field patients (those who did not list SO/GI information), respectively. During mode 2, average modified CCAT scores were 95.6 (SD, 11.9; 95% CI, 92.7-98.5; P < .05), 93.2 (SD, 13.6; 95% CI, 89.9-96.5; P = .591), and 93.6 (SD, 14.7; 95% CI, 90.0-97.2; P = .703) among SGM, non-SGM, and Blank Field patients. Modified CCAT scores were significantly higher among SGM patients only. The SAB was involved in analyzing these results and provided guidance on their interpretation and communication.
The vast majority of patients are willing to provide SO/GI in the ED. Patients report higher satisfaction and comfort with ED encounters when SO/GI is collected along with other demographic information via a nonverbal, written method during patient registration. Our results highlight the importance of patient-centered approaches to training, technical assistance, and implementation for effectively integrating SO/GI information collection in ED settings.
Sexual and gender minority (SGM) populations face significant disparities in access to health care and outcomes. Recent estimates indicate that 3.5% of Americans identify as sexual minorities1 and 0.6% of Americans identify as transgender.2 SGM is an umbrella term; each group is distinct with specific health care needs.3 Many SGM individuals are part of other communities facing disparities related to race, ethnicity, and socioeconomic status, and may be vulnerable to the cumulative negative health impact of these factors.4
SGM populations experience significant health disparities that are inextricably linked to social stigma and discrimination, including higher risk of all-cause mortality, cardiovascular and immune conditions, depression, suicide, substance abuse, and HIV.5-10 Gay men, particularly within communities of color, have a higher risk of HIV and other STDs.11 Lesbians and bisexual women are more likely to be overweight or obese,12 and lesbians are less likely to receive preventive services for cancer.13 Transgender individuals arguably experience the most acute health care access and outcomes disparities. Current research indicates that transgender individuals are at greater risk for physical victimization, substance abuse, and attempted suicide.14,15 According to one systematic review, transgender people experience a high prevalence of HIV and other sexually transmitted diseases.16 The overwhelming array of health disparities experienced by the SGM community, compounded by the lack of available data, recently led the US Department of Health and Human Services to identify SGM individuals as a target group for improvement in Healthy People 2020 and prioritize the need to identify appropriate data collection systems for SGM health.17
Stigma and discrimination are fundamental causes of SGM health disparities.18-20 Unfortunately, social stigma and discrimination against SGM individuals extend into the health care system. Many SGM patients report that negative past experiences with their health care providers or fear of provider homophobia lead them to delay seeking care even when they are ill or injured.21,22 In 2009, the report When Health Care Isn't Caring: Lambda Legal's Survey on Discrimination Against SGM People and People Living With HIV published findings from a survey of 4916 people representing SGM communities and people living with HIV that indicated disturbingly high levels of discrimination and substandard health care.23 More than half of all respondents reported experiencing at least 1 of the following instances of discrimination in care: being refused needed health care; having a health care provider who refused to touch them or used excessive precautions; experiencing harsh or abusive language from the health care provider; or experiencing physically rough or abusive treatment from the health care provider. The report also found that 51.9% of transgender and 9.1% of lesbian, gay, and bisexual (LGB) respondents feared they would be refused medical service. Nearly 75% of transgender and 28.5% of LGB respondents were concerned that health care providers would treat them differently than non-SGM people.23 Another study of 427 SGM physicians found that 65% had heard derogatory comments about SGM individuals and 34% had witnessed discriminatory care of an SGM patient.24 A review of literature on nurses’ attitudes toward SGM patients showed evidence of negative attitudes in all 17 included articles.25 The high prevalence of stigma and discrimination in the health care setting highlights the importance of promoting high-quality, patient-centered care for SGM patients and addressing SGM health care disparities.
Lack of data on sexual orientation and gender identity (SO/GI) is a major challenge to understanding and addressing SGM health disparities.26 Furthermore, health care providers need to know a patient's SO/GI information as it may be clinically relevant.27 For example, a female-to-male transgender patient who presents to the emergency department (ED) with pelvic pain and does not disclose his GI—and, most importantly, is not asked—is in danger of receiving poorer-quality or delayed health care simply due to his providers’ lack of awareness.
EDs are the source of nearly half of inpatient admissions in the United States and the primary point of entry for uninsured and underinsured patients.28 In the United States, there are nearly 130 million ED visits every year.29 Few hospital EDs routinely collect SO or GI information27; however, the potential impact of routinely collecting SO/GI information in the ED is tremendous given the high volume of patients.30,31 In the Human Rights Campaign's study of health care facilities that use the Healthcare Equality
Index online survey to measure themselves against the established criteria for SGM patient-centered care, only 26% of facilities recorded SO in print or in the electronic health record (EHR).27 Health care institutions may not routinely ask for SO/GI information, to avoid potential discomfort among providers and patients.27 Although there are significant challenges to asking SO/GI information in a patient-centered, inclusive, and sensitive way, the failure to inclusively and sensitively address SO/GI information in the hospital emergency setting effectively creates conditions for a kind of invisibility among SGM individuals—both within the examining room and within health outcomes data. In essence, much of the American health care system is operating under a “Don't Ask, Don't Tell” policy,27 with many health care providers assuming that patients are heterosexual.32 Without a safe and “normalized” way to disclose SO/GI information, many patients are put in the uncomfortable position of either voluntarily disclosing their SO/GI information while wondering if the lack of routine collection is a sign of institutional stigma, or alternatively, remaining silent.
Efforts to routinely collect patient SO/GI information have been concentrated in the primary health care setting, particularly within medical centers that are known to be SGM-friendly. Since 2016, the Health Resources and Services Administration (HRSA) Bureau of Primary Care has required federally qualified health centers to report on patients’ SO/GI.33 Efforts at University of California, Davis (UC Davis), one of the first US academic health centers to formally introduce patient SO/GI demographic data into its EHR, provide a framework for overcoming initial resistance and engaging patients and providers in health systems change.34 While many primary health care patients may elect to go to SGM-friendly facilities or providers whom they trust and confide in, ED patients typically have little choice over which hospital or provider treats them. In the absence of choice of the hospital or provider giving care, patients still deserve to receive high-quality, patient-centered, and inclusive care. There are currently no recommendations or methods standards to collect SO/GI data in EDs. Despite the importance of SO/GI collection for providing high-quality, patient-centered care and the opportunity to collect a high volume of SO/GI information for SGM disparities research, routine collection of SO/GI in the ED setting is rare and the optimal patient-centered approaches for collecting this information remain unclear.
Given the importance of SO to population health and clinical care, the objectives of this study were to (1) gather qualitative input on perceived facilitators, barriers, and preferred approaches to collecting SO/GI in the ED setting; (2) develop and prioritize patient-centered approaches for collecting SO/GI; and (3) evaluate the comparative effectiveness of patient-centered approaches for collecting SO/GI in the ED setting.
