Corey A. Kalbaugh
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Corey A. Kalbaugh has a total of 85 co-authors. Here are the top co-authors by number of shared publications:
- Caitlin W. Hicks
Division of Vascular Surgery and Endovascular Therapy, Department of Surgery, The Johns Hopkins Hospital, Baltimore, MD, USA
5 shared publications Brian Witrick
Department of Public Health Sciences, Clemson University, Clemson, South Carolina
4 shared publicationsKatharine L. McGinigle
Department of Surgery, School of Medicine, The University of North Carolina at Chapel Hill, Chapel Hill
4 shared publicationsKerry A. Howard
Department of Public Health Sciences, Clemson University, Clemson, South Carolina
4 shared publicationsAshley Clark
Center for Survey Research and O’Neill School of Public and Environmental Affairs, Indiana University–Bloomington, Bloomington
3 shared publications- Erika Teresa Beidelman
Robert N. Butler Columbia Aging Center, Columbia University Mailman School of Public Health , New York, NY
3 shared publications
Latest Publications
- Open Access
Trajectories of Ankle-Brachial Index Values and Their Relation to Cardiovascular Health Measured by Life’s Essential 8 in the Atherosclerosis Risk in Communities Study
Background: Peripheral artery disease (PAD) is an occlusive arterial disease primarily affecting the lower extremities. It impacts over 230 million people worldwide and is associated with significant morbidity and mortality. The ankle brachial index (ABI) test is a non-invasive method to detect PAD that compares the blood pressure in the ankle and arm to evaluate lower extremity blood flow. An estimated 20-50% of individuals with detectable PAD are asymptomatic and remain undiagnosed; however, ABI screening in high-risk, asymptomatic populations is not currently guideline-recommended. Few studies have evaluated change in ABI over time in asymptomatic populations. Therefore, we aimed to identify distinct trajectories of ABI values from mid-to late-life. Methods: We utilized data from the Atherosclerosis Risk in Communities (ARIC) study; a longitudinal cohort study initiated in 1987 that enrolled 15,792 participants aged 45-64. ABI measurements were collected at five visits over a 30-year period. We used group-based trajectory modeling to identify trajectories of ABI from mid-to late-life. Final model selection was based on visual fit, statistical criteria, group sizes, and substantive knowledge. Lastly, we compared baseline demographics, social determinants of health, and overall cardiovascular (CV) health, assessed using the American Heart Association’s Life’s Essential 8 (LE8) framework, across trajectory groups. Results: We identified 4,121 participants with ≥3 ABI measurements over the study period in at least one limb. At baseline, participants had an average age of 51.4 ± 4.9 years, were 57.3% female, 22.2% Black, and had an average overall LE8 score of 68.0 ± 13.9 points. Our final model identified three linear trajectories: low-normal, high-normal, and declining. Overall LE8 scores varied significantly across trajectory groups: 67.3 ± 10.7 (high-normal), 61.9 ± 13.3 (low-normal), and 50.1 ± 15.8 points (declining). Women had lower average ABI values, were more likely to experience a declining ABI trajectory, and had a delayed onset of decline compared to men. A greater proportion of Black participants experienced declining ABIs, with earlier, faster, and more severe declines than White participants. Conclusions: Poor overall CV health and common CV risk factors are associated with ABI decline. Targeted ABI screening in middle age may help detect PAD in its beginning stages and support early intervention.
Open AccessTrajectories of Ankle-Brachial Index Values and Their Relation to Cardiovascular Health Measured by Life’s Essential 8 in the Atherosclerosis Risk in Communities Study
Background: Peripheral artery disease (PAD) is an occlusive arterial disease primarily affecting the lower extremities. It impacts over 230 million people worldwide and is associated with significant morbidity and mortality. The ankle brachial index (ABI) test is a non-invasive method to detect PAD that compares the blood pressure in the ankle and arm to evaluate lower extremity blood flow. An estimated 20-50% of individuals with detectable PAD are asymptomatic and remain undiagnosed; however, ABI screening in high-risk, asymptomatic populations is not currently guideline-recommended. Few studies have evaluated change in ABI over time in asymptomatic populations. Therefore, we aimed to identify distinct trajectories of ABI values from mid-to late-life. Methods: We utilized data from the Atherosclerosis Risk in Communities (ARIC) study; a longitudinal cohort study initiated in 1987 that enrolled 15,792 participants aged 45-64. ABI measurements were collected at five visits over a 30-year period. We used group-based trajectory modeling to identify trajectories of ABI from mid-to late-life. Final model selection was based on visual fit, statistical criteria, group sizes, and substantive knowledge. Lastly, we compared baseline demographics, social determinants of health, and overall cardiovascular (CV) health, assessed using the American Heart Association’s Life’s Essential 8 (LE8) framework, across trajectory groups. Results: We identified 4,121 participants with ≥3 ABI measurements over the study period in at least one limb. At baseline, participants had an average age of 51.4 ± 4.9 years, were 57.3% female, 22.2% Black, and had an average overall LE8 score of 68.0 ± 13.9 points. Our final model identified three linear trajectories: low-normal, high-normal, and declining. Overall LE8 scores varied significantly across trajectory groups: 67.3 ± 10.7 (high-normal), 61.9 ± 13.3 (low-normal), and 50.1 ± 15.8 points (declining). Women had lower average ABI values, were more likely to experience a declining ABI trajectory, and had a delayed onset of decline compared to men. A greater proportion of Black participants experienced declining ABIs, with earlier, faster, and more severe declines than White participants. Conclusions: Poor overall CV health and common CV risk factors are associated with ABI decline. Targeted ABI screening in middle age may help detect PAD in its beginning stages and support early intervention.
