1
|
Perets O, Stagno E, Yehuda EB, McNichol M, Anthony Celi L, Rappoport N, Dorotic M. Inherent Bias in Electronic Health Records: A Scoping Review of Sources of Bias. MEDRXIV : THE PREPRINT SERVER FOR HEALTH SCIENCES 2024:2024.04.09.24305594. [PMID: 38680842 PMCID: PMC11046491 DOI: 10.1101/2024.04.09.24305594] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Indexed: 05/01/2024]
Abstract
Objectives 1.1Biases inherent in electronic health records (EHRs), and therefore in medical artificial intelligence (AI) models may significantly exacerbate health inequities and challenge the adoption of ethical and responsible AI in healthcare. Biases arise from multiple sources, some of which are not as documented in the literature. Biases are encoded in how the data has been collected and labeled, by implicit and unconscious biases of clinicians, or by the tools used for data processing. These biases and their encoding in healthcare records undermine the reliability of such data and bias clinical judgments and medical outcomes. Moreover, when healthcare records are used to build data-driven solutions, the biases are further exacerbated, resulting in systems that perpetuate biases and induce healthcare disparities. This literature scoping review aims to categorize the main sources of biases inherent in EHRs. Methods 1.2We queried PubMed and Web of Science on January 19th, 2023, for peer-reviewed sources in English, published between 2016 and 2023, using the PRISMA approach to stepwise scoping of the literature. To select the papers that empirically analyze bias in EHR, from the initial yield of 430 papers, 27 duplicates were removed, and 403 studies were screened for eligibility. 196 articles were removed after the title and abstract screening, and 96 articles were excluded after the full-text review resulting in a final selection of 116 articles. Results 1.3Systematic categorizations of diverse sources of bias are scarce in the literature, while the effects of separate studies are often convoluted and methodologically contestable. Our categorization of published empirical evidence identified the six main sources of bias: a) bias arising from past clinical trials; b) data-related biases arising from missing, incomplete information or poor labeling of data; human-related bias induced by c) implicit clinician bias, d) referral and admission bias; e) diagnosis or risk disparities bias and finally, (f) biases in machinery and algorithms. Conclusions 1.4Machine learning and data-driven solutions can potentially transform healthcare delivery, but not without limitations. The core inputs in the systems (data and human factors) currently contain several sources of bias that are poorly documented and analyzed for remedies. The current evidence heavily focuses on data-related biases, while other sources are less often analyzed or anecdotal. However, these different sources of biases add to one another exponentially. Therefore, to understand the issues holistically we need to explore these diverse sources of bias. While racial biases in EHR have been often documented, other sources of biases have been less frequently investigated and documented (e.g. gender-related biases, sexual orientation discrimination, socially induced biases, and implicit, often unconscious, human-related cognitive biases). Moreover, some existing studies lack causal evidence, illustrating the different prevalences of disease across groups, which does not per se prove the causality. Our review shows that data-, human- and machine biases are prevalent in healthcare and they significantly impact healthcare outcomes and judgments and exacerbate disparities and differential treatment. Understanding how diverse biases affect AI systems and recommendations is critical. We suggest that researchers and medical personnel should develop safeguards and adopt data-driven solutions with a "bias-in-mind" approach. More empirical evidence is needed to tease out the effects of different sources of bias on health outcomes.
Collapse
|
2
|
Morris SM, Gupta A, Kim S, Foraker RE, Gutmann DH, Payne PRO. Predictive Modeling for Clinical Features Associated With Neurofibromatosis Type 1. Neurol Clin Pract 2022; 11:497-505. [PMID: 34987881 PMCID: PMC8723929 DOI: 10.1212/cpj.0000000000001089] [Citation(s) in RCA: 4] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/06/2020] [Accepted: 02/25/2021] [Indexed: 12/23/2022]
Abstract
Objective To perform a longitudinal analysis of clinical features associated with
neurofibromatosis type 1 (NF1) based on demographic and clinical
characteristics and to apply a machine learning strategy to determine
feasibility of developing exploratory predictive models of optic pathway
glioma (OPG) and attention-deficit/hyperactivity disorder (ADHD) in a
pediatric NF1 cohort. Methods Using NF1 as a model system, we perform retrospective data analyses using a
manually curated NF1 clinical registry and electronic health record (EHR)
information and develop machine learning models. Data for 798 individuals
were available, with 578 comprising the pediatric cohort used for
analysis. Results Males and females were evenly represented in the cohort. White children were
more likely to develop OPG (odds ratio [OR]: 2.11, 95% confidence interval
[CI]: 1.11–4.00, p = 0.02) relative to their
non-White peers. Median age at diagnosis of OPG was 6.5 years
(1.7–17.0), irrespective of sex. Males were more likely than females
to have a diagnosis of ADHD (OR: 1.90, 95% CI: 1.33–2.70,
p < 0.001), and earlier diagnosis in males
relative to females was observed. The gradient boosting classification model
predicted diagnosis of ADHD with an area under the receiver operator
characteristic (AUROC) of 0.74 and predicted diagnosis of OPG with an AUROC
of 0.82. Conclusions Using readily available clinical and EHR data, we successfully recapitulated
several important and clinically relevant patterns in NF1 semiology
specifically based on demographic and clinical characteristics. Naive
machine learning techniques can be potentially used to develop and validate
predictive phenotype complexes applicable to risk stratification and disease
management in NF1.