Before submitting the funding proposal to PCORI, our team conducted preliminary work to gather input from patients and health care providers on perceived barriers, facilitators, and preferred approaches to collecting SO/GI information. We distributed a standardized survey to 47 SGM students, staff, and faculty of the Johns Hopkins Medical Institutions, and 22 members of the Johns Hopkins Patient- and Family-centered Care Council, which comprises current or former patients and their caregivers from all medical departments as well as faculty and staff at Johns Hopkins Hospital. This preliminary work revealed that most patients and clinicians agree it is important for health care providers to have information on patients’ SO and GI. Additionally, patients reported they would feel most comfortable providing SO/GI information via a written or electronic form or answering SO/GI questions verbally to a physician or nurse.
To develop the stakeholder advisory board (SAB), we first conducted a comprehensive search of major SGM health and advocacy organizations, including Fenway Health, the Gay and Lesbian Medical Association, and Services and Advocacy for GLBT Elders. Consulting with representatives from these organizations who joined the SAB, we continued to identify board members who had a wide range of expertise. Our efforts resulted in a 25-member SAB comprising patients, patient advocates, clinicians, health care administrators, and directors of some of the most influential national and local SGM advocacy organizations. Patients composed 28% of the SAB. To learn more about best practices for SO/GI information collection, we included a representative from UC Davis, one of the first academic medical centers in the country to implement SO/GI collection throughout its system. The SAB was designed to incorporate views from patient, provider, hospital, and policy perspectives.
SAB members provided extensive feedback and guidance during the development of the grant application, and the core investigative team continued to work closely with the SAB to monitor and evaluate study conduct. SAB meetings occurred quarterly, during which study plans, measures, instruments, and results were presented to gain feedback from SAB members.
In phase 1, SAB meetings were held in person and via teleconference. During phase 1, SAB members provided feedback on the qualitative interview guides and the national survey drafts; we used that feedback to revise the guides and surveys. SAB members also provided insights into the qualitative and quantitative interview results and commented on the impact of the results within the wider political and health care climate in the United States. Insights from the SAB were integral to understanding the balance among identifying a “best” method, patient-centered methods, and methods that are feasible in the ED.
During phase 2, SAB members participated in modified Delphi rounds, as described in “Phase 2” in the Methods section, to identify the preferred methods of SO/GI collection to test. Contact between participants is discouraged during Delphi rounds, so we suspended SAB meetings during this time. Instead, SAB members completed online surveys that required members to prioritize phase 3 study methods and provide open-ended feedback. Two Delphi rounds were completed, and SAB member consensus was reached at the end of the second round. The Delphi rounds resulted in the identification of the phase 3 comparators: SO/GI collection via nurse verbal collection vs SO/GI collection via registrar written collection.
In phase 3, SAB members resumed quarterly in-person and teleconference meetings, providing feedback on ED staff training curricula and participating in staff education efforts. SAB members also gave insight into preliminary phase 3 results and suggested actionable steps for the study team to tackle implementation challenges. Finally, SAB members helped develop dissemination and implementation plans.
In addition to the SAB, we included a patient coinvestigator on our study team. As a coinvestigator, they represented the perspective of an LGBT (lesbian, gay, bisexual, and transgender) patient who had received ED care and they helped to recruit LGBT patients for in-depth qualitative interviews. The core investigative team also included experts in disparities and outcomes research, qualitative research methodology, epidemiology, and emergency medicine. Further details on study team and SAB members can be found in Appendix 1.
The purpose of the Emergency Department Query for Patient-centered Approaches to Sexual Orientation and Gender Identity (EQUALITY) study was to identify the optimal patient-centered method to collect SO and GI in the ED. We chose a mixed-methods design because of the rich body of data that mixed-methods studies produce, allowing us to identify and understand patient-centered methods for SO/GI collection. An overview of our study design is provided below, followed by an in-depth explanation of each phase.
In phase 1, we gathered qualitative input on perceived facilitators, barriers, and preferred approaches for collecting SO/GI in the ED setting. To do this, we explored themes about SO/GI collection among SGM and heterosexual, cisgender (identifying with the same gender as assigned at birth) patients and providers using semistructured, in-depth interviews. We then used the findings from these interviews to develop and disseminate a nationwide survey for SGM and heterosexual, cisgender patients and providers.
In this phase, we developed and prioritized patient-centered approaches for collecting SO/GI information. SAB members participated in a consensus-building protocol to identify patient-centered methods of SO/GI collection.
Phase 3 aimed to evaluate the comparative effectiveness of patient-centered approaches for collecting SO/GI in the ED setting. The SO/GI collection methods identified in phase 2 were implemented in the participating EDs. A diagram of the EQUALITY study design can be found in Figure 1.
EQUALITY Study Design: Multiphase Mixed-Methods Study.
Phase 1 of the EQUALITY study used an exploratory sequential mixed-methods design. In the first qualitative portion, we conducted in-depth interviews with a sample of patients and ED providers to identify themes regarding perceived facilitators, barriers, and preferred approaches for collecting SO/GI in the ED setting. The results from the qualitative portion informed the development of the second, quantitative portion of phase 1, in which we conducted a national survey of patients and ED providers. SAB members also decided to exclude the original transgender sample from the national surveys, because >90% of participants from the initial national survey sample no longer qualified as transgender after cross-validation with a 2-question prompt of sex assigned at birth and current GI. We conducted a second national survey to obtain data from an accurate transgender sample.
We recruited a purposive sample of participants for semistructured, in-depth, one-on-one interviews. We used community outreach, flyers, and social media in Baltimore to recruit patient participants with experience seeking emergency medical care. Individuals aged 18 and older who were willing and able to provide informed consent were eligible for study inclusion. Patients contacted the researchers, who completed a telephone screen; screening criteria included questions about age, race, ethnicity, SO, GI, and sex assigned at birth. Interview recruitment was designed to ensure approximately equal participation of participants who identified as lesbian, gay, bisexual, and heterosexual, with variation in age and race/ethnicity within each stratum.
We recruited ED provider participants, including physicians, nurses, and advanced practice providers, from 3 community and 2 academic medical centers. Provider participants were eligible for inclusion if they were aged 18 or older, currently employed at a Johns Hopkins Hospital campus, and willing and able to provide written informed consent. We designed interview recruitment to ensure that a variety of staff roles within the ED setting, as well as diversity across age and race/ethnicity, were included.
Qualitative interviews took place between July and September 2014. One of 2 researchers conducted all interviews in person in a private location. The researchers informed all participants that the purpose of the research was to understand the risks, benefits, and preferred approaches to collecting SO/GI information in the ED. Informed consent was obtained before beginning the interviews, which were tape recorded. Interviews lasted between 30 minutes and 2 hours, with most interviews lasting approximately 1 hour. Interviewers wrote summaries after each interview. Audio data were deidentified and transcribed verbatim before analysis. Interview guides can be found in Appendix 2.
We first used open coding to develop a codebook. Two researchers independently coded the data using the constant comparative method, a commonly used method for qualitative analysis that is frequently used as part of grounded theory, which can be especially helpful for analyses with minimal a priori theory.35 The researchers used an iterative coding process to identify themes, and they examined those themes for similarities and differences across participants, with attention to sample diversity. The coding schemes for patients and providers emerged from the data, so they differed slightly but did have similar themes and structure. Transcribed interviews were entered into ATLAS.ti (https://atlasti.com/) for data management.36 We ensured consistency by double coding every fifth interview, and coders met regularly with senior staff. Having 2 analysts and a planned communication structure richened the process and created a mechanism for dialogue, with the aim of identifying areas of inconsistency or external influences. Data were summarized within and across themes and shared with the SAB for feedback and further insight.