- 27 October 2025
- Published by Springer Nature
- p. 19-37
National and International Trends in Prevalence and Incidence of Intermittent Claudication Since 1950
- 27 October 2025
- Published by Springer Nature
- p. 19-37
National and International Trends in Prevalence and Incidence of Intermittent Claudication Since 1950
- 7 July 2025
- Published by Wolters Kluwer Health in Epidemiology
- Vol. 36 (6) , 841-848
Disaggregating Health Differences and Disparities With Machine Learning and Observed-to-expected Ratios: Application to Major Lower Limb Amputation
Background: Major lower limb amputation is a devastating but preventable complication of peripheral artery disease. It is unclear whether racial and ethnic and rural differences in amputation rates are due to clinical, hospital, or structural factors. Methods: We included all peripheral artery disease hospitalizations of patients ≥40 years old between 2017 and 2019 in Florida, Georgia, Maryland, Mississippi, or New York (HCUP State Inpatient Databases). We estimated the expected number of amputations using three models: (1) unadjusted, (2) adjusted for clinical factors, and (3) adjusted for clinical factors, hospital factors, and social determinants of health using least absolute shrinkage and selection operator (LASSO). We calculated and compared observed-to-expected ratios and quantified the role of these factors in amputation rates. Results: Overall, 1,577,061 hospitalizations (990,152 unique patients) and 21,233 major lower limb amputations (1.4%) were included. After accounting for clinical differences, we observed amputation disparities among rural Black, Hispanic, Native American, and White patients and nonrural Black and Native American patients. After accounting for hospital factors and social determinants of health, disparities were no longer present among rural White adults (0.93, 95% confidence interval [CI]: 0.77, 1.09); however, disparities persisted among rural Black (1.26, 95% CI: 1.01, 1.51), Hispanic (1.50, 95% CI: 0.89, 2.12), and Native American patients (1.13, 95% CI: 0.68, 1.58) and nonrural Black (1.12, 95% CI: 1.09, 1.15) and Native American (1.15, 95% CI: 0.86, 1.44) patients. Conclusion: Clinical factors did not fully explain differences in amputation rates, and hospital factors and social determinants of health did not fully explain disparities. These findings provide additional evidence that implicit bias is associated with amputation disparities.
- 7 July 2025
- Published by Wolters Kluwer Health in Epidemiology
- Vol. 36 (6) , 841-848
Disaggregating Health Differences and Disparities With Machine Learning and Observed-to-expected Ratios: Application to Major Lower Limb Amputation
Background: Major lower limb amputation is a devastating but preventable complication of peripheral artery disease. It is unclear whether racial and ethnic and rural differences in amputation rates are due to clinical, hospital, or structural factors. Methods: We included all peripheral artery disease hospitalizations of patients ≥40 years old between 2017 and 2019 in Florida, Georgia, Maryland, Mississippi, or New York (HCUP State Inpatient Databases). We estimated the expected number of amputations using three models: (1) unadjusted, (2) adjusted for clinical factors, and (3) adjusted for clinical factors, hospital factors, and social determinants of health using least absolute shrinkage and selection operator (LASSO). We calculated and compared observed-to-expected ratios and quantified the role of these factors in amputation rates. Results: Overall, 1,577,061 hospitalizations (990,152 unique patients) and 21,233 major lower limb amputations (1.4%) were included. After accounting for clinical differences, we observed amputation disparities among rural Black, Hispanic, Native American, and White patients and nonrural Black and Native American patients. After accounting for hospital factors and social determinants of health, disparities were no longer present among rural White adults (0.93, 95% confidence interval [CI]: 0.77, 1.09); however, disparities persisted among rural Black (1.26, 95% CI: 1.01, 1.51), Hispanic (1.50, 95% CI: 0.89, 2.12), and Native American patients (1.13, 95% CI: 0.68, 1.58) and nonrural Black (1.12, 95% CI: 1.09, 1.15) and Native American (1.15, 95% CI: 0.86, 1.44) patients. Conclusion: Clinical factors did not fully explain differences in amputation rates, and hospital factors and social determinants of health did not fully explain disparities. These findings provide additional evidence that implicit bias is associated with amputation disparities.
