Large language models to identify social determinants of health in electronic health records.

Social determinants of health (SDoH) play a critical role in patient outcomes, yet their documentation is often missing or incomplete in the structured data of electronic health records (EHRs). Large language models (LLMs) could enable high-throughput extraction of SDoH from the EHR to support research and clinical care. However, class imbalance and data limitations present challenges for this sparsely documented yet critical information. Here, we investigated the optimal methods for using LLMs to extract six SDoH categories from narrative text in the EHR: employment, housing, transportation, parental status, relationship, and social support. The best-performing models were fine-tuned Flan-T5 XL for any SDoH mentions (macro-F1 0.71), and Flan-T5 XXL for adverse SDoH mentions (macro-F1 0.70). Adding LLM-generated synthetic data to training varied across models and architecture, but improved the performance of smaller Flan-T5 models (delta F1 + 0.12 to +0.23). Our best-fine-tuned models outperformed zero- and few-shot performance of ChatGPT-family models in the zero- and few-shot setting, except GPT4 with 10-shot prompting for adverse SDoH. Fine-tuned models were less likely than ChatGPT to change their prediction when race/ethnicity and gender descriptors were added to the text, suggesting less algorithmic bias (p < 0.05). Our models identified 93.8% of patients with adverse SDoH, while ICD-10 codes captured 2.0%. These results demonstrate the potential of LLMs in improving real-world evidence on SDoH and assisting in identifying patients who could benefit from resource support.

Sexual Health Assessment in Women with Lung Cancer study: Sexual health assessment in women with lung cancer.

BACKGROUND: Sexual health is understudied and underreported in patients with lung cancer, and most data precede the approval of widely used targeted therapies and immune checkpoint inhibitors. The authors sought to evaluate the prevalence of sexual dysfunction in women with lung cancer in our current clinical environment. METHODS: This cross-sectional survey study was administered online to 249 women via the GO2 for Lung Cancer (GO2) Registry, using the Patient-Reported Outcomes Measurement Information System Sexual Function and Satisfaction Measures questionnaire. Participants were recruited between June 2020 to June 2021. Eligibility criteria included age >18 years, self-identifying as a woman, fluency in English, and a lung cancer diagnosis within 10 years. RESULTS: Most (67%) had stage IV lung cancer and 47% were receiving targeted therapy; 66% were undergoing active treatment. Despite 54% of participants reporting "recent" sexual activity, most (77%) indicated having little to no interest in sexual activity and 48% reported recent minimal satisfaction with their sex life. The most common reasons negatively affecting participants' satisfaction with their sex life included fatigue (40%) and feeling sad/unhappy (28%). Common reasons for lack of recent sexual activity included lack of interest (68%) and vaginal dryness or pain (30%). Compared to pre-diagnosis, women were significantly less likely to have recent interest in sexual activity. In multivariable logistic-regression, vaginal dryness showed a significant negative association with recent interest in sexual activity. CONCLUSIONS: Sexual dysfunction is prevalent in women with lung cancer. Sexual health should be integrated into routine care for patients with lung cancer.

Radiation Oncology Residency Training Program Integration of Diversity, Equity, and Inclusion: An Association of Residents in Radiation Oncology Equity and Inclusion Subcommittee Inaugural Program Director Survey.