Researchers used Gesellschaft für Konsumforschung (GfK), an international market research firm, to conduct national surveys of patients and providers. Researchers recruited patient survey participants through GfK's KnowledgePanel, and they recruited provider survey participants through GfK's Physician Consulting Network. The GfK Knowledge Panel is the largest probability-based survey panel in the United States and has been used for research on numerous public health–related topics.37-39 GfK recruits members for its KnowledgePanel using random digit dialing and address-based sampling techniques to ensure inclusion of households with landlines, cell phone–only households, and households both with and without internet access.40 Participants were eligible if they were aged 18 or older, lived in the United States, and consented to participate in the survey. Surveys took place between March and April 2015.
Based on qualitative data collected during the in-depth interviews, we developed 2 new comprehensive, English-language surveys for this study in collaboration with our SAB. Because the EQUALITY study is the first interventional trial to investigate SO/GI information collection in the ED, a novel setting, our SAB advised us to use new measures. We designed survey measures to solicit information on the importance of SO/GI collection; barriers, facilitators, risks, and benefits to SO/GI collection; willingness to disclose and collect SO/GI information; and preferred methods to disclose and collect SO/GI. Survey items were multiple choice, Likert scale, or open ended. All survey items were pilot tested with students and faculty from the Johns Hopkins Bloomberg School of Public Health before beginning quantitative data collection and have not yet undergone psychometric testing. Surveys can be found in Appendix 3.
We calculated descriptive statistics of survey items. We used chi-square tests to determine differences in views of routine collection of SO between straight and lesbian, gay, and bisexual patients. We calculated the odds of refusing to provide SO using stepwise logistic regression; the primary independent variable was SO and covariates included sex, SO, age, educational level, race, marital status, rental status, head of household, work status, and income. We conducted all analyses using Stata Version 13 (Stata Statistical Software: Release 2013).41 Phase 1 results are detailed in “Phase 1: Quantitative Results” in the Results section.
Phase 2 of the EQUALITY study consisted of modified Delphi rounds to determine 2 optimal patient-centered approaches for SO/GI information collection in the ED. The Delphi process is a structured, consensus-building technique in which a panel of experts answers questionnaires in multiple rounds. Developed by the RAND Corporation during the Cold War to identify and prioritize strategic targets for launching nuclear weapons, the Delphi method is now widely used in health care research to set priorities and gain concensus.42,43 We used a modified Delphi assessment to prioritize patient-centered approaches for collecting SO/GI information to be tested and to finalize how they will be evaluated.
Members of the SAB participated in the modified Delphi rounds. Processes were standardized across SAB members regardless of profession.
Each questionnaire used multiple-choice questions with additional free response options (Appendix 4). We developed questionnaire language to reflect findings from phase 1 qualitative and quantitative interviews and surveys. Requested input included participants’ ratings about who should collect SO/GI information in the ED, what question phrasings were most and least preferred overall, and proportions of participants who were comfortable with SO/GI information being included in their medical record. We defined consensus as agreement by at least 75% of the SAB.
We used results from previous phases of the study to develop a list of the preferred methods and modes for collecting SO/GI information. We developed a list of options for the potential methods of inclusion of SO/GI information in the medical record. SAB members reviewed these lists and added options that they considered to be high priority. Stakeholder responses were compiled and sent for prioritization in subsequent rounds.
We sent SAB members the compiled list of identified roles of questioners, phrasing of questions, and modes of collection developed in round 1. Each stakeholder selected the highest-priority role of the person collecting information and the highest-priority mode for collection. Each stakeholder also selected the 2 highest-priority phrasing of questions. If consensus was not achieved in the first round of prioritization, we sent the compiled responses back to the stakeholders to review and reprioritize, to build consensus. We repeated this process until at least 2 high-priority roles, 2 high-priority modes, and 2 to 3 high-priority questions were identified.
Phase 2 results are detailed in the “Phase 2: Modified Delphi Rounds Results” section.
After the modified Delphi rounds were completed, we identified and compared 2 questioners and 2 modes of collection via a multisite trial. Four hospital EDs in Boston, Massachusetts, and Baltimore, Maryland (2 academic hospitals and 2 community hospitals), participated in phase 3 of the study. As per the SAB's recommendations, ED staff at each location received training before mode 1 and mode 2 implementation of SO/GI information collection. Study staff conducted standardized, in-person or remote trainings for 185 ED nurses, registrars, residents, physician assistants, and physicians that included a short discussion on SGM health disparities, how SO/GI information collection can alleviate these disparities, and SGM terminology. Staff were given informational handouts to take with them. Given the demographic differences of Baltimore vs Boston, the SAB recommended implementing both written and verbal modes at all sites. Mode 1 and mode 2 were implemented sequentially at each site to focus on training and technical assistance, with the staff implementing SO/GI collection during that particular mode: Mode 1 was implemented for 6 to 8 months, depending on how long it took to reach the sample size, then mode 2 was implemented until the sample size was reached. Mode 1 consisted of the verbal collection of SO/GI information by nurses during the patient's clinical encounter. Mode 2 consisted of written collection of SO/GI information by registrars during the registration process. We chose these interventions through SAB member input, weighing both the patient-centeredness and the feasibility in the ED.
All English-speaking adult patients who entered the EDs during the trial period were eligible for inclusion. We excluded patients if they were diagnosed with a psychiatric condition, under the influence of drugs or alcohol, had an Emergency Severity Index (ESI)44 less than 2 (minimum score of 1, maximum score of 5, with lower numbers indicating urgent and high-risk cases), or had a hand or arm injury (mode 2 only) that would limit their ability to use an iPad.
Because the EQUALITY study focuses on SGM patient experiences, research assistants attempted to invite every patient who identified as SGM and met inclusion criteria to participate in follow-up surveys. Every patient who identified as an SGM was invited to complete outcome surveys. SGM patients who consented and were enrolled were matched 1:1 by age (±5 years) and illness severity (ESI44 ±1) to patients who identified as heterosexual and cisgender (non-SGM) and to patients for whom SO/GI information was missing (Blank Field). Matched non-SGM and Blank Field patients were also invited to complete outcome surveys. We chose not to match patients on wait time, because of difficulty finding matches, and we excluded race/ethnicity and socioeconomic status from match criteria because this information was not routinely collected at all sites.