- Open Access
- 1 May 2025
- Published by American Medical Association (AMA) in JAMA Surgery
- Vol. 160 (5) , 536
Implicit Racial Bias and Unintentional Harm in Vascular Care
Importance: Implicit bias may influence physician treatment decisions and contribute to Black-White health disparities. There are limited data linking implicit bias with actual care delivery and outcomes.Objective: To determine whether implicit racial bias is associated with potentially harmful surgical treatment selection for a cohort of patients with peripheral artery disease–related claudication.Design, Setting, and Participants: This survey study, linked with observational registry data, included eligible clinicians who participate in the Vascular Quality Initiative (VQI) among 960 centers. The VQI includes academic medical centers, teaching hospitals, community hospitals, and private practices. Eligible participants included all vascular specialist VQI members (N = 2512), of whom 218 completed the race implicit association test (IAT) and were linkable to procedure-level data. The study was conducted between October 2021 and October 2022.Exposure: Race IAT.Main Outcomes and Measures: Clinician-level implicit bias results were linked to patient-level registry data of peripheral revascularization procedures performed for claudication. The adjusted odds of performance of any infrapopliteal procedure by specialist implicit bias and patient race were measured via mixed-effects logistic regression models. Implicit bias as a moderator of the association of infrapopliteal procedures for claudication and patient race with 1-year amputation was assessed as a secondary outcome.Results: Among 218 vascular specialists (mean [SD] age, 46 [9] years; 160 [73%] male), 157 (72%) had a pro-White bias. Black patients treated by a physician with pro-White bias had a significant increase in the odds of receiving an infrapopliteal procedure compared with the total sample (adjusted odds ratio [AOR], 1.67; 95% CI, 1.12-2.48). When treated by a specialist with pro-White bias, Black patients had increased odds of 1-year amputation, regardless of anatomic location treated, compared with White patients (AOR, 2.34; 95% CI, 1.20-4.55). Conversely, Black patients treated by a specialist with no bias had similar odds of an infrapopliteal procedure (AOR, 0.93; 95% CI, 0.68-1.26) as the full patient sample and similar odds of 1-year amputation (AOR, 1.29; 95% CI, 0.33-4.99) as White patients.Conclusions and Relevance: These findings indicate that implicit bias is associated with potentially harmful infrapopliteal procedures for Black patients and contributes to Black-White outcome disparities in the US. These results suggest the need for system-level interventions that transparently identify and warn of procedures not aligned with best practices to reduce the negative influence of implicit bias.
- 1 May 2025
- Published by American Medical Association (AMA) in JAMA Surgery
- Vol. 160 (5) , 536
Open AccessImplicit Racial Bias and Unintentional Harm in Vascular Care
Importance: Implicit bias may influence physician treatment decisions and contribute to Black-White health disparities. There are limited data linking implicit bias with actual care delivery and outcomes.Objective: To determine whether implicit racial bias is associated with potentially harmful surgical treatment selection for a cohort of patients with peripheral artery disease–related claudication.Design, Setting, and Participants: This survey study, linked with observational registry data, included eligible clinicians who participate in the Vascular Quality Initiative (VQI) among 960 centers. The VQI includes academic medical centers, teaching hospitals, community hospitals, and private practices. Eligible participants included all vascular specialist VQI members (N = 2512), of whom 218 completed the race implicit association test (IAT) and were linkable to procedure-level data. The study was conducted between October 2021 and October 2022.Exposure: Race IAT.Main Outcomes and Measures: Clinician-level implicit bias results were linked to patient-level registry data of peripheral revascularization procedures performed for claudication. The adjusted odds of performance of any infrapopliteal procedure by specialist implicit bias and patient race were measured via mixed-effects logistic regression models. Implicit bias as a moderator of the association of infrapopliteal procedures for claudication and patient race with 1-year amputation was assessed as a secondary outcome.Results: Among 218 vascular specialists (mean [SD] age, 46 [9] years; 160 [73%] male), 157 (72%) had a pro-White bias. Black patients treated by a physician with pro-White bias had a significant increase in the odds of receiving an infrapopliteal procedure compared with the total sample (adjusted odds ratio [AOR], 1.67; 95% CI, 1.12-2.48). When treated by a specialist with pro-White bias, Black patients had increased odds of 1-year amputation, regardless of anatomic location treated, compared with White patients (AOR, 2.34; 95% CI, 1.20-4.55). Conversely, Black patients treated by a specialist with no bias had similar odds of an infrapopliteal procedure (AOR, 0.93; 95% CI, 0.68-1.26) as the full patient sample and similar odds of 1-year amputation (AOR, 1.29; 95% CI, 0.33-4.99) as White patients.Conclusions and Relevance: These findings indicate that implicit bias is associated with potentially harmful infrapopliteal procedures for Black patients and contributes to Black-White outcome disparities in the US. These results suggest the need for system-level interventions that transparently identify and warn of procedures not aligned with best practices to reduce the negative influence of implicit bias.
