Demographic Disparities Research Articles

Demographic Bias in Low-Resolution Deep Face Recognition in the Wild

Face biometrics play a primary role in smart cities, from consumer- to organizational-level applications. This class of technologies has been recently shown to emphasize performance disparities across gender and ethnic groups. Prior work on demographic bias in deep face recognition has tended to focus exclusively on high-resolution images, leaving out low-resolution images captured, for example, by a certain surveillance camera at a distance. Notable reasons for this focus include the lack of low-resolution face image data sets that report user's demographic attributes. Therefore, demographic bias in low-resolution deep face recognition in the wild still remains under-explored. In this paper, we propose a framework for exploring demographic disparities in low-resolution face recognition at a distance. To this end, we devised a deep generative approach to turn high-resolution face images to their low-resolution counterpart. We then trained state-of-the-art face recognition models on different combinations of high- and low-resolution images. Finally, we assessed the model demographic disparities on artificially degraded images from five data sets. Our results show that face images artificially degraded through our generative approach are more realistic than those obtained with other statistical changes. Key disparities across gender and ethnic groups exist and urge timely interventions. Source code: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://cutt.ly/FChLfOC</uri>

Disparities in Melanoma Demographics, Tumor Stage, and Metastases in Hispanic and Latino Patients: A Retrospective Study

Disparities in Melanoma Demographics, Tumor Stage, and Metastases in Hispanic and Latino Patients: A Retrospective Study

A review of the current literature of ethnic, gender, and socioeconomic disparities in venous disease

Venous disease is prevalent, undertreated, and frequently unrecognized. During the past two decades, new treatment modalities have changed how venous disease is approached. Some of these treatment modalities are only available in certain centers or locations and access to care could be inequitable. Although venous disease affects millions in the United States, we have little understanding of the gender, socioeconomic, and ethnic disparities in both superficial and deep venous disease presentation. In an effort to better understand the treatment of male and female patients from different gender, ethnic, and socioeconomic backgrounds, literature searches were conducted to investigate how these patients were evaluated and treated. PubMed was used to search literature using the terms “venous insufficiency,” “superficial venous disease,” “venous thromboembolism,” "deep vein thrombosis,” “DVT,” “May-Thurner,” and “pulmonary embolism,” with gender, sex, racial, and socioeconomic disparities and differences within the keywords. In addition, once articles were discovered, the “similar articles” function was used to expand the search. The included studies were restricted to those reported from 1995 to the present. Given the paucity of data, no study was excluded. It is readily apparent that there is not enough data to make decisions that would modify treatment to affect the outcomes of patients with differing backgrounds and gender. Studies currently are limited to evaluating patients by sex assigned at birth without interrogation of their identified gender. It is imperative that consideration is given to evaluating gender and ethnic differences, because treatment options might need to be tailored accordingly. Outreach and education for underserved patient populations with improvement in access to care must also be incorporated into the healthcare system. Additional work in this area is required. Further data collection and research related to demographic disparities among patients with venous disease is necessary to better understand the differences that could change treatment algorithms tailored to specific groups.

Equity and Inclusion in Pediatric Gastroenterology Telehealth: A Study of Demographic, Socioeconomic and Digital Disparities.

The purpose of our study is to compare in-person and telehealth pediatric care ambulatory visits for gastroenterology (GI) at the Nemours Children's Health System in the Delaware Valley (NCH-DV) based on geospatial, demographic, socioeconomic, and digital disparities. Characteristics of 26,565 patient encounters from January 2019 - December 2020 were analyzed. U.S. Census Bureau geographic identifiers (GEOIDs) were assigned to each participant and aligned with the American Community Survey (2015-2019) socioeconomic and digital outcomes. Reported odds ratios (OR) are telehealth encounter / in person encounter. GI telehealth usage increased 145-fold in 2020 compared to 2019 for NCH-DV. Comparing telehealth to in-person usage in 2020 revealed that GI patients who required a language translator were 2.2-fold less likely to choose telehealth (individual level adjusted OR (I-ORa) [95% C.I.], 0.45[0.30,0.66], p<0.001). Individuals of Hispanic ethnicity or non-Hispanic Black or African American race are 1.3-1.4-fold less likely to utilize telehealth than non-Hispanic Whites (I-ORa [95% C.I.], 0.73[0.59,0.89], p=0.002 and 0.76[0.60,0.95], p=0.02, respectively). Households in census block-groups (BG) that are more likely to utilize telehealth: have broadband access (BG-OR = 2.51[1.22,5.31], p=0.014); are above the poverty level (BG-OR = 4.44[2.00,10.24], p<0.001); own their own home (BG-OR = 1.79[1.25,2.60], p=0.002); and have a bachelor's degree or higher (BG-OR = 6.55[3.25,13.80], p<0.001). Our study is the largest reported pediatric GI telehealth experience in North America that describes racial, ethnic, socioeconomic, and digital inequities. Advocacy and research for pediatric GI focused on telehealth equity and inclusion is urgently needed.