Research assistants collected 2 types of data from patients during phase 3: patient self-reported SO/GI and patient-reported comfort during the ED visit. ED staff asynchronously collected SO and GI information according to mode (ie, nurse verbal collection in mode 1 followed sequentially after 6 to 8 months by registrar written collection in mode 2). Although slight variations in question wording occurred because of differences in EHRs, patients were asked their SO (straight/heterosexual, gay/lesbian/homosexual, bisexual, queer, questioning/unsure, pansexual, prefer to speak with nurse, decline to state, or other) and GI (female, male, transgender female-to-male, transgender male-to-female, queer/genderqueer, questioning/unsure, prefer to speak with nurse, decline to state, or other). Although study staff highly encouraged nurses and registrars to collect SO/GI information, we could not mandate it without approval from the nurses’ union, which was not feasible in the research study's time frame. A proportion of patients were not asked about SO/GI, as evidenced by blank fields in the EHR; these patients were categorized as Blank Field. Similarly, patients were not required to provide SO/GI information when asked by nurses or registrars, and those few patients (n = 20) who declined were subsequently not invited to complete a follow-up survey.
To collect patient-reported comfort during the ED visit, research assistants (RAs) invited selected patients to complete a follow-up survey after the patient had been in the ED for at least an hour, regardless of whether they had previously been asked for SO/GI information. RAs ran EHR reports every 15 minutes (when on duty between 7 am and 10 pm) to allow them to determine which patients reported their SO/GI as SGM, which identified as non-SGM, and which were not asked. When an RA identified an eligible patient (SGM, non-SGM, or Blank Field), this triggered an outcome survey. If the patient had been in the ED for at least an hour, the RA approached the patient and invited him or her to complete the follow-up survey. This survey included general questions about patient experiences from the Communication Climate Assessment Toolkit (CCAT)45 as well as Likert scale questions specific to the experience of providing SO/GI during the ED visit. We did not collect SO/GI information from any patients during the follow-up survey, including those who were not asked for this information by the nurse or registrar. Patient surveys can be found in Appendix 5. Because most of the RAs also collected data for several different studies and/or did not staff the ED full time, it was not feasible for them to wait until a patient received discharge orders to complete an outcome survey.
This study's primary outcome was patient satisfaction as measured by a scale modified from the CCAT patient survey, an assessment of attitudes toward organizational climate and provider/patient communication.45 The CCAT is reliable, has been validated in geographically and ethnically diverse health care organizations, and accurately predicts patient-reported quality and trust.46 Containing 5 out of 7 items from the full CCAT, our prespecified modified scale included only those questions that were applicable to the ED population. For example, we kept the question “Do you feel welcome at the hospital?” but eliminated the question “Was it easy to reach someone on the phone if you had a question?” from analyses. Each scale item was scored as a 0 (most disagreement), 1/2 (neutral), or 1 (agreement), resulting in a scale score ranging from 0 to 5; higher scores were considered more favorable. If a patient skipped or answered “unsure” to an item, we excluded the patient's survey data from the analysis. We calculated the overall score by averaging the modified scale score and multiplying by 20, resulting in an overall score out of 100 points. We distinguished between score groups by a difference of 10 points, which corresponds to meaningful changes in patients’ beliefs that they were receiving high-quality care.47
Secondary outcomes to assess patient satisfaction included overall patient comfort, patient experiences, and patient comfort with SO/GI collection. Secondary outcome survey items were collected on Likert scales, and we grouped responses into 2 or 3 categories for analysis, depending on the nature of the question. For example, we categorized “How comfortable did ED staff seem to be when interacting with you today?” into 3 categories: very uncomfortable/uncomfortable, neither comfortable nor uncomfortable, and comfortable/very comfortable. Whereas we categorized “How concerned were you about your privacy while answering questions today?” into 2 categories: not at all concerned vs a little concerned/somewhat concerned/concerned/very concerned. To understand the overall acceptability of each method of SO/GI collection within staff workflow, we assessed the proportion of patients from whom SO/GI was collected with each collection method.
We calculated descriptive data on patient participants by intervention mode and patient match group (SGM, non-SGM, and Blank Field patients) using analysis of variance (ANOVA) for continuous variables and chi-square tests for categorical variables. Although patients were not required to provide SO/GI information, less than 5% declined to respond if they were asked. Because of their low numbers, this subset of patients was not enrolled.
We calculated primary and secondary outcome results using ANOVA or chi-square tests between modes for each patient match group. Because the modified CCAT is an ordinal categorical variable based on a summation of subset scores, we used multivariable ordered logistic regression to examine whether intervention mode was associated with modified CCAT scores after controlling for potential confounding variables: age, race, illness severity, and hospital site. We distinguished between score groups by a difference of 10 points, which corresponds to meaningful changes in patients’ beliefs that they were receiving high-quality care.47 Therefore, the outcome categories were multiples of 10 (0, 10, 20, 30, etc) ranging from 0 to 100. Although patients completing outcome surveys were matched on age, we controlled for age as a continuous variable in the regression model to further adjust for any intragroup variation. We also controlled for race using 3 categories: White/Caucasian, Black/African American, and other.
We considered several other possible options for examining variability in CCAT scores, including linear regression or CCAT score dichotomization–based logistic regression models. We based our final decision to rely on ordered logistic regression on the following aspects of the data: (1) The CCAT score is not normally distributed and is subject to a strong ceiling effect, with a substantial number of individuals answering at the top of the scale; (2) the scale is categorical, with answers possible only within 1 of 10 possible bins, which are ordered in the manner of a Likert scale for each contributing factor; (3) individual-specific sums of multiple domain scores also follow a pattern that we felt would be best addressed by continuing to consider the CCAT as an ordered categorical variable; and (4) multivariable linear regression was likely to introduce bias as a function of the non-normal distribution of individual item and individual-level combined mean CCAT scores. A priori, we decided to control for patient race/ethnicity, patient age, patient presentation severity, and study location (as a fixed effect). Other analyses are possible within and between specific subsets of patients; however, power may be insufficient to support the identification of significant findings in smaller subsets without the risk of type 1 or type 2 error.
We used complete case analyses for our final results. Even though we had a very low proportion of missing data, we conducted a series of sensitivity analyses on all eligible patients in which we sequentially did not include one of the potential confounders we had controlled for (ie, age, race, illness severity, and hospital site). No changes to our results were noted. For this study, matching was performed without replacement.48 By matching on factors such as age and medical severity, we purposely limited our ability to generalize findings in the non-SGM groups in favor of focusing on the SGM group for which the study questions were most relevant. We considered methods for replacing missing data, including multiple imputation, for the subset of patients who were not matched because they had missing values for CCAT responses, for example. Based on the predictive abilities of the independent variables available, we decided that imputing the outcome variables (eg, CCAT question responses) was as likely to introduce bias as to reduce it.49
We performed all statistical analyses using SAS software (SAS Institute; 2019) and Stata 14.2 (Stata Statistical Software: Release 14; 2015).
Phase 3 results are detailed in “Patient Outcomes” in the Results section.
The final study protocol as implemented can be found in Appendix 6. Relevant changes to the protocol over the course of the study included additions to staff and patient outcome surveys and changes to enrollment dates to facilitate attainment of target numbers. This study was approved by the Johns Hopkins Medicine IRB and the Partners Health Institutional Review Board.