- Open Access
- 9 January 2025
- Published by JMIR Publications Inc. in JMIR Perioperative Medicine
- Vol. 8 (1) , e59422
Development and Validation of a Routine Electronic Health Record-Based Delirium Prediction Model for Surgical Patients Without Dementia: Retrospective Case-Control Study
Background: Postoperative delirium (POD) is a common complication after major surgery and is associated with poor outcomes in older adults. Early identification of patients at high risk of POD can enable targeted prevention efforts. However, existing POD prediction models require inpatient data collected during the hospital stay, which delays predictions and limits scalability. Objective: This study aimed to develop and externally validate a machine learning-based prediction model for POD using routine electronic health record (EHR) data. Methods: We identified all surgical encounters from 2014 to 2021 for patients aged 50 years and older who underwent an operation requiring general anesthesia, with a length of stay of at least 1 day at 3 Indiana hospitals. Patients with preexisting dementia or mild cognitive impairment were excluded. POD was identified using Confusion Assessment Method records and delirium International Classification of Diseases (ICD) codes. Controls without delirium or nurse-documented confusion were matched to cases by age, sex, race, and year of admission. We trained logistic regression, random forest, extreme gradient boosting (XGB), and neural network models to predict POD using 143 features derived from routine EHR data available at the time of hospital admission. Separate models were developed for each hospital using surveillance periods of 3 months, 6 months, and 1 year before admission. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC). Each model was internally validated using holdout data and externally validated using data from the other 2 hospitals. Calibration was assessed using calibration curves. Results: The study cohort included 7167 delirium cases and 7167 matched controls. XGB outperformed all other classifiers. AUROCs were highest for XGB models trained on 12 months of preadmission data. The best-performing XGB model achieved a mean AUROC of 0.79 (SD 0.01) on the holdout set, which decreased to 0.69-0.74 (SD 0.02) when externally validated on data from other hospitals. Conclusions: Our routine EHR-based POD prediction models demonstrated good predictive ability using a limited set of preadmission and surgical variables, though their generalizability was limited. The proposed models could be used as a scalable, automated screening tool to identify patients at high risk of POD at the time of hospital admission.
- 9 January 2025
- Published by JMIR Publications Inc. in JMIR Perioperative Medicine
- Vol. 8 (1) , e59422
Open AccessDevelopment and Validation of a Routine Electronic Health Record-Based Delirium Prediction Model for Surgical Patients Without Dementia: Retrospective Case-Control Study
Background: Postoperative delirium (POD) is a common complication after major surgery and is associated with poor outcomes in older adults. Early identification of patients at high risk of POD can enable targeted prevention efforts. However, existing POD prediction models require inpatient data collected during the hospital stay, which delays predictions and limits scalability. Objective: This study aimed to develop and externally validate a machine learning-based prediction model for POD using routine electronic health record (EHR) data. Methods: We identified all surgical encounters from 2014 to 2021 for patients aged 50 years and older who underwent an operation requiring general anesthesia, with a length of stay of at least 1 day at 3 Indiana hospitals. Patients with preexisting dementia or mild cognitive impairment were excluded. POD was identified using Confusion Assessment Method records and delirium International Classification of Diseases (ICD) codes. Controls without delirium or nurse-documented confusion were matched to cases by age, sex, race, and year of admission. We trained logistic regression, random forest, extreme gradient boosting (XGB), and neural network models to predict POD using 143 features derived from routine EHR data available at the time of hospital admission. Separate models were developed for each hospital using surveillance periods of 3 months, 6 months, and 1 year before admission. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC). Each model was internally validated using holdout data and externally validated using data from the other 2 hospitals. Calibration was assessed using calibration curves. Results: The study cohort included 7167 delirium cases and 7167 matched controls. XGB outperformed all other classifiers. AUROCs were highest for XGB models trained on 12 months of preadmission data. The best-performing XGB model achieved a mean AUROC of 0.79 (SD 0.01) on the holdout set, which decreased to 0.69-0.74 (SD 0.02) when externally validated on data from other hospitals. Conclusions: Our routine EHR-based POD prediction models demonstrated good predictive ability using a limited set of preadmission and surgical variables, though their generalizability was limited. The proposed models could be used as a scalable, automated screening tool to identify patients at high risk of POD at the time of hospital admission.