Author Response: The Incidence of and Demographic Disparities among Fifth Metatarsal Fracture Nonunions

Author Response: The Incidence of and Demographic Disparities among Fifth Metatarsal Fracture Nonunions

Letter to the Editor: The Incidence of and Demographic Disparities among Fifth Metatarsal Fracture Nonunions

Letter to the Editor: The Incidence of and Demographic Disparities among Fifth Metatarsal Fracture Nonunions

PD40-09 16 YEAR TRENDS IN THE USE OF PARTIAL NEPHRECTOMY: ARE DEMOGRAPHIC DISPARITIES IMPROVING?

You have accessJournal of UrologyCME1 Apr 2023PD40-09 16 YEAR TRENDS IN THE USE OF PARTIAL NEPHRECTOMY: ARE DEMOGRAPHIC DISPARITIES IMPROVING? Steven Monda, Hans Lui, Gabriela Gonzalez, Manolis Pratsinis, and Marc Dall'Era Steven MondaSteven Monda More articles by this author , Hans LuiHans Lui More articles by this author , Gabriela GonzalezGabriela Gonzalez More articles by this author , Manolis PratsinisManolis Pratsinis More articles by this author , and Marc Dall'EraMarc Dall'Era More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003345.09AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Beginning in the late 1990s partial nephrectomy (PN) became increasingly common for treatment of renal cell carcinoma as an alternative to radical (or total) nephrectomy (RN). Early work established disparities within whom received partial nephrectomy with black, female, and low-income patients less likely to receive partial nephrectomy. Recent trends in the continued adoption and demographic disparities within PN are not known. METHODS: Using 2004-2019 SEER data we identified 131,527 renal cell carcinoma patients with tumors between 1.0 and 10 cm who underwent nephrectomy, PN or RN. We assessed predictors of partial nephrectomy using multivariable logistic regression assessing tumor size, treatment year, income, metropolitan status, race, and ethnicity as independent predictors. We also report proportions of nephrectomies performed by PN within different groups across this period. RESULTS: Between 2004 and 2019 the mean tumor size treated with PN increased from 3.0 cm (95% CI 2.9-3.1) to 3.3 cm (95% CI 3.2-3.3) and the percent of excised T1a tumors, ≤4 cm, treated with PN increased from 37.6% to 70.6% (Figures 1 & 2). While controlling for tumor size and treatment year on multivariable logistic regression, Black, Hispanic, non-metropolitan, and lower income patients were all less likely to receive partial nephrectomy (Odds ratios of 0.82, 0.84, 0.83, and 0.70, all p<0.001). The gap in percentage of patient receiving partial nephrectomy for T1a tumors appeared to be eliminated for female patients, narrow for black patients, but widen for poor and rural patients across the study period (Table 1). CONCLUSIONS: Partial nephrectomy for renal cell carcinoma has become increasingly common over the past 16 years with some plateau over the last 5 years. Despite this, demographic disparities in utilization of partial nephrectomy remain pronounced, particularly for low-income and rural patients. Source of Funding: none © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e1055 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Steven Monda More articles by this author Hans Lui More articles by this author Gabriela Gonzalez More articles by this author Manolis Pratsinis More articles by this author Marc Dall'Era More articles by this author Expand All Advertisement PDF downloadLoading ...

Predictive Factors for Cancer Treatment Delay in a Racially Diverse and Socioeconomically Disadvantaged Urban Population.