We asked participants, both patients and health care team members, to discuss the risks and benefits of several SO/GI collection mechanisms. In general, SO and GI may be collected either on a form or verbally. Figure 2 identifies the points in the ED encounter during which SO/GI data could be collected. If these data are collected on a form, possibilities include paper, electronic tablet, electronic kiosk, smartphone application, or home computer application (eg, MyChart). If collected verbally, different health care team members may collect this information (registrar, nurse, or physician).
Potential SO and GI Collection Points in the ED.
Figure 3 describes interview participant enrollment, and Figure 4 depicts national survey participant flow. Patient and provider demographics can be found in Table 1. Six major themes emerged from the interviews: privacy, medical relevance, normalization, recognition, the patient–provider relationship, and ED flow. A description of themes can be found in Tables 2 and 3, and intersections between themes and collection mechanisms can be found in Table 4.
Phase 1: Interview Participant Flow.
Phase 1: National Survey Participant Flow.
Phase 1: Demographics of Patient and Provider Interview Participants.
Phase 1: Qualitative Interview Themes.
Phase 1: Thematic Intersections With Collection Mechanisms.
Phase 1: Patient and Provider Quotes Describing Interview Themes Regarding Sexual Orientation and Gender Identity Collection Methods.
Throughout the interviews, participants made a clear distinction between self-disclosure on a form and answering a verbal question from a registrar or provider. Patients viewed form collection as routine or standard policy and, in the setting of form collection, SO and GI were viewed as basic demographic data rather than a clinically relevant piece of information. Participants noted that form collection of SO/GI would be more private than answering questions out loud in a public setting. Form collection on paper, however, was also noted to be not entirely private. Having these questions on intake forms also signaled to participants that the hospital or clinic recognizes and respects SGMs. Finally, many participants felt that the impersonal nature of answering written questions would remove the fear of needing to “come out” to a provider with an unknown level of acceptance and knowledge about SO and GI minority health. Many providers, however, felt that they would be better suited to ask these questions verbally because of their ability to build rapport with patients. Some providers felt that sensitive questions are better addressed verbally as part of a discussion between the patient and provider. Representative quotes from patients and providers regarding form collection can be found in Table 4.
The findings related to verbal collection by the registrar were the most variable. Many participants (both patients and providers) did not feel registrars were appropriate for collecting SO/GI because of privacy concerns, yet registrar interactions emerged as a strongly viable site of collection. Many patient participants envisioned the registrar as a data collector and felt that allowing the registrar to collect SO and GI would normalize the collection of these data as standard demographics. Representative quotes related to registrar collection of SO/GI are found in Table 4.
Patients viewed verbal collection by the nurse as generally favorable. Both patients and members of the health care team felt that the nurse would be able to collect SO and GI in a private area and could manage this information in a sensitive and respectful way. One patient commented, “I was going to say I immediately go to: I want the nurse asking that.” Another agreed, “A nurse is a professional, so that would be fine with a nurse asking me.” Further representative quotes describing nurse verbal collection of SO/GI can be found in Table 4.
Patients most often equated verbal collection of SO and GI by the physician with medical relevance and sexual health–related concerns. Similarly, physicians felt these data were relevant in particular circumstances based on the patient's health concern. Physicians indicated they would be willing to ask if they themselves perceived the data to be medically relevant. Representative quotes describing physician verbal collection of SO/GI can be found in Table 4.
The survey response rates were 70.1%, 89.4%, and 86.2% for LGB, transgender, and provider participants, respectively. Nationally weighted demographic data for patients and providers are found in Table 5. Because of incomplete data, 141 surveyed providers were excluded; the survey organization was unable to provide crucial demographic information, which raised questions about the validity of these data, leading to their exclusion from the study sample. Demographics for transgender patients and providers are found in Tables 6 and 7, respectively. Results from the national survey confirmed the generalizability of our qualitative findings from a subset of patients and providers to the national level.
Phase 1: Demographics of Lesbian, Gay, Bisexual, and Straight Cisgender National Survey Participants.
Phase 1. Demographics of Transgender National Survey Participants.
Phase 1: Demographics of Provider National Survey Participants.
Data on willingness to provide SO and GI when routinely collected in the ED are presented in Tables 8 and 9. Among patients, only 10.1% would refuse to provide SO if asked, whereas 77.8% of clinicians believed patients would refuse to provide SO. Similarly, 7.4% of patients would refuse to provide GI if asked, whereas 73.4% of clinicians believed patients would refuse to provide GI.
Phase 1: Patient and Provider Views on Willingness to Provide SO.
Phase 1: Patient and Provider Views of Willingness to Provide GI.
In unadjusted logistic regression, bisexual patients had 1.73 times the odds of refusing to provide SO compared with straight patients (95% CI, 1.01-2.98; P = .048) and patients ≥60 years old had 2.19 greater odds of refusing to provide SO information (95% CI, 1.04-4.62; P = .04). After adjusting for other demographic characteristics, only bisexual patients continued to have significantly increased odds of refusing to provide SO compared with straight patients (odds ratio [OR], 2.40; 95% CI, 1.26-4.56; P = .008).
Both patients and providers indicated nonverbal self-report as their preferred method of SO and GI information collection (Tables 10 and 11). This preference is supported by the qualitative interviews, in which patients identified normalization and recognition as major facilitators of SO/GI collection.
Phase 1: Preferred Method of Sexual Orientation Collection in the ED by Patients and Providers.
Phase 1: Preferred Method of Gender Identity Collection in the ED by Patients and Providers.
Twenty-two of 25 SAB members participated in round 1 (participation rate: 84.6%); all 22 SAB members who participated in round 1 also participated in round 2 (participation rate: 100%).
A tabular summary of Delphi round results can be found in Table 12. Most SAB members agree that SO/GI should be collected by the same method, at the same time during the ED visit, and with the same frequency. Free responses emphasized the importance of allowing patients the opportunity to update their SO/GI information at a frequency of their own discretion; however, the most appropriate frequency for collecting SO/GI varied. SAB members recognized that SO and GI are very different attributes; nonetheless, for the purpose of implementation most SAB members agreed that these 2 data points should be collected simultaneously.
Phase 2: Results of Delphi Rounds With SAB.
Multimodal collection (ie, nurse verbal collection in mode 1 and registrar written collection in mode 2) was the most preferred method for collecting SO/GI. No SAB members prioritized verbal collection as the preferred method for collecting SO/GI. Free responses recognized that for feasibility and operations, the same method should be employed to collect both SO and GI. This presents a challenge because the needs of sexual and gender minorities may differ. Free responses recognized that each ED may need to adapt SO/GI collection based on feasibility and operational factors.
SAB members were most supportive of collecting SO/GI once at the first ED visit but allowing patients to update this information at their own discretion at subsequent visits. Free responses highlighted other ideas: Ideally, SO/GI would already be in the patient's EHR (through primary care or outpatient care) and would not need to be collected in the ED. Few SAB members felt SO/GI should be collected at every visit.