- Open Access
- 16 July 2024
- Published by Wolters Kluwer Health in Journal of the American Heart Association
- Vol. 13 (14) , e033463
Trends and Factors Associated With Peripheral Vascular Interventions for the Treatment of Claudication From 2011 to 2022: A National Medicare Cohort Study
Background: Previous cross‐sectional studies have identified wide practice pattern variations in the use of peripheral vascular interventions (PVIs) for the treatment of claudication. However, there are limited data on longitudinal practice patterns. We aimed to describe the temporal trends and charges associated with PVI use for claudication over the past 12 years in the United States. Methods and Results: We conducted a retrospective analysis using 100% Medicare fee‐for‐service claims data to identify all patients who underwent a PVI for claudication between January 2011 and December 2022. We evaluated the trends in utilization and Medicare‐allowed charges of PVI according to anatomic level, procedure type, and intervention settings using generalized linear models. Multinomial logistic regressions were used to evaluate factors associated with different levels and types of PVI. We identified 599 197 PVIs performed for claudication. The proportional use of tibial PVI increased 1.0% per year, and atherectomy increased by 1.6% per year over the study period. The proportion of PVIs performed in ambulatory surgical centers/office‐based laboratories grew at 4% per year from 12.4% in 2011 to 55.7% in 2022. Total Medicare‐allowed charges increased by $11 980 035 USD/year. Multinomial logistic regression identified significant associations between race and ethnicity and treatment setting with use of both atherectomy and tibial PVI. Conclusions: The use of tibial PVI and atherectomy for the treatment of claudication has increased dramatically in in ambulatory surgical center/office‐based laboratory settings, non‐White patients, and resulting in a significant increase in health care charges. There is a critical need to improve the delivery of value‐based care for the treatment of claudication.
- 16 July 2024
- Published by Wolters Kluwer Health in Journal of the American Heart Association
- Vol. 13 (14) , e033463
Open AccessTrends and Factors Associated With Peripheral Vascular Interventions for the Treatment of Claudication From 2011 to 2022: A National Medicare Cohort Study
Background: Previous cross‐sectional studies have identified wide practice pattern variations in the use of peripheral vascular interventions (PVIs) for the treatment of claudication. However, there are limited data on longitudinal practice patterns. We aimed to describe the temporal trends and charges associated with PVI use for claudication over the past 12 years in the United States. Methods and Results: We conducted a retrospective analysis using 100% Medicare fee‐for‐service claims data to identify all patients who underwent a PVI for claudication between January 2011 and December 2022. We evaluated the trends in utilization and Medicare‐allowed charges of PVI according to anatomic level, procedure type, and intervention settings using generalized linear models. Multinomial logistic regressions were used to evaluate factors associated with different levels and types of PVI. We identified 599 197 PVIs performed for claudication. The proportional use of tibial PVI increased 1.0% per year, and atherectomy increased by 1.6% per year over the study period. The proportion of PVIs performed in ambulatory surgical centers/office‐based laboratories grew at 4% per year from 12.4% in 2011 to 55.7% in 2022. Total Medicare‐allowed charges increased by $11 980 035 USD/year. Multinomial logistic regression identified significant associations between race and ethnicity and treatment setting with use of both atherectomy and tibial PVI. Conclusions: The use of tibial PVI and atherectomy for the treatment of claudication has increased dramatically in in ambulatory surgical center/office‐based laboratory settings, non‐White patients, and resulting in a significant increase in health care charges. There is a critical need to improve the delivery of value‐based care for the treatment of claudication.