PURPOSE: The aim of this study was to investigate United States (US) radiation oncology (RO) program directors' (PDs) attitudes and practices regarding racial/ethnic diversity, equity, and inclusion (DEI) to better understand potential effects on underrepresented in medicine (UIM) residents in RO. METHODS AND MATERIALS: A 28-item survey was developed using the validated Ethnic Harassment Experiences Scale and the Daily Life Experiences subscale, as well as input from DEI leaders in RO. The survey was institutional review board-approved and administered to RO PDs. PDs were provided with the American Association of Medical Colleges definition of UIM, that is, "Underrepresented in medicine means those racial and ethnic populations that are underrepresented in the medical profession relative to their numbers in the general population." Descriptive statistics were used in analysis. RESULTS: The response rate was 71% (64/90). Institutional Culture and Beliefs: 42% responded that they had a department DEI director. A minority (17%, n = 11) agreed "I believe that people from UIM backgrounds have equal access to quality tertiary education in the US." The majority (97%, n = 62) agreed "My program values residents from UIM backgrounds." Support and Resources: The majority (78%, n = 50) agreed "My program has resources in place to assist/provide support for resident physicians from UIM backgrounds." Interview and Recruitment: Most PDs (53%) had not taken part in activities aimed at recruiting UIM residents and 17% had interviewed no UIM applicants in the past 5 years for residency. Resident Experiences of Racism: 17% (n = 11) agreed "UIM residents in my program have reported incidents of racism to me," and 28% (n = 18) agreed "I believe that UIM residents in my program have been treated differently because of their race/ethnicity by faculty, staff, coresidents or patients." CONCLUSIONS: Most PDs reported that they did not believe that UIM residents were treated differently in their department because of their race/ethnicity, and only a minority had received reports of racial discrimination experienced by residents. These data contrast resident experiences and suggest a disconnect between DEI perceptions and resident experiences among US RO PDs that should be addressed through increased programmatic action and evaluation.

Tailored Mentorship for the Underrepresented and Allies in Radiation Oncology: The Association of Residents in Radiation Oncology Equity and Inclusion Subcommittee Mentorship Experience.

PURPOSE: There are limited opportunities for mentorship for underrepresented in medicine (URM) trainees and physicians in radiation oncology (RO). The purpose of this study was to create and evaluate a formal mentorship program open to URMs and allies with interests in diversity, equity, and inclusion. METHODS AND MATERIALS: A mentorship program incorporating a virtual platform was designed by the Association of Residents in Radiation Oncology Equity and Inclusion Subcommittee. It was structured to include 6 sessions over 6 months with matched mentor-mentee pairs based on responses to a publicized online interest form. A compilation of evidence-based guidelines was provided to optimize the mentorship relationship. Linked pre- and postprogram surveys were administered to collect demographic data, define baseline goals and level of support, and evaluate program satisfaction. RESULTS: Thirty-five mentor-mentee pairs were matched; 31 mentees completed the preprogram survey and 17 completed the postprogram survey. Preprogram, only 3 mentees (9.7%) reported satisfaction with current mentorship and 5 (16%) reported mechanisms or mentorship in place at their program to support URMs. On the postprogram survey, mentees reported high satisfaction with areas of mentorship, mentor attributes, and the program overall. Opportunities for improvement include implementation of mechanisms to enhance communication with mentor-mentee pairs and maintain longitudinal engagement. CONCLUSIONS: In the first tailored mentorship program in RO for URMs and those with diversity, equity, and inclusion interests, our results demonstrate that there is self-reported interest for better mentorship for URMs in RO, and that a nationwide structured mentorship program can address participants' goals with high satisfaction. Program expansion could provide URMs and allies in RO more opportunities for career development and promote a greater sense of community and inclusion within the field.

Radiation Oncology Resident Perspectives on Diversity, Equity, and Inclusion: An Association of Residents in Radiation Oncology Equity and Inclusion Subcommittee Inaugural Resident Survey.