The most preferred multimodal approach was verbal SO/GI information collection by the nurse followed by electronic form collection administered by the nurse. The nurse was selected as the preferred person to administer the form for collecting SO and GI as part of any multimodal approach. Rationales for choosing nurses over other health care providers included their clinical expertise, their rapport with patients, and the amount of time they spent with patients as compared with registrars and physicians. In general, there was more support for multimodal approaches involving the nurse and registrar; there was less support for multimodal approaches involving the physician. Most SAB members agreed that if the preferred approach uses form collection followed by verbal confirmation, the form should include an option that indicates the patient is not willing to disclose on the form and not willing to participate in a verbal discussion with ED staff. There was clear consensus that if the preferred approach uses form collection followed by verbal confirmation, the form should include an option that indicates the patient is not willing to disclose on the form but is willing to participate in a verbal discussion with ED staff. Most SAB members supported the idea that SO/GI data should be documented and available within the EHR.
We asked SAB members to determine the best methods for implementing SO/GI collection. Options included multimodal collection (a combination of different methods, eg, form with option to provide information verbally at a later time), form collection (eg, filling out a paper form, filling out an electronic form at a kiosk, filling out an online form from a computer at home), or verbal collection (a conversation between a member of the ED team and the patient). The most common free response answer was multimodal vs form collection.
Because several SAB members’ expertise included SGM cultural competence training, participants were asked for input on best practices for educating and training ED staff. The most preferred education format was in-service training for all ED staff, and 95% of SAB members expected this training to be mandatory. Of SAB members, 50% thought annual staff training was appropriate. Many suggested that training at hire and at least annually thereafter (if not more frequently) was best.
The study was registered under ClinicalTrials.gov, number NCT 02701049. All ClinicalTrials.gov tables are included in Appendix 7.
Between February 8, 2016, and March 29, 2017, we tested 2 SO/GI information collection approaches sequentially: nurse verbal collection during the clinical encounter (mode 1) and then registrar written collection during registration (mode 2). During mode 1, a total of 109 994 patients were seen in the ED at the participating sites. Of these patients, 19 742 (18%) had SO/GI recorded in the EHR by a nurse or other clinician. During mode 2, a total of 88 143 patients were seen in the EDs. Of these, 3630 (4%) provided SO/GI information via a form administered by the registrar. Thus, SO/GI data were collected from 23 372 patients during phase 3. Because our IRB approval did not permit us to analyze the EHR charts of those patients whom we did not enroll, we are unable to report patient demographics for all eligible patients.
Among patients who had SO/GI collected, 22 699 patients did not identify as SGM and 673 patients identified as SGM. In total, 551 of 673 SGM patients (82%) were eligible to participate in the study. From those, 122 were excluded because they were not consentable based on medical or demographic exclusion criteria (n = 95) or they were identified as “other” during SO/GI collection (n = 27); these exclusion criteria are detailed in the Limitations section of the discussion. Among eligible SGM patients, 318 were not enrolled because a research assistant was not available (n = 289), the clinical team advised study staff not to approach the patient (n = 24), or the patient had previously enrolled in the study (n = 5). Of the 233 SGM patients invited to participate in the study, 213 (92%) completed outcome surveys.
After identifying all SGM patients who completed outcome surveys, we matched these 213 participants on hospital site, age (±5 years), and ESI (±1) to non-SGM and to Blank Field patients, resulting in 207 triads. Of the 621 patients in fully matched triads, 540 (87%) were included in final analyses: 342 from mode 1 (nurse verbal collection) and 198 from mode 2 (registrar written collection). Figures 5a and 5b provide full details on enrollment during each intervention mode.
Patient Enrollment During EQUALITY Intervention Mode 1.
Patient Enrollment During EQUALITY Intervention Mode 2.
Overall, patients enrolled during mode 2 were significantly younger and had a lower illness severity (Emergency Severity Index 4 or 5) than mode 1 patients (Table 13). When comparing patients by match group and mode, patients in mode 2 were significantly younger than patients in mode 1. Additionally, a significantly higher proportion of patients had an illness severity of 4 in mode 2 compared with mode 1 across all match groups. A higher proportion of patients who identified as female participated in mode 2 among the SGM match group only. Additional characteristics of patients by match group and mode are found in Table 14.
Phase 3: Characteristics of Patients Enrolled During the EQUALITY Intervention.
Phase 3: Characteristics of Patients Enrolled During the EQUALITY Intervention by Patient Match Group and Intervention Mode.
The intervention also included educating and training ED physicians, physician assistants, nurses, and registrars on SGM health disparities and terminology.
Nurses (mode 1) and registrars (mode 2) were asked, but not required, to collect SO/GI information as part of their workflow. During mode 1, nurses entered SO/GI information directly into the EHR, from which analytic reports were run to identify the number and proportion of patients from whom SO/GI was collected. During mode 2, registrars asked patients to complete SO/GI information as part of a demographics collection form that was administered electronically or on paper, dependent on site. Research staff entered forms that were completed on paper, which were then linked to EHR data to determine the number and proportion of patients from whom SO/GI information was collected. Nurse verbal collection of SO/GI information was integrated into the workflow at every study site and was considered standard of care in other departments by the hospital systems. Registrar written collection was a new process at every site, requiring workflow reorganization and modification in addition to staff training on SGM health.
Average modified CCAT scores during nurse verbal collection were 89.5 (SD, 20.5; 95% CI, 85.7-93.3), 91.8 (SD, 18.9; 95% CI, 88.3-95.3), and 92.7 (SD, 15.9; 95% CI, 89.8-95.7) for SGM, non-SGM, and Blank Field patients, respectively. During registrar written collection, average modified CCAT scores were 95.6 (SD, 11.9; 95% CI, 92.7-98.5), 93.2 (SD, 13.6; 95% CI, 89.9-96.5), and 93.6 (SD, 14.7; 95% CI, 90.0-97.2) among SGM, non-SGM, and Blank Field patients, respectively. Differences between modified CCAT scores during mode 1 vs mode 2 were significant among SGM patients only, with a 6.1-point difference (P < .05). We found no significant differences between modes 1 and 2 for any of the secondary outcome measures (Table 15).
Phase 3: Primary and Secondary Outcomes of the EQUALITY Intervention by Patient Match Group and Intervention Mode.
Because we were interested in the experiences of SGM patients in particular, we conducted ordered logistic regression analyses stratified by patient match group (SGM, non-SGM, and Blank Field). In unadjusted regression models stratifying by patient match group, SGM patients had 1.98 increased odds of a better CCAT score category, a difference of 10 points, between modes 1 and 2 (95% CI, 0.99-3.98; P = .05). After adjusting for covariates, the strength of the association increased to 2.57 (95% CI, 1.13-5.82; P = .02; Table 16a). The odds of the CCAT score increasing between modes 1 and 2 among non-SGM patients or patients not asked to provide SO/GI were not significant (Table 16b).
Phase 3: Ordered Logistic Regression of Increasing Patient Satisfaction Score for All Patients.
Phase 3: Ordered Logistic Regression of Increasing Patient Satisfaction Score by Patient Group.
To test the robustness of our findings, we conducted multiple sensitivity analyses. We examined ED wait time as a predictor of CCAT scores because wait time is a predictor of satisfaction with ED encounters.50,51 Wait time was missing among 4% of patients, which would result in losing 12% of the sample after dropping the matched triads (SGM, non-SGM, and Blank Field). We completed the regression analyses with wait time both as a continuous variable and as a dichotomous variable (wait time missing: yes/no). Because neither of these wait time variables was a significant predictor of modified CCAT score overall or in the stratified analyses, we chose not to include it in the final model.