- Open Access
Investigating Unconscious Race Bias and Bias Awareness Among Vascular Surgeons
Background Implicit bias can influence behavior and decision-making. In clinical settings, implicit bias may influence treatment decisions and contribute to health disparities. Given documented Black-White disparities in vascular care, the purpose of this study was to examine the prevalence and degree of unconscious bias and awareness of bias among vascular surgeons treating peripheral artery disease (PAD). Methods The sampling frame included all vascular surgeons who participate in the Vascular Quality Initiative (VQI). Participants completed a survey which included demographic questions, the race implicit association test (IAT) to measure magnitude of unconscious bias, and six bias awareness questions to measure conscious bias. The magnitude of unconscious bias was no preference; or slight, moderate, or strong in the direction of pro-White or pro-Black. Data from participants were weighted to account for nonresponse bias and known differences in the characteristics of surgeons who chose to participate compared to the full registry. We stratified unconscious and conscious findings by physician race/ethnicity, physician sex, and years of experience. Finally, we examined the relationship between unconscious and conscious bias. Results There were 2,512 surgeons in the VQI registry, 304 of whom completed the survey, including getting IAT results. Most participants (71.6%) showed a pro-White bias with 73.0% of this group in the moderate and strong categories. While 77.5% of respondents showed conscious awareness of bias, of those whose conscious results showed lack of awareness, 67.8% had moderate or strong bias, compared to 55.7% for those with awareness. Bias magnitude varied based on physician race/ethnicity and years of experience. Women were more likely than men to report awareness of biases and potential impact of bias on decision-making. Conclusions Most people have some level of unconscious bias, developed from early life reinforcements, social stereotypes, and learned experiences. Regarding health disparities, however, these are important findings in a profession that takes care of patients with PAD due to heavy burden of comorbid conditions and high proportion of individuals from structurally vulnerable groups. Given the lack of association between unconscious and conscious awareness of biases, awareness may be an important first step in mitigation to minimize racial disparities in healthcare. What is new? This is the first study to examine unconscious and conscious bias in vascular surgeons, an important population for treating peripheral artery disease which disproportionately affects structurally vulnerable groups. We found that the majority of vascular surgeons show a pro-White bias, there is a lack of association between unconscious bias and conscious awareness of bias, and those who do not report conscious awareness of bias may also show greater magnitude of unconscious bias. What are the clinical implications? These findings offer important considerations for attentiveness to both unconscious biases and enhancement of awareness of the existence of biases among a surgery community that provides care to a diverse population of patients with PAD and disparities in health outcomes. Incorporating information on the awareness of biases and structural changes to facilitate behavior change based on these findings may be helpful within training programs for vascular surgeons. With awareness as an important first step in mitigation, cognizance of the existence of biases, as identified by this study, can aid in efforts to minimize racial disparities in health care.
Open AccessInvestigating Unconscious Race Bias and Bias Awareness Among Vascular Surgeons
Background Implicit bias can influence behavior and decision-making. In clinical settings, implicit bias may influence treatment decisions and contribute to health disparities. Given documented Black-White disparities in vascular care, the purpose of this study was to examine the prevalence and degree of unconscious bias and awareness of bias among vascular surgeons treating peripheral artery disease (PAD). Methods The sampling frame included all vascular surgeons who participate in the Vascular Quality Initiative (VQI). Participants completed a survey which included demographic questions, the race implicit association test (IAT) to measure magnitude of unconscious bias, and six bias awareness questions to measure conscious bias. The magnitude of unconscious bias was no preference; or slight, moderate, or strong in the direction of pro-White or pro-Black. Data from participants were weighted to account for nonresponse bias and known differences in the characteristics of surgeons who chose to participate compared to the full registry. We stratified unconscious and conscious findings by physician race/ethnicity, physician sex, and years of experience. Finally, we examined the relationship between unconscious and conscious bias. Results There were 2,512 surgeons in the VQI registry, 304 of whom completed the survey, including getting IAT results. Most participants (71.6%) showed a pro-White bias with 73.0% of this group in the moderate and strong categories. While 77.5% of respondents showed conscious awareness of bias, of those whose conscious results showed lack of awareness, 67.8% had moderate or strong bias, compared to 55.7% for those with awareness. Bias magnitude varied based on physician race/ethnicity and years of experience. Women were more likely than men to report awareness of biases and potential impact of bias on decision-making. Conclusions Most people have some level of unconscious bias, developed from early life reinforcements, social stereotypes, and learned experiences. Regarding health disparities, however, these are important findings in a profession that takes care of patients with PAD due to heavy burden of comorbid conditions and high proportion of individuals from structurally vulnerable groups. Given the lack of association between unconscious and conscious awareness of biases, awareness may be an important first step in mitigation to minimize racial disparities in healthcare. What is new? This is the first study to examine unconscious and conscious bias in vascular surgeons, an important population for treating peripheral artery disease which disproportionately affects structurally vulnerable groups. We found that the majority of vascular surgeons show a pro-White bias, there is a lack of association between unconscious bias and conscious awareness of bias, and those who do not report conscious awareness of bias may also show greater magnitude of unconscious bias. What are the clinical implications? These findings offer important considerations for attentiveness to both unconscious biases and enhancement of awareness of the existence of biases among a surgery community that provides care to a diverse population of patients with PAD and disparities in health outcomes. Incorporating information on the awareness of biases and structural changes to facilitate behavior change based on these findings may be helpful within training programs for vascular surgeons. With awareness as an important first step in mitigation, cognizance of the existence of biases, as identified by this study, can aid in efforts to minimize racial disparities in health care.