PURPOSE: In this study, radiation oncology residents were surveyed on perceptions of diversity, equity, inclusion, and belonging in their residency training programs. METHODS AND MATERIALS: A 23-item survey was developed by the Association of Residents in Radiation Oncology Equity and Inclusion Subcommittee resident members and faculty advisors. The survey was divided into 4 sections: institutional culture, support and resources, interview and recruitment, and experiences of bias. The survey was sent individually to residents from all Accreditation Council for Graduate Medical Education-accredited radiation oncology programs. RESULTS: The survey was issued to 757 residents. A total of 319 residents completed the survey, for a response rate of 42%. All postgraduate years and geographic regions were represented. Significant racial, ethnic, and gender differences were present in survey response patterns. White residents (94%, 164 of 174) and male residents (96%, 186 of 194) were more likely to strongly agree/agree that they were treated with respect by their colleagues and their coworkers than other racial groups (P < .005) or gender groups (P < .008). Only 3% (5 of 174) of White residents strongly agreed/agreed that they were treated unfairly because of their race/ethnicity, while 31% (5 of 16) of Black residents and 10% (9 of 94) of Asian residents strongly agreed/agreed (P < .0001). Similarly, Hispanic residents were more likely to strongly agree/agree (24%, 5 of 21) than non-Hispanic residents (7%, 20 of 298) (P = .003). Regarding mentorship, there were no differences by gender or ethnicity. There were differences by race in residents reporting that they had a supportive mentor (P = .022), with 89% (154 of 174) of White residents who strongly agreed/agreed, 88% (14 of 16) of Black residents, and 91% of Asian residents (86 of 94). CONCLUSIONS: This survey reveals that experiences of support, mentorship, inclusion, and bias vary significantly among radiation oncology residents based on race, ethnicity, and gender. Radiation oncology has opportunity for growth to ensure an equitable experience for all residents.

The Current State of Departmental Diversity, Equity, and Inclusion Efforts Within US Academic Radiation Oncology Departments.

PURPOSE: Promoting a diverse workforce of health care professionals that delivers equitable patient care is an important goal in oncology, as in all of medicine. Although most medical schools have a diversity office and associated initiatives, little is known about radiation oncology (RO) department-level efforts to promote diversity, equity, and inclusion (DEI). We describe the current state of DEI leadership and initiatives in RO departments in the US to guide future policies and programs. METHODS AND MATERIALS: A total of 124 US RO departments affiliated with a medical school were contacted to identify departmental DEI leadership. Identified DEI leaders were asked to complete an anonymous survey assessing characteristics of their departmental DEI leadership, committee/organizational structure, activities, and perceived barriers to, and effect of, their work. Descriptive statistics are reported. RESULTS: Among 85 RO departments that responded (68.5% response rate), 48 (56.5%) reported having a departmental DEI leader. Thirty-four DEI leaders completed the survey (70.8%). Of those who answered each survey question, most DEI leaders were assistant or associate professors (n = 24, 82.8%), women (n = 19, 73.1%), and identified with at least one non-White race or Hispanic ethnicity (n = 15, 53.6%). Nineteen (57.6%) had an associated departmental DEI committee; with 10 of these starting in 2020 or later. Few DEI leaders had administrative support (38.2%), funding (29.4%), protected time (23.5%), or increased compensation for added duties Fifteen (50.0%) believed their DEI-focused efforts were considered for promotion. The most reported initiatives included offering programming/education, supporting students from backgrounds underrepresented in medicine, improving recruitment practices/hiring, and implementing pipeline/pathway projects. The perceived impact of DEI initiatives included an increased culture of respect (89.7%), improved health care disparity awareness (75.9%), and improved systemic/structural racism awareness (79.3%). CONCLUSIONS: Departmental DEI efforts are increasingly common within RO, however, the structure, resources, and recognition associated with DEI work are variable. Additional dedicated resources and recognition for these efforts will help ensure a culture of inclusive excellence for the RO workforce and patients.

Dónde Están? Latinx/Hispanic Representation in the Oncology Workforce: Present and Future.

This manuscript reviews the status of Hispanic/Latinx physicians in oncology, the benefits and challenges to achieving equitable representation, and potential solutions and actions to increase diversity in the oncology workforce. Persons of Hispanic/Latinx origin comprise 18.7% of the population and 16.8% of the adult population in the United States but are only 4.7% of practicing oncologists. The reasons for the lack of representation of Hispanic/Latinx individuals in medicine are multifaceted and include discrimination and biases, exclusionary practices, financial barriers, and lack of role modeling. As a result, patients are deprived the benefits of a representative workforce, such as improved access, enhanced culturally and linguistically competent care, and minimization of health disparities. Solutions included in the manuscript include a description of efforts by ASCO to improve the representativeness of the oncology workforce through its awards programs and educational efforts, especially for Hispanic/Latinx clinicians. The manuscript also outlines individual actions that attending physicians, senior oncologists, oncology leaders, and hospital/cancer center leadership can take to improve the diversity of the oncology workforce and support our Latinx/Hispanic trainees and colleagues. Improving the representativeness of the oncology workforce will require collective action by institutions, medical societies, and individuals. Nevertheless, widespread commitment to creating an inclusive and supported workforce is necessary to ensure the quality of care for minority patients, reduce existing cancer care disparities, and advance innovation in oncology.