Incremental delays in time to treatment initiation (TTI) have been shown to cause a proportional, increased independent risk of disease-specific mortality for breast cancer, colorectal cancer (CRC), head and neck cancer (HNC), non-small-cell lung cancer (NSCLC), and pancreatic cancer. Studies suggest that delays are associated with racial and socioeconomic disparities. We evaluated associations between patient factors and TTI to identify those associated with delay. This is a retrospective cohort study at an urban community-based academic center of patients diagnosed with or referred for curative-intent treatment of breast cancer, CRC, HNC, NSCLC, and pancreatic cancer from January 2019 to December 2021. Variables of interest included Charlson Comorbidity Index (CCI) score, insurance type, language preference, and inpatient admission 30 days before diagnosis. Factors associated with TTI delay, defined as TTI ≥ 30 days, were assessed using multivariable logistic regression. Among 2,543 patients (69% female), the mean age was 63.4 years and the median TTI was 25 days (IQR, 6-44). Within multivariable models, patients treated as outpatient and not admitted 30 days before diagnosis experienced statistically significant greater delay for CRC (odds ratio [OR], 2.82; 95% CI, 1.71 to 4.66) and NSCLC (OR, 2.11; 95% CI, 1.31 to 3.39). Higher CCI score was associated with delay for HNC (OR, 2.63; 95% CI, 1.04 to 6.66) and NSCLC (OR, 1.75; 95% CI, 1.14 to 2.71). For breast cancer, uninsured and Spanish-speaking patients (OR, 1.79; 95% CI, 1.21 to 2.67) experienced increased TTI. Care coordination/compliance (eg, inpatient 30 days before diagnosis), clinical (eg, medical comorbidities), and socioeconomic (eg, uninsured status) predictors for delayed TTI were identified and may inform delay minimizing interventions. Our data support evidence that TTI delays are associated with demographic and socioeconomic disparities. Existing disparities are likely exacerbated by delays that disproportionately affect patients with care coordination/compliance issues, multiple comorbidities, and lower socioeconomic status.

Narrowing the Divide: The Role of Telehealth in Type 1 Diabetes Care for Marginalized Communities.

Widespread uptake of telehealth in response to the COVID-19 pandemic has highlighted geographic, demographic, and economic disparities in access to virtual care. However, research studies and clinical programs that predate the pandemic demonstrate the potential for telehealth-based interventions to improve access to and outcomes of type 1 diabetes (T1D) care for individuals in geographically or socially marginalized communities. In this expert commentary, we discuss telehealth-based care models that have been successful in improving care for marginalized T1D populations. We also outline the policy changes needed to expand access to such interventions to reduce established disparities in T1D care and promote better health equity among people living with T1D.

Abstract P577: Demographic Disparities in the Distribution and Severity of Coronary Artery Disease- An Urban-Suburban Comparative Study

Background: Some small powered studies have suggested a paradoxical finding of less severe coronary artery obstruction in African American (AA) males as compared to Caucasian male patients despite a higher prevalence of CVD mortality in AAs. We performed a retrospective study to further identify the difference in the extent of coronary artery obstruction and the prevalence of multivessel disease between an urban hospital and a suburban hospital. Methods: A chart review of cardiac catheterization reports was performed on 3,579 patients from 2 different hospitals (1 urban and 1 suburban center). Statistical analysis was done using R and Stata. Results: Analyzing the main characteristics of the patient population we found that the suburban patient population (SPP) was 7 years older than the urban patient population (UPP) (P-valve= &lt;0.001), 78.4% of the suburban population were Caucasian while 79.7% of the urban center patient population were African Americans (P value= &lt;0.001). 96% of the patient population at the suburban center were insured while 85% of patients at the urban center were insured (P= &lt;0.001). Comorbidity burden was overall higher in the SPP ( P-Value= &lt; 0.001). SPPs were found to have a higher prevalence of severe CAD than UPP (Table 1). However, no significant difference was noted when comparing the prevalence of multivessel disease (Table 2). Conclusion: SPPs are more likely to have severe CAD than UPPs but with a similar level of CAD distribution.

Local population changes as a spatial varying multiscale process: The Italian case

The population dynamics in Italy show a strong spatial heterogeneity within a framework of persistent demographic territorial disparities. From a local point of view, it is necessary to understand what demographic determinants govern this process. In the paper, we model the population change according to a local (i.e., spatial varying coefficients) multiscale approach. To this aim, local demographic growth rates of each Italian municipality for the period 2011 – 2019 were estimated and modeled by means of a classic a-spatial global model (i.e., ordinary least-square), and a multiscale geographically weighted regression. The multiscale dimensions of local population changes are therefore analyzed by means of three sub-dimensions: Level of influence, scalability, and specificity. The results show that the determinants of local population changes are not spatially constant and that they vary in their effect at different geographical scales.

