Key difference-making conditions distinguishing ASCO members’ institutions that collect sexual orientation and gender identity (SOGI) data from those that do not.

Abstract

e18554 Background: Despite ASCO and NIH calling it a priority, sexual orientation and gender identity (SOGI) data are not routinely collected in most cancer care settings. The aim of this analysis was to identify the key difference-making conditions that distinguish ASCO members’ institutions that collect SOGI data from those that do not. Methods: In fall of 2020, an anonymous 54-item electronic survey was distributed to ASCO members in oncology practice or research, presumably from non-overlapping institutions. We used configurational analysis to identify difference-making configurations for the presence and absence of institutional SOGI data collection. This approach applies Boolean logic and set theory, and like machine learning, is both novel and agnostic. Results: Three factors consistently distinguished sites (n = 45) that reported collecting both SO and GI data (n = 25) from those that did not (n = 20). The positive model consisted of two solution pathways: (1) replying “always” or “most of the time” to the question “Do you ask your patients what pronouns they want you to use for them?” or (2) replying “strongly agree” or “agree” to the statement “Institutional leadership supports collecting sexual orientation and gender identity (SOGI) data from patients” and replying “yes, the health record has a specific page/tab/section for SOGI data” to the question “Does the electronic health record at your institution have a specific section to collect information about patients' SOGI?” (consistency = .87, coverage = .80). The negative model consisted of three solution pathways: (3) replying “strongly disagree” or “disagree” that leadership supports collecting SOGI data OR (4) replying “no, the health record does not allow for collection of SOGI data” or “not sure” to whether the EHR supports SOGI data collection OR (5) replying “no, the health record does not have a specific section, but it can be recorded as unstructured data in notes” to whether their EHR supports SOGI data collection AND replying “sometimes” or “occasionally” to how often they ask patients about pronoun preferences (consistency = .94, coverage = .80). Conclusions: Leadership support, electronic health record structure, and clinicians asking patients about pronouns consistently distinguished institutions that collected SOGI data from those that did not in this sample. Additional data should be collected to confirm the results, given the small sample size in this study.