Collapse
Affiliation(s)
- Stephanie M Morris
- Department of Neurology (DHG), Washington University, St. Louis, MO; and Institute for Informatics (SMM, AG, SK, REF, PROP), Washington University, St. Louis, MO
| | - Aditi Gupta
- Department of Neurology (DHG), Washington University, St. Louis, MO; and Institute for Informatics (SMM, AG, SK, REF, PROP), Washington University, St. Louis, MO
| | - Seunghwan Kim
- Department of Neurology (DHG), Washington University, St. Louis, MO; and Institute for Informatics (SMM, AG, SK, REF, PROP), Washington University, St. Louis, MO
| | - Randi E Foraker
- Department of Neurology (DHG), Washington University, St. Louis, MO; and Institute for Informatics (SMM, AG, SK, REF, PROP), Washington University, St. Louis, MO
| | - David H Gutmann
- Department of Neurology (DHG), Washington University, St. Louis, MO; and Institute for Informatics (SMM, AG, SK, REF, PROP), Washington University, St. Louis, MO
| | - Philip R O Payne
- Department of Neurology (DHG), Washington University, St. Louis, MO; and Institute for Informatics (SMM, AG, SK, REF, PROP), Washington University, St. Louis, MO
| |
Collapse
|
3
|
Lechtholz-Zey E, Bonney PA, Cardinal T, Mendoza J, Strickland BA, Pangal DJ, Giannotta S, Durham S, Zada G. Systematic Review of Racial, Socioeconomic, and Insurance Status Disparities in the Treatment of Pediatric Neurosurgical Diseases in the United States. World Neurosurg 2021; 158:65-83. [PMID: 34718199 DOI: 10.1016/j.wneu.2021.10.150] [Citation(s) in RCA: 14] [Impact Index Per Article: 4.7] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/31/2021] [Revised: 10/21/2021] [Accepted: 10/22/2021] [Indexed: 12/18/2022]
Abstract
BACKGROUND Increasing light is being shed on how race, insurance, and socioeconomic status (SES) may be related to outcomes from disease in the United States. To better understand the impact of these health care disparities in pediatric neurosurgery, we performed a systematic review of the literature. METHODS We conducted a systematic review using PRISMA guidelines and MeSH terms involving neurosurgical conditions and racial, ethnic, and SES disparities. Three independent reviewers screened articles and analyzed texts selected for full analysis. RESULTS Thirty-eight studies were included in the final analysis, of which all but 2 were retrospective database reviews. Thirty-four studies analyzed race, 22 analyzed insurance status, and 13 analyzed SES/income. Overall, nonwhite patients, patients with public insurance, and patients from lower SES were shown to have reduced access to treatment and greater rates of adverse outcomes. Nonwhite patients were more likely to present at an older age with more severe disease, less likely to undergo surgery at a high-volume surgical center, and more likely to experience postoperative morbidity and mortality. Underinsured and publicly insured patients were more likely to experience delay in surgical referral, less likely to undergo surgical treatment, and more likely to experience inpatient mortality. CONCLUSIONS Health care disparities are present within multiple populations of patients receiving pediatric neurosurgical care. This review highlights the need for continued investigation into identifying and addressing health care disparities in pediatric neurosurgery patients.