Disparities in Pediatric Medical and Childcare Disruption Due to COVID-19

This cross-sectional study uses National Survey of Children’s Health data to assess demographic disparities in medical and childcare disruptions during the COVID-19 pandemic.

Geographic, Demographic, and Socioeconomic Disparities and Factors Associated With Cancer Literacy in China: National Cross-sectional Study.

Cancer literacy is associated with several health-related behaviors and outcomes. However, there is still a lack of nationwide surveys for cancer literacy in China. This study aims to evaluate cancer literacy in China, explore disparities, and provide scientific evidence for policy makers. A cross-sectional survey was conducted in mainland China in 2021 using the multistage probability proportional to the size sampling method. Both the reliability and validity of the questionnaire were evaluated. The awareness levels were adjusted by sampling weights and nonrepresentativeness weights to match the actual population distributions. The Rao-Scott adjusted chi-square test was applied to test geographic, demographic, and socioeconomic disparities. A generalized linear model was used to explore potential factors. A total of 80,281 participants aged 15-74 years were finally enrolled from 21 provinces, with an overall response rate of 89.32%. The national rate of cancer literacy was 70.05% (95% CI 69.52%-70.58%). The rates were highest regarding knowledge of cancer management (74.96%, 95% CI 74.36%-75.56%) but were lowest regarding basic knowledge of cancer (66.77%, 95% CI 66.22%-67.33%). Cancer literacy was highest in East China (72.65%, 95% CI 71.82%-73.49%), Central China (71.73%, 95% CI 70.65%-72.81%), and North China (70.73%, 95% CI 68.68%-72.78%), followed by Northeast (65.38%, 95% CI 64.54%-66.22%) and South China (63.21%, 95% CI 61.84%-64.58%), whereas Southwest (59.00%, 95% CI 58.11%-59.89%) and Northwest China (57.09%, 95% CI 55.79%-58.38%) showed a need for improvement. Demographic and socioeconomic disparities were also observed. Urban dwellers, the Han ethnic group, and population with higher education level or household income were associated with prior knowledge. The questionnaire showed generally good internal and external reliability and validity. It remains important for China to regularly monitor levels of cancer literacy, narrow disparities, and strengthen health education for dimensions with poor performance and for individuals with limited knowledge to move closer to the goal of Healthy China 2030.

Telemedicine Utilization Among Transgender and Gender-Diverse Adolescents Before and After the COVID-19 Pandemic.

Introduction: Little is known about how expansion of telemedicine services during the COVID-19 pandemic has affected access to gender-affirming care for transgender and gender-diverse (TGD) youth. The purpose of this study was to explore differences in demographic characteristics and visit completion rates at a multidisciplinary gender clinic before and after telemedicine implementation in March 2020 and among telemedicine users and nonusers. Methods: Data were from electronic health records of Seattle Children's Gender Clinic (SCGC) patients seen between April 2019 and February 2021. We assessed differences in demographic characteristics and care utilization (i.e., encounter type and status) between April 2019 and February 2020 (pre-telemedicine) and April 2020 and February 2021 (post-telemedicine). Results: Of the 1,051 unique patients seen at SCGC during this time period, majority groups were as follows: 62% identified as transmasculine/male, 68% were non-Hispanic White, and 76% resided within 50 miles of the clinic. Statistically significant differences were observed in patient pronouns and insurance type when comparing the pre- and post-telemedicine periods (p < 0.01). Half (52%) of post-telemedicine period encounters were conducted through telemedicine, and telemedicine encounters were significantly more likely to be completed (72% vs. 50%) and less likely to be canceled (21% vs. 46%) compared with in-person encounters. Conclusions: Telemedicine services facilitated continued access to gender-affirming care services for TGD youth during the COVID-19 pandemic. Although the introduction of telemedicine did not exacerbate demographic disparities in access to this care, further research and interventions are warranted to address the ongoing disparities in access to gender-affirming care for youth of color and rural youth.

Characteristics of Atypical Sleep Durations Among Older Compared to Younger Adults: Evidence from the New Zealand Health Survey.