Additionally, we completed ordered logistic regression of the full 7-item CCAT score. Among the 30% of patients (n = 162) who answered questions that were not relevant to their ED experience (enabling a calculation of the full 7-item CCAT score), patients participating during mode 2 were 2.8 times more likely (P = .02) to answer in a higher CCAT score category, a difference of 10 points corresponding to increased satisfaction with their ED experience, than individuals answering in mode 1.
Of note, only 16 individuals from each patient match group provided complete CCAT survey data in mode 2, and, interestingly, Blank Field patients represented the only group significantly more likely to have answered in a higher CCAT score category during mode 2 than during mode 1 (OR, 6.74; 95% CI, 1.09, 41.6; P = 0.04). Because of low numbers of complete surveys, we included incomplete surveys in the analyses.
SGM populations report poorer health30,31 and less access to health insurance and health services13 compared with heterosexual, cisgender populations. Approximately 10 million Americans now identify as SGM,52 and although some disparities have been identified, the lack of data on SO/GI is a major challenge to understanding and addressing SGM health inequities.53-55 To begin addressing this issue, SGM individuals have been identified as a target group for health improvement by Healthy People 2020,56 and the US Department of Health and Human Services, National Academy of Medicine, Health Resources and Services Administration, and Joint Commission now recommend routine collection of SO/GI data in federally funded population health surveys and in EHRs.4,23,33,47
Obtaining data through population surveys is important for identifying and addressing health disparities; in addition, collecting data in a patient's EHR can help clinicians provide more complete care.27,58 In 2015, the Centers for Medicare & Medicaid Services released Meaningful Use Stage 3 Guidelines that require all certified EHR systems to have the capacity to record SO/GI data,59 but very few health systems or hospitals report routinely collecting such data.27 EDs are the source of nearly half of inpatient admissions in the United States and the primary point of entry for uninsured and underinsured patients,28 as well as SGM patients.60 However, few hospital EDs routinely collect SO/GI information,27 and there are no evidence-based methods standards to collect these data in a patient-centered manner. Despite the importance of SO/GI collection for providing high-quality, patient-centered care and the opportunity to collect a large volume of SO/GI information for SGM disparities research, routine collection of data on SO/GI in the ED setting is rare, and optimal patient-centered approaches for collecting this information remain unclear.
Little is known about the effect of SO/GI data collection on patient outcomes; yet, given the importance of SO/GI to population health4,57,61 and clinical care,22,62-64 the objectives of this study were to (1) gather qualitative input on perceived facilitators, barriers, and preferred approaches for collecting SO/GI in the ED setting; (2) develop and prioritize patient-centered approaches for collecting SO/GI; and (3) evaluate the comparative effectiveness of patient-centered approaches for collecting SO/GI in the ED setting. The results of the EQUALITY study will help clinicians and ED leadership make decisions about how to implement SO/GI collection in their organizations in a patient-centered manner.
The results of phase 1 of the EQUALITY study revealed that most Americans are willing to disclose SO in the ED setting, regardless of their SO. The survey data are supported by in-depth, qualitative interviews that suggest discordance between patient and provider views on routine collection of SO; although most providers believe patients will refuse to provide SO information, few patients report that they would refuse to provide such information. Routine collection of SO from all patients signals normalization of SGM within society, and both patients and providers identify nonverbal self-report as the preferred method of collection.
Routine collection of SO information in the health care setting is important not only for individual patients but also for the normalization of SGM individuals within society.65 Normalizing the collecting of SO/GI information for every patient creates a dialogue between patients and providers,66 and promotes a welcoming, inclusive environment.67 Previous research in community health settings shows that patients think it is important for providers to know their SO and that they are willing to provide such information when asked.68 Phase 1 of the EQUALITY study extends these findings to the ED and further iterates not only that patients are willing to provide SO/GI information when asked, but that very few patients will take offense to or refuse to participate in routine collection of SO/GI information.
Findings from phase 2 of the EQUALITY study highlighted potential challenges of SO/GI collection that have been reported previously, including the practicality of collection, unknown willingness of patients to disclose, and use of the information after it has been disclosed.69 Further, discontinuity of EHRs between health systems—and even between departments within 1 system—poses a challenge for SO/GI collection because patients may be prompted for SO/GI information more than once or providers might skip collection, assuming it had occurred in an earlier or separate clinical encounter.
EQUALITY study phase 3 assessed 2 potential methods to collect SO/GI in the ED and found that SGM patients reported significantly higher satisfaction with their ED experience during registrar written collection of SO/GI compared with nurse verbal collection. The difference in mean CCAT scores between nurse verbal collection and registrar written collection (>10) suggests clinically significant differences in patients’ beliefs about the quality of care that they were receiving. No significant differences in satisfaction occurred among non-SGM patients or Blank Field patients. We also found that most patients—across modes and patient match groups—said it was important for all patients to report SO/GI information: 62.4% of patients reported that it was important for all patients to provide SO information (95% CI, 0.580-0.669; P < .001) and 70.6% of patients reported that it was important for all patients to provide GI information (95% CI, 0.664-0.748; P < .001). These findings indicate that collection is important to patients and that the collection method may not matter to patients who are not affected by self-identifying as SGM, but nonverbal self-report is clearly favored by patients who disclose their SGM status during an implementation trial.
Many researchers have called for routine SO/GI collection to create health care environments that facilitate disclosure and recognition of SGM patients,53,70,71 and Meaningful Use Stage 3 Guidelines set the expectation that many health care systems should have the ability to collect these data. Implementing SO/GI collection in a patient-centered manner helps create an environment that facilitates disclosure and provides the opportunity for SGM patients to be recognized and acknowledged within the health care setting.71
Previous research has indicated that many SGM patients feel that disclosing SO/GI to a clinician is as difficult as disclosing to other people in their lives,72 so the findings of EQUALITY phase 3 are critical to implementing SO/GI collection in a patient-centered manner.
The current study indicates that SGM patients have much more positive experiences when SO/GI is collected via nonverbal self-report. Based on findings from phase 1, patients are willing to provide SO/GI in the ED if the information is collected safely and appropriately. Emphasis on standard collection for the purposes of population health facilitates disclosure, so standard scripts for staff may help alleviate patient concerns regarding SO/GI collection.
Staff training has been identified as a central consideration for implementing SO/GI collection, both in the EQUALITY study and by other researchers.54,55 Focusing educational messaging on 1 or 2 critical messages has been successful in implementing SO/GI collection in large US academic medical centers.34
Another major consideration when implementing SO/GI collection is the impact on staff workflow. Patient-centered SO/GI collection will be successful only if it is routinely collected by staff. Overall SO/GI collection from all patients was lower during nonverbal self-report, but this can be explained by the nature of the interventions. Nurse verbal collection of SO/GI information was integrated into the workflow at every study site. In contrast, registrar written collection was a new process at every site, requiring workflow reorganization and modification in addition to staff training on SGM health. This finding speaks to the importance of integrating SO/GI collection smoothly into the workflow in order to make SO/GI collection a success. Additionally, because our findings indicate that patients prefer nonverbal self-report of SO/GI, hospitals and EDs must use great care when designing their SO/GI collection systems, to ensure that clinicians are able to view and use the important personal information that patients provide.