- Open Access
- 11 April 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Development and Validation of a Routine Electronic Health Record-Based Delirium Prediction Model for Surgical Patients Without Dementia: Retrospective Case-Control Study (Preprint)
BACKGROUND Postoperative delirium (POD) is a common complication after major surgery and is associated with poor outcomes in older adults. Early identification of patients at high risk of POD can enable targeted prevention efforts. However, existing POD prediction models require inpatient data collected during the hospital stay, which delays predictions and limits scalability. OBJECTIVE This study aimed to develop and externally validate a machine learning-based prediction model for POD using routine electronic health record (EHR) data. METHODS We identified all surgical encounters from 2014 to 2021 for patients aged 50 years and older who underwent an operation requiring general anesthesia, with a length of stay of at least 1 day at 3 Indiana hospitals. Patients with preexisting dementia or mild cognitive impairment were excluded. POD was identified using Confusion Assessment Method records and delirium International Classification of Diseases (ICD) codes. Controls without delirium or nurse-documented confusion were matched to cases by age, sex, race, and year of admission. We trained logistic regression, random forest, extreme gradient boosting (XGB), and neural network models to predict POD using 143 features derived from routine EHR data available at the time of hospital admission. Separate models were developed for each hospital using surveillance periods of 3 months, 6 months, and 1 year before admission. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC). Each model was internally validated using holdout data and externally validated using data from the other 2 hospitals. Calibration was assessed using calibration curves. RESULTS The study cohort included 7167 delirium cases and 7167 matched controls. XGB outperformed all other classifiers. AUROCs were highest for XGB models trained on 12 months of preadmission data. The best-performing XGB model achieved a mean AUROC of 0.79 (SD 0.01) on the holdout set, which decreased to 0.69-0.74 (SD 0.02) when externally validated on data from other hospitals. CONCLUSIONS Our routine EHR-based POD prediction models demonstrated good predictive ability using a limited set of preadmission and surgical variables, though their generalizability was limited. The proposed models could be used as a scalable, automated screening tool to identify patients at high risk of POD at the time of hospital admission.
- 11 April 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessDevelopment and Validation of a Routine Electronic Health Record-Based Delirium Prediction Model for Surgical Patients Without Dementia: Retrospective Case-Control Study (Preprint)
BACKGROUND Postoperative delirium (POD) is a common complication after major surgery and is associated with poor outcomes in older adults. Early identification of patients at high risk of POD can enable targeted prevention efforts. However, existing POD prediction models require inpatient data collected during the hospital stay, which delays predictions and limits scalability. OBJECTIVE This study aimed to develop and externally validate a machine learning-based prediction model for POD using routine electronic health record (EHR) data. METHODS We identified all surgical encounters from 2014 to 2021 for patients aged 50 years and older who underwent an operation requiring general anesthesia, with a length of stay of at least 1 day at 3 Indiana hospitals. Patients with preexisting dementia or mild cognitive impairment were excluded. POD was identified using Confusion Assessment Method records and delirium International Classification of Diseases (ICD) codes. Controls without delirium or nurse-documented confusion were matched to cases by age, sex, race, and year of admission. We trained logistic regression, random forest, extreme gradient boosting (XGB), and neural network models to predict POD using 143 features derived from routine EHR data available at the time of hospital admission. Separate models were developed for each hospital using surveillance periods of 3 months, 6 months, and 1 year before admission. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC). Each model was internally validated using holdout data and externally validated using data from the other 2 hospitals. Calibration was assessed using calibration curves. RESULTS The study cohort included 7167 delirium cases and 7167 matched controls. XGB outperformed all other classifiers. AUROCs were highest for XGB models trained on 12 months of preadmission data. The best-performing XGB model achieved a mean AUROC of 0.79 (SD 0.01) on the holdout set, which decreased to 0.69-0.74 (SD 0.02) when externally validated on data from other hospitals. CONCLUSIONS Our routine EHR-based POD prediction models demonstrated good predictive ability using a limited set of preadmission and surgical variables, though their generalizability was limited. The proposed models could be used as a scalable, automated screening tool to identify patients at high risk of POD at the time of hospital admission.