Understanding and supporting sleep is important across the life span. Disparities in sleep status are well documented in mid-life but under-explored among older populations. Data from 40,659 adults pooled from the New Zealand Health Surveys were used; 24.2% were "older adults" (aged ≥65years), 57% were female, and 20.5% of Māori ethnicity. "Long," "short," or "typical" sleep categories were based on age-related National Sleep Foundation guidelines. Multinomial logistic regression examined predictors of atypical sleep, including sociodemographic characteristics, lifestyle factors, and health status. Prevalence of short and long sleep among older adults was 296 (3.0%) and 723 (7.4%), respectively. Correspondingly, prevalence among younger adults was 2521 (8.2%) and 364 (1.2%). Atypical sleep was more significantly associated with indicators of reduced socioeconomic status and ethnicity among younger rather than older adults. Within both age groups, lower physical activity was associated with long sleep status. Higher physical activity and smoking were related to short sleep status among younger adults only. Within both age groups, atypical sleep was associated with SF-12 scores indicating poorer physical and mental health. Having ≥3 health conditions was related to short sleep among the older adults, while for young adults, it was related to both atypical durations. Indicators of negative lifestyle and health factors remain consistent predictors of atypical sleep with aging. However, demographic disparities are less apparent among older atypical sleepers. This study highlights individual and contextual factors associated with atypical sleep patterns which may be important for age-appropriate recognition and management of sleep problems.

Exploring demographic differences and outcomes in young-onset colorectal cancer.

35 Background: Colorectal cancer (CRC) is the second leading cause of cancer-related deaths in the United States (1). The incidence of young-onset CRC is on the rise based on recent data (2). Previous research on other cancers has shown that patients living in rural areas have worse outcomes (3). This study aimed to investigate the geographic and socio-demographic disparities in young-onset CRC patients. Methods: We conducted a retrospective study on colorectal cancer patients between 1975 and 2016 using the Surveillance, Epidemiology, and End Results (SEER) database. Both univariate and multivariable analyses were performed to evaluate overall survival (OS) and disease-specific survival (DSS). Socio-demographic factors, including the location of residence [metro area (MA) or rural area (RA)], gender, race, insurance status, and marital status, were analyzed. Young-onset colorectal cancer (CRC) was defined as the age of diagnosis less than 50 years. The program used for analysis was SAS software 9.4 (SAS Institute Inc., Cary, NC, USA.) Univariate and multivariable models were analyzed using Cox proportional models. Demographic differences between urban and rural population were assessed using Wilcoxon Rank Sum test (continuous variable) and Chi-square test (categorical variables). Results: A total of 73,378 [RA, N = 7,636(10.6%); MA, N = 64,605(89.4%)] young-onset CRC patients were included. RA had more Caucasian patients compared to MA (80.5% vs 60%, p &lt; 0.001). Patients living in RA were more likely to be uninsured (4.8% vs 3%, p &lt; 0.001) than MA. During the study period, the incidence and mortality rates were consistently higher in RA vs. MA. Univariate and multivariable analysis showed that RA had worse OS (multivariate HR = 1.14; p &lt; 0.01) and DSS (multivariate HR = 1.15; p &lt; 0.001) compared to MA. Similarly, males, single and uninsured patients had worse OS and DSS compared to females, married, and insured patients, respectively. Conclusions: Our study identified social and demographic disparities in young-onset CRC incidence and outcomes. Potential causes may include access to healthcare, diet, health behavior, and environmental factors. Future research is needed to understand and attenuate such disparities.

A global portrait of expressed mental health signals towards COVID-19 in social media space.

Globally, the COVID-19 pandemic has induced a mental health crisis. Social media data offer a unique opportunity to track the mental health signals of a given population and quantify their negativity towards COVID-19. To date, however, we know little about how negative sentiments differ across countries and how these relate to the shifting policy landscape experienced through the pandemic. Using 2.1 billion individual-level geotagged tweets posted between 1 February 2020 and 31 March 2021, we track, monitor and map the shifts in negativity across 217 countries and unpack its relationship with COVID-19 policies. Findings reveal that there are important geographic, demographic, and socioeconomic disparities of negativity across continents, different levels of a nation's income, population density, and the level of COVID-19 infection. Countries with more stringent policies were associated with lower levels of negativity, a relationship that weakened in later phases of the pandemic. This study provides the first global and multilingual evaluation of the public's real-time mental health signals to COVID-19 at a large spatial and temporal scale. We offer an empirical framework to monitor mental health signals globally, helping international authorizations, including the United Nations and World Health Organization, to design smart country-specific mental health initiatives in response to the ongoing pandemic and future public emergencies.