Results from phase 1 national surveys indicate that few patients would refuse to provide SO/GI information, regardless of SO and GI. Phase 3 of the EQUALITY study took place in both academic and community hospitals in the Northeastern and Mid-Atlantic United States, and the results indicate that SGM patients reported more comfort with their ED experience when written SO/GI collection was implemented as compared with verbal collection. Patients’ preferred method for collection may differ in other areas of the United States, but considering that nonverbal self-report is the more conservative method, it would likely be acceptable in any region of the United States. Additionally, SO/GI collection can be customized to each organization's unique environment, as UC Davis demonstrated when it implemented both online self-report and provider verbal collection of SO/GI.34
The results of the EQUALITY study clearly indicate that SGM patients prefer nonverbal self-report to provide SO/GI information in the ED, whereas non-SGM patients appear to have no preference for collection. Results from the phase 1 national survey indicated that bisexual patients are significantly more likely to refuse to provide SO/GI; however, phase 3 was not powered to look at differences in preferred collection methods among lesbian, gay, bisexual, and transgender patients. Additionally, transgender patients are particularly vulnerable to health disparities and discrimination in clinical settings,71,73 so understanding patient-centered ways to collect and use GI information in the ED is critical.
Each phase of the EQUALITY study had potential limitations. In phase 1, interviews were conducted in 1 urban area of the United States, thus potentially identifying themes unique to SGM patients in 1 region or in urban areas vs rural areas. However, the subsequent national survey, which confirmed the qualitative themes on the national level, helped to alleviate those concerns. In addition, all interviewed providers identified as heterosexual and cisgender. Nonheterosexual and transgender providers could have different views on SO/GI data collection, although the medical community lacks data on the proportion of providers who identify as SGM and, consequently, their attitude toward SGM health care.
In phase 3, patients who identified their SO/GI as “other” were not eligible for inclusion in the outcome surveys because some patients who did not necessarily fall into the SGM spectrum were being classified as “other” in the EHR. For example, at least 1 priest and 1 nun were listed as “other.” RAs often did not know this until they approached these patients; because enrolling them was taking away from recruiting other patients, we decided to exclude patients with this classification. Thus, phase 3 missed out on data from patients who have fluid SO or gender identities, or who choose not to define themselves in a particular category. It is unclear how these patients felt about each mode of SO/GI collection and their experiences with SO/GI collection in the ED, thus potentially attenuating the findings.
Additionally, in phase 3, patients who were admitted to the ED with a psychiatric diagnosis were not eligible for participation because of potential issues confounding both comfort and satisfaction with the ED encounter. SGM patients are particularly vulnerable to poor mental health,57 and excluding these patients loses the voice of patients who may benefit most from improved SGM sensitivity. Excluding SGM patients admitted for mental health diagnoses likely attenuated the phase 3 findings toward the null. Non-English–speaking patients were also ineligible for participation. Because EDs are disproportionately important sources of care for many without insurance, including many immigrants, refugees, and asylees,74 excluding these patients’ experiences creates a gap in fully understanding the experience of SO/GI collection in ED settings.
Further, under advisement of our SAB, we decided against collecting follow-up surveys from patients who declined to provide SO/GI information to registrars and nurses. Although this was a small proportion of patients with not enough statistical power to include in analyses, they may have had important contributions to understanding experiences of comfort and satisfaction with care provided in the ED.
A final limitation was that we could not mandate SO/GI collection by registrars and nurses; consequently, a substantive proportion (88%) of patients were not asked their SO/GI information. Because we did not investigate reasons why ED staff did not ask some patients for SO/GI information and our IRB approval did not permit us to analyze demographic information about eligible but unenrolled patients, we cannot know for sure if systematic bias occurred in SO/GI reporting among ED patients (eg, if ED staff used contextual clues to decide whether to ask patients for SO/GI information). However, <0.001% of patients of whom SO/GI was asked declined to provide this information, and “declined” patients did not differ significantly from “asked” patients. Because we achieved a priori sample sizes, it is unlikely that this affected our major finding as we found no systematic biases in data collection other than staff inclination to collect data. Although our results would be more robust if more patients had SO/GI collected, every method of analysis resulted in SGM patients demonstrating significantly better CCAT scores during written collection vs verbal collection. Implementation research is necessary to explore factors affecting routing SO/GI collection in a busy clinical setting.
Understanding the preferences of the most vulnerable patients, including transgender and SGM patients with mental health diagnoses, is particularly important. Future research should examine patient preferences about SO/GI collection and use, specifically among communities we were unable to survey in our study.
Because a program is only as good as its implementation, future research should examine how best to implement SO/GI collection while examining community-, hospital-, and staff-level factors. Providing flexible and adaptable guidance to hospitals, along with training and technical assistance, will help elucidate how to implement SO/GI collection effectively in a variety of ED settings.
In summary, few patients overall stated that they would refuse to provide SO/GI when asked in the ED (see Tables 8 and 9), in contrast to providers’ beliefs that patients would refuse to provide SO/GI information. Both patients and providers indicate nonverbal self-report as a comfortable and appropriate method of SO/GI information collection. Despite excluding both patients who do not define themselves by 1 particular SO/GI category and SGM patients admitted to the ED with mental health diagnoses, our results support existing literature that suggests that patients who are asked about SO/GI do not have lower satisfaction than Blank Field patients.
ED clinicians and leadership will find these results particularly useful as many move toward routine collection of SO/GI in the ED. But, even though the findings indicate that patients report more comfort and satisfaction with their care during self-report SO/GI information collection, organizational factors must be considered when implementing SO/GI collection. Ensuring that patients, patient advocates, clinicians, and ED staff are all included from the beginning planning stages is critical to making the implementation of standard SO/GI collection a success.
Future research should examine the experiences and preferences of those SGM patients who are most in need of sensitive, patient-centered care in the ED. Additionally, research examining how community, hospital, and staff factors influence implementation of SO/GI collection is critical to ensuring successful implementation throughout the United States.
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#AD-110114-IC) Further information available at: https://www.pcori.org/research-results/2013/comparing-ways-ask-patients-about-sexual-orientation-and-gender-identity
Study Team and Stakeholder Advisory Board Members (PDF, 181K)
Qualitative Interview Guides (PDF, 188K)
National Surveys (PDF, 329K)
Delphi Round Questionnaires (PDF, 221K)
Patient Outcome Survey (PDF, 211K)
Final Study Protocol (PDF, 239K)
ClinicalTrials.gov Tables (PDF, 201K)
Haider A, Schneider E, Schuur J, et al. (2019). Comparing Ways to Ask Patients about Sexual Orientation and Gender Identity in the Emergency Room—The EQUALITY Study. Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/7.2019.AD.110114IC
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