Proyecto ELEVAR: Evaluating the Latinx Experience With the Health Care System Through Variables on Access and Resources.

PURPOSE: Hispanic/Latinx communities encounter disparities in access and availability of appropriate timely health care services, negatively affecting cancer outcomes. Limited data exist on the role of these factors in personal experiences of Hispanic/Latinx with health care systems. We aim to better understand Hispanic/Latinx experience with the health care system, focusing on the impact of the Affordable Care Act (ACA) and applications to clinical practice in patients diagnosed with cancer. METHODS: Competitively obtained data from 2010 to 2019 Association of American Medical Colleges Consumer Survey of Health Care Access (a US population-based sample of potential health care consumers needing health care at least once in the past year) were retrospectively reviewed. Descriptive statistics and univariate and multivariate comparisons of patterns pre- versus post-ACA were performed for respondents self-identifying as Hispanic/Latinx. RESULTS: In a total cohort of 53,523 respondents, Hispanic/Latinx comprised 7.5% (n = 4,094). Post-ACA demographics showed a higher proportion of female and an increase in employment, educational attainment, income, and private insurance coverage, with a decrease in uninsured. Adjusted models showed that post-ACA Hispanic/Latinx patients are less likely to use emergency room if sick or needing medical advice (odds ratio [OR] 0.33; P < .001) and more likely to report chronic conditions involving cancer (OR 2.43; P < .001) or being limited in activities because of mental or emotional problems (OR 1.36; P = .005). CONCLUSION: Although ACA implementation was associated with decreased health care access barriers in Hispanic/Latinx, they continue to experience disparities in chronic health conditions, including cancer, with higher limitations because of emotional or mental health. Ongoing efforts that build upon current strides, toward improved patient satisfaction and overall outcomes, are of paramount importance.

A Paradigm Shift in Radiation Oncology Training.

The coronavirus disease 2019 pandemic has been intertwined with the movement for racial justice in the United States and has highlighted and risks aggravating educational and workforce disparities within radiation oncology. We discuss wide-ranging changes within radiation oncology training that are essential to developing and maintaining diversity, including utilization of competency-based educational models that allow for streamlining of training and examinations; responsiveness to the needs of residents and medical students of different gender, racial/ethnic, and socioeconomic groups; and technological integration to increase educational efficiency and decrease barriers.

Prostate cancer incidence across stage, NCCN risk groups, and age before and after USPSTF Grade D recommendations against prostate-specific antigen screening in 2012.

BACKGROUND: We sought to determine the extent to which US Preventive Services Task Force (USPSTF) 2012 Grade D recommendations against prostate-specific antigen screening may have impacted recent prostate cancer disease incidence patterns in the United States across stage, National Comprehensive Cancer Network (NCCN) risk groups, and age groups. METHODS: SEER*Stat version 8.3.4 was used to calculate annual prostate cancer incidence rates from 2010 to 2015 for men aged ≥50 years according to American Joint Committee on Cancer stage at diagnosis (localized vs metastatic), NCCN risk group (low vs unfavorable [intermediate or high-risk]), and age group (50-74 years vs ≥75 years). Age-adjusted incidences per 100,000 persons with corresponding year-by-year incidence ratios (IRs) were calculated using the 2000 US Census population. RESULTS: From 2010 to 2015, the incidence (per 100,000 persons) of localized prostate cancer decreased from 195.4 to 131.9 (Ptrend  < .001) and from 189.0 to 123.4 (Ptrend  < .001) among men aged 50-74 and ≥75 years, respectively. The largest relative year-by-year decline occurred between 2011 and 2012 in NCCN low-risk disease (IR, 0.77 [0.75-0.79, P < .0001] and IR 0.68 [0.62-0.74, P < .0001] for men aged 50-74 and ≥75 years, respectively). From 2010-2015, the incidence of metastatic disease increased from 6.2 to 7.1 (Ptrend  < .001) and from 16.8 to 22.6 (Ptrend  < .001) among men aged 50-74 and ≥75 years, respectively. CONCLUSIONS: This report illustrates recent prostate cancer "reverse migration" away from indolent disease and toward more aggressive disease beginning in 2012. The incidence of localized disease declined across age groups from 2012 to 2015, with the greatest relative declines occurring in low-risk disease. Additionally, the incidence of distant metastatic disease increased gradually throughout the study period.

