The study drew on a uniquely long and rich dataset. Between May 1, 2009, and March 27, 2025, a total of 6,237 nurses, resident and fellow physicians, medical and pharmacy students, faculty physicians, and other staff voluntarily completed the ISP’s web-based stress and depression questionnaire. The program, launched at UC San Diego in 2009 to proactively identify healthcare team members at risk for suicide, invites participants to create self-assigned user IDs and passwords to preserve anonymity. Counselors review each completed questionnaire, write customized responses, and, when indicated, refer individuals to treatment. Because responses cannot be linked across time, the data are cross-sectional at the individual level, with repeat users counted only once using their most recent survey.

The screening instrument stratifies suicide risk into three tiers based on distress, symptom severity, and day-to-day functioning. Tier 1, the highest-risk category, includes respondents scoring 15 or higher on the nine-item Patient Health Questionnaire (PHQ-9) or reporting current suicidal ideation; a score of 10 to 14 also qualifies if accompanied by a prior suicide attempt, intense feelings of anxiety, panic, rage, desperation, or loss of control, or severe functional impairment. Depression severity was assessed with the PHQ-9, a validated measure aligned with DSM-5 criteria, in which respondents rate how often they experienced each depressive symptom over the previous two weeks on a scale from 0 (“not at all”) to 3 (“most or all the time”). A score of 10 or above, the conventional threshold for major depressive disorder, classified a respondent as depressed. Burnout was measured with three items adapted from the Maslach Burnout Inventory, capturing feeling burned out, emotionally drained, or increasingly callous toward people; a positive screen required endorsing “a lot of the time” or more on at least two items. Intense affective states, specifically feeling intensely lonely and feeling hopeless, were drawn from the Affective State Questionnaire, and increased alcohol use was captured with a single item about drinking more than usual over the preceding four weeks.

The scale of distress was striking even before any comparisons were made. Fully 51.9 percent of respondents were classified in the highest suicide risk tier, with an average PHQ-9 score of 8.9 falling in the mild depression range. Nearly 43 percent screened positive for burnout, 24.2 percent reported feeling intensely lonely, 19.9 percent reported feeling hopeless, and 6.7 percent reported drinking alcohol more than usual. The respondents skewed young, with a mean age of 37.1 years, and 70.7 percent were female. They included 1,420 nurses, 915 resident and fellow physicians, 682 medical students, 385 faculty or staff physicians, 329 pharmacy students, and 2,506 others.

To test whether distress and help-seeking differed by race and ethnicity, the investigators coded all outcomes as binary variables and used binary logistic regression, entering race/ethnicity as a seven-level categorical predictor alongside covariates of age, sex, academic position, and whether the survey was completed before or after January 1, 2020. When the omnibus race/ethnicity effect was significant, they ran post-hoc pairwise comparisons between every pair of groups, applying the Benjamini-Yekutieli procedure to control the false discovery rate at p < 0.05 within each outcome.

The results were unambiguous. Compared with non-Hispanic White respondents, the odds of being in the highest suicide risk tier were 43 percent higher for Asian American/Pacific Islander individuals (odds ratio 1.43), 46 percent higher for Hispanic/Latino individuals (OR 1.46), and more than double for multiracial individuals (OR 2.08). Multiracial respondents also outpaced nearly every other group in high-risk classification, exceeding even those identifying as Hispanic/Latino, AAPI, other race/ethnicity, and those who declined to disclose. The same pattern held for clinically meaningful depression, which was 24 to 91 percent more likely among AAPI, Hispanic/Latino, and multiracial respondents than among their non-Hispanic White peers. Intense loneliness was 38 to 73 percent more likely in those groups, and hopelessness was similarly elevated, with Hispanic/Latino respondents showing 69 percent higher odds and multiracial respondents 83 percent higher odds of frequent hopelessness compared with non-Hispanic White respondents.

Just as troubling was what happened next, or did not happen. AAPI respondents were 49 to 63 percent less likely than Black or African American, Hispanic/Latino, multiracial, and non-Hispanic White respondents to report taking prescribed psychiatric medication, a pattern consistent with distress that is substantial yet undertreated. Respondents who preferred not to disclose their racial or ethnic identity were 57 to 60 percent less likely to report medication use than Hispanic/Latino, multiracial, or non-Hispanic White respondents, and they were twice as likely as AAPI respondents to report burnout, alongside greater loneliness and hopelessness than non-Hispanic White respondents. By contrast, psychotherapy use showed no significant racial or ethnic differences overall, nor did perceived stigma about seeking care or judgments about whether institutional mental health resources were adequate, suggesting the disparities run deeper than attitudes alone.

The authors argue that systemic and interpersonal discrimination plausibly accounts for much of the gap. Prior research has documented bullying, racism, and even patients refusing treatment from non-White staff, stressors layered on top of the demanding schedules, exposure to suffering, and moral distress that all healthcare workers face. For AAPI workers, the picture is complicated by the “model minority” stereotype, which casts them as high-achieving and emotionally resilient while silencing intergenerational trauma, immigration stress, and shame about seeking help. Although AAPI individuals are well represented in the healthcare workforce, their numbers thin sharply in medical education and hospital leadership, a phenomenon sometimes called the “bamboo ceiling,” and the category itself masks more than 40 distinct ethnolinguistic groups. The surge in anti-Asian sentiment during the COVID-19 pandemic may have further eroded belonging. For Hispanic/Latino workers, socioeconomic disadvantage, immigration-related stressors, underrepresentation in leadership, and stigma around professional help-seeking are likely contributors. For multiracial workers, the elevated suicide risk echoes the interpersonal theory of suicide, which identifies thwarted belongingness and identity conflict as central drivers of suicidal ideation; national data show multiracial individuals report the highest rates of major depressive episodes of any group.

The nondisclosure group may be the most quietly alarming finding. Earlier work has framed withholding demographic information as “privacy-related behavior,” and found it associated with higher rates of recent suicidal behavior, possibly reflecting efforts to preserve anonymity among people at elevated risk who fear confidentiality breaches, professional repercussions, or mandated interventions. The new findings reinforce that interpretation: people who withhold demographic information appear both more distressed and less engaged with formal treatment, making anonymity-preserving support options not a privacy nicety but a potential lifeline.

The study has limits the authors readily acknowledge. The ISP was built as a screening and outreach tool, not a validated research instrument, and it is unknown whether voluntary respondents differ systematically from healthcare workers who never took the screener, or how response rates varied by racial or ethnic group. The questionnaire offered no Middle Eastern or North African category and did not disaggregate the AAPI group, while small samples for Black or African American and American Indian/Alaskan Native respondents limited statistical power. Race and ethnicity, the authors note, likely serve as proxies for unmeasured cultural, structural, and experiential factors.

Still, the message is hard to miss. Ancillary findings showed that younger respondents, women, those screened after 2020, and nurses reported more distress across nearly every measure, with nurses significantly more likely than faculty physicians to occupy the highest risk tier, score above the depression threshold, and report loneliness and hopelessness. The authors call for proactive, culturally responsive screening; peer-support programs; leadership modeling that normalizes help-seeking; and greater diversity in workforce and leadership to ease the disproportionate burdens on minority professionals. Recognizing and addressing the unequal weight carried by already marginalized members of the healthcare team, they conclude, is essential to building a culture of psychological safety for everyone who cares for patients.