- Open Access
Implicit Bias And Unintentional Harm In Vascular Care: The Case For Intervention
Introduction: Implicit (or unconscious) bias may influence physician treatment decisions and contribute to Black-White health disparities. While implicit bias has been linked with low quality care via clinical vignettes, some worry that these studies are not representative of the ‘real world.’ There is limited data that has attempted to link implicit bias with actual care delivery and outcomes. We sought to understand if implicit bias is associated with potentially harmful surgical treatment selection in a cohort of patients with peripheral artery disease (PAD)-related claudication undergoing below-knee lower extremity revascularization as captured in a ‘real world’ procedural registry. Methods: We invited vascular specialists from the Vascular Quality Initiative (VQI) to take the race Implicit Association Test (IAT). The IAT asks participants to associate images of Black and White Americans with either positive or negative attributes. Based on reaction time differences across sequential tests, participants were grouped into race-based implicit bias categories: pro-White bias, no bias, or pro-Black bias. Our provider-level implicit bias results were linked to patient-level registry data of peripheral revascularization procedures performed for claudication. We measured the adjusted odds of performance of below knee procedures by specialist implicit bias and patient race via mixed effects logistic regression models. We assessed implicit bias as a moderator of the association of below-knee procedures and patient race with 1-yr amputation. Results: 218 vascular specialists in the United States completed the IAT and 157 (72%) had a pro-White bias. Black patients treated by a physician with pro-White bias had a 74% increase in the odds of receiving a below-knee procedure compared to the total sample (aOR: 1.74, 95% CI: 1.33-2.15). When treated by a specialist with pro-White bias, Black patients had 3 times the odds of 1-yr amputation – regardless of anatomic location treated – compared to White patients (aOR: 3.04, 95% CI: 1.68-5.51). Conversely, Black patients treated by a specialist with no bias had similar odds of a below-knee procedure (aOR: 0.99, 95% CI: 0.67-1.30) and 1-year amputation (aOR: 1.31, 95% CI: 0.35-4.96) as the full patient sample. Conclusions: Implicit bias is associated with potentially harmful below-knee procedures for Black patients and contributes to Black-White outcome disparities in the United States. These results suggest the need for system-level interventions that transparently identify and warn of procedures not aligned with best practices to reduce the negative effect of implicit bias. Clinical Perspective: What is new?: Pro-White bias is associated with low-value care for Black patients. Black patients treated by physicians with pro-White implicit preferences also had worse 1-year amputation rates. We found little race-based variation in treatment decisions for below-knee revascularization procedures among physicians with no implicit bias as measured by the IAT. What are the clinical implications?: Physicians who treat vascular diseases consider their own practices and where they may be falling short of standards, particularly for their Black patients. Health care leaders must begin to evaluate how and why their own health care systems make it possible for practitioners’ unconscious biases to negatively impact care. Policy changes that enhance payment for evidence-based procedures and reduce payment for potentially harmful procedures should be considered.
Open AccessImplicit Bias And Unintentional Harm In Vascular Care: The Case For Intervention
Introduction: Implicit (or unconscious) bias may influence physician treatment decisions and contribute to Black-White health disparities. While implicit bias has been linked with low quality care via clinical vignettes, some worry that these studies are not representative of the ‘real world.’ There is limited data that has attempted to link implicit bias with actual care delivery and outcomes. We sought to understand if implicit bias is associated with potentially harmful surgical treatment selection in a cohort of patients with peripheral artery disease (PAD)-related claudication undergoing below-knee lower extremity revascularization as captured in a ‘real world’ procedural registry. Methods: We invited vascular specialists from the Vascular Quality Initiative (VQI) to take the race Implicit Association Test (IAT). The IAT asks participants to associate images of Black and White Americans with either positive or negative attributes. Based on reaction time differences across sequential tests, participants were grouped into race-based implicit bias categories: pro-White bias, no bias, or pro-Black bias. Our provider-level implicit bias results were linked to patient-level registry data of peripheral revascularization procedures performed for claudication. We measured the adjusted odds of performance of below knee procedures by specialist implicit bias and patient race via mixed effects logistic regression models. We assessed implicit bias as a moderator of the association of below-knee procedures and patient race with 1-yr amputation. Results: 218 vascular specialists in the United States completed the IAT and 157 (72%) had a pro-White bias. Black patients treated by a physician with pro-White bias had a 74% increase in the odds of receiving a below-knee procedure compared to the total sample (aOR: 1.74, 95% CI: 1.33-2.15). When treated by a specialist with pro-White bias, Black patients had 3 times the odds of 1-yr amputation – regardless of anatomic location treated – compared to White patients (aOR: 3.04, 95% CI: 1.68-5.51). Conversely, Black patients treated by a specialist with no bias had similar odds of a below-knee procedure (aOR: 0.99, 95% CI: 0.67-1.30) and 1-year amputation (aOR: 1.31, 95% CI: 0.35-4.96) as the full patient sample. Conclusions: Implicit bias is associated with potentially harmful below-knee procedures for Black patients and contributes to Black-White outcome disparities in the United States. These results suggest the need for system-level interventions that transparently identify and warn of procedures not aligned with best practices to reduce the negative effect of implicit bias. Clinical Perspective: What is new?: Pro-White bias is associated with low-value care for Black patients. Black patients treated by physicians with pro-White implicit preferences also had worse 1-year amputation rates. We found little race-based variation in treatment decisions for below-knee revascularization procedures among physicians with no implicit bias as measured by the IAT. What are the clinical implications?: Physicians who treat vascular diseases consider their own practices and where they may be falling short of standards, particularly for their Black patients. Health care leaders must begin to evaluate how and why their own health care systems make it possible for practitioners’ unconscious biases to negatively impact care. Policy changes that enhance payment for evidence-based procedures and reduce payment for potentially harmful procedures should be considered.