Collapse
Affiliation(s)
- Elizabeth Lechtholz-Zey
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| | - Phillip A Bonney
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| | - Tyler Cardinal
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA.
| | - Jesse Mendoza
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| | - Ben A Strickland
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| | - Dhiraj J Pangal
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| | - Steven Giannotta
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| | - Susan Durham
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA; Division of Neurosurgery, Children's Hospital Los Angeles, Los Angeles, California, USA
| | - Gabriel Zada
- Department of Neurosurgery, Keck School of Medicine of University of Southern California, Los Angeles, California, USA
| |
Collapse
|
4
|
Barber Doucet H, Ward VL, Johnson TJ, Lee LK. Implicit Bias and Caring for Diverse Populations: Pediatric Trainee Attitudes and Gaps in Training. Clin Pediatr (Phila) 2021; 60:408-417. [PMID: 34308661 DOI: 10.1177/00099228211035225] [Citation(s) in RCA: 6] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/15/2022]
Abstract
The objective of this study was to determine the attitudes, skill level, and preferred educational interventions of pediatric residents related to implicit bias and caring for diverse patient populations. A cross-sectional survey of pediatric residents at a single, large urban residency program was utilized. Surveys were completed by 88 (55%) residents who were 69% female and 35% non-White or mixed race. Almost all residents felt that it was very or extremely important to receive training on health disparities, diverse patient populations, and implicit bias. Self-assessment of skill level revealed that residents felt confident in areas often covered by cultural competency curricula, such as interpreter use, but were less confident in other areas. The top 3 areas identified for further training included implicit bias, working with transgender and gender nonconforming patients, and weight bias. For the majority of diversity and bias-related skills, prior training was significantly correlated with higher skill level (P < .05).
Collapse
Affiliation(s)
| | | | | | - Lois K Lee
- Boston Children's Hospital, Boston, MA, USA
| |
Collapse
|
5
|
Lopez-Rivera V, Dono A, Abdelkhaleq R, Sheth SA, Chen PR, Chandra A, Ballester LY, Esquenazi Y. Treatment trends and overall survival in patients with grade II/III ependymoma: The role of tumor grade and location. Clin Neurol Neurosurg 2020; 199:106282. [PMID: 33045626 DOI: 10.1016/j.clineuro.2020.106282] [Citation(s) in RCA: 8] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/06/2020] [Revised: 09/08/2020] [Accepted: 10/04/2020] [Indexed: 10/23/2022]
Abstract
BACKGROUND Treatment of ependymoma (EPN) is guided by associated tumor features, such as grade and location. However, the relationship between these features with treatments and overall survival in EPN patients remains uncharacterized. Here, we describe the change over time in treatment strategies and identify tumor characteristics that influence treatment and survival in EPN. METHODS AND MATERIALS Using the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) 18 Registries (1973-2016) database, we identified patients with EPN microscopically confirmed to be grade II (EPN-GII) or III (EPN-GIII) tumors between 2004-2016. Overall survival (OS) was analyzed using Kaplan-Meier survival estimates and multivariable Cox proportional hazard models. A sub-analysis was performed by tumor location (supratentorial, posterior fossa, and spine). Change over time in rates of gross total resection (GTR), radiotherapy (RT), and chemotherapy (CS) were analyzed using linear regression, and predictors of treatment were identified using multivariable logistic regression models. RESULTS Between 2004-2016, 1,671 patients were diagnosed with EPN, of which 1,234 (74 %) were EPN-GII and 437 (26 %) EPN-GIII. Over the study period, EPN-GII patients underwent a less aggressive treatment (48 % vs 27 %, GTR; 60 % vs 30 %, RT; 22 % vs 2%, CS; 2004 vs 2016; p < 0.01 for all). Age, tumor size, location, and grade were positive predictors of undergoing treatment. Univariate analysis revealed that tumor grade and location were significantly associated with OS (p < 0.0001 for both). In multivariable Cox regression, tumor grade was an independent predictor of OS among patients in the cohort (grade III, HR 3.89 [2.84-5.33]; p < 0.0001), with this finding remaining significant across all tumor locations. CONCLUSIONS In EPN, tumor grade and location are predictors of treatment and overall survival. These findings support the importance of histologic WHO grade and location in the decision-making for treatment and their role in individualizing treatment for different patient populations.
Collapse
Affiliation(s)
- Victor Lopez-Rivera
- Department of Neurology, The University of Texas Health Science Center at Houston, Houston, TX, USA
| | - Antonio Dono
- Vivian L. Smith Department of Neurosurgery, The University of Texas Health Science Center at Houston, Houston, TX, USA; Department of Pathology and Laboratory Medicine, The University of Texas Health Science Center at Houston