Temporal variation of excess deaths from diabetes during the COVID-19 pandemic in the United States

BackgroundAlthough the COVID-19 pandemic has persisted for more than two years with the evident excess mortality from diabetes, few studies have investigated its temporal patterns. This study aims to estimate the excess deaths from diabetes in the United States (US) during the COVID-19 pandemic and evaluate the excess deaths by spatiotemporal pattern, age groups, sex, and race/ethnicity. MethodsDiabetes as one of multiple causes of death or an underlying cause of death were both considered into analyses. The Poisson log-linear regression model was used to estimate weekly expected counts of deaths during the pandemic with adjustments for long-term trend and seasonality. Excess deaths were measured by the difference between observed and expected death counts, including weekly average excess deaths, excess death rate, and excess risk. We calculated the excess estimates by pandemic wave, US state, and demographic characteristic. ResultsFrom March 2020 to March 2022, deaths that diabetes as one of multiple causes of death and an underlying cause of death were about 47.6 % and 18.4 % higher than the expected. The excess deaths of diabetes had evident temporal patterns with two large percentage increases observed during March 2020, to June 2020, and June 2021 to November 2021. The regional heterogeneity and underlying age and racial/ethnic disparities of the excess deaths were also clearly observed. ConclusionsThis study highlighted the increased risks of diabetes mortality, heterogeneous spatiotemporal patterns, and associated demographic disparities during the pandemic. Practical actions are warranted to monitor disease progression, and lessen health disparities in patients with diabetes during the COVID-19 pandemic.

Characteristics of Occupational Environmental Heat Injuries/Illnesses: Survey of Occupational Injuries and Illnesses, 2011 to 2019.

This study describes the characteristics of workers experiencing occupational heat-related injuries/illnesses in the United States and explores the associations between states' average annual temperatures and heat-related injury/illness rates. The number and rate of occupational environmental heat injuries/illnesses were calculated according to age group, gender, race/ethnicity, occupation group, and state from 2011 to 2019. Injury/illness rates were higher among Black and Hispanic workers. Workers in farming, fishing, and forestry; installation, maintenance, and repair; and construction/extraction occupations had the highest rates. There was a positive correlation between states' average annual temperatures and heat-related injury/illness rates. There are demographic and occupational disparities in occupational environmental heat-related injuries/illnesses in the United States and a correlation between these injuries/illnesses and state average annual temperatures. There is a need for policies and other interventions to protect workers from occupational environmental heat injuries/illnesses.

Medical and economic burden of delirium on hospitalization outcomes of acute respiratory failure: A retrospective national cohort

Although delirium in patients with acute respiratory failure (ARF) may evolve in any hospital setting, previous studies on the impact of delirium on ARF were restricted to those in the intensive care unit (ICU). The data about the impact of delirium on ARF hospitalizations outside of the ICU is limited. Therefore, we conducted the first national study to examine the effect-magnitude of delirium on ARF in all hospital settings, that is, in the ICU as well as on the general medical floor. We searched the 2016 and 2017 National Inpatient Sample databases for ARF hospitalizations and created "Delirium" and "No delirium" groups. The outcomes of interest were mortality, endotracheal intubation, length of stay (LOS), and hospitalization costs. We also aimed to explore any potential demographic, racial, or healthcare disparities that may be associated with the diagnosis of delirium among ARF patients. Multivariable logistic regression was used to control for demographics and comorbidities. Delirium was present in 12.7% of the sample. Racial disparities among African Americans were also significant. Delirious patients had more comorbidities, higher mortality, and intubation rates (17.5% and 9.2% vs 10.6% and 6.1% in the "No delirium" group [P < .001], respectively). Delirious patients had a longer LOS and higher hospitalization costs (5.9 days and $15,395 USD vs 3.7 days and $9393 USD in "No delirium" [P < .001], respectively). Delirium was associated with worse mortality (adjusted odds ratio 1.49, confidence interval [CI] = 1.41, 1.57), higher intubation rates (adjusted odds ratio 1.46, CI = 1.36, 1.56), prolonged LOS (adjusted mean ratio 1.40, CI = 1.37, 1.42), and increased hospitalization costs (adjusted mean ratio 1.49, CI = 1.46, 1.52). A racial disparity in the diagnosis of delirium among African Americans hospitalized with ARF was noted in our sample. Patients in small, non-teaching hospitals were diagnosed with delirium less frequently compared to large, urban, teaching centers. Delirium predicts worse mortality and morbidity for ARF patients, regardless of bed placement and severity of the respiratory failure.