Referral and follow-up after mental health screening in commercially insured adolescents.

PURPOSE: Although mental health screening is recommended for adolescents, little is known about the predictors of referral to mental health services or engagement in treatment. We examined predictors of mental health referral from primary care and service use for commercially insured youth who had been screened using the Pediatric Symptom Checklist or Youth-Pediatric Symptom Checklist. METHODS: A retrospective chart review was conducted of commercially insured patients 14-17 years of age who were newly identified by the Pediatric Symptom Checklist or Youth-Pediatric Symptom Checklist at a well-child visit. Comparisons were made with propensity-matched negative adolescents meeting the same criteria. Bivariate analyses were conducted to examine differences between positives and negatives and between referred and nonreferred positives. Logistic regression analyses were performed to assess predictors of mental health referral for positive youth. RESULTS: Medical records of 117 positive and 110 negative youth were examined. Compared with negative youth, positive youth were significantly more likely to be referred for mental health treatment (p < .0001) and receive specialty mental health services (p < .0001). Of the positives, 54% were referred for mental health care and 67% of them accepted. However, only 18% completed a face-to-face mental health visit in the next 180 days. Pediatric Symptom Checklist score (odds ratio, 1.21; confidence interval, 1.03-1.42), parental or personal concern (odds ratio, 10.87; confidence interval, 2.70-43.76), and having depressive symptoms (odds ratio, 9.18; confidence interval, 1.49-56.60) were predictive of referral. CONCLUSIONS: Despite identification after behavioral health screening, limited treatment engagement by referred patients persists. Primary care physicians and mental health specialists must enhance their efforts to engage and monitor identified patients.

Through our eyes: exploring African-American men's perspective on factors affecting transition to manhood.

BACKGROUND: Premature mortality and disparities in morbidity observed in African-American men may be associated with factors in their social, economic, and built environments that may be especially influential during the transition to adulthood. OBJECTIVE: To have young, African-American men from Los Angeles County identify and prioritize factors associated with their transition to manhood using photovoice methodology and pile-sorting exercises. DESIGN: Qualitative study using community-based participatory research (CBPR) and photovoice PARTICIPANTS: Twelve African-American men, ages 16-26 years, from Los Angeles County, California. APPROACH: We used CBPR principles to form a community advisory board (CAB) whose members defined goals for the partnered project, developed the protocols, and participated in data collection and analysis. Participants were given digital cameras to take 50-300 photographs over three months. Pile-sorting techniques were used to facilitate participants' identification and discussion of the themes in their photos and selected photos of the group. Pile-sorts of group photographs were analyzed using multidimensional scaling and hierarchical cluster analysis to systematically compare participants' themes and identify patterns of associations between sorted photographs. Sub-themes and related quotes were also elicited from the pile-sorting transcripts. The CAB and several study participants met periodically to develop dissemination strategies and design interventions informed by study findings. KEY RESULTS: Four dominant themes emerged during analysis: 1) Struggles face during the transition to manhood, 2) Sources of social support, 3) Role of sports, and 4) Views on Los Angeles lifestyle. The project led to the formation of a young men's group and community events featuring participants. CONCLUSIONS: CBPR and photovoice are effective methods to engage young, African-American men to identify and discuss factors affecting their transition to manhood, contextualize research findings, and participate in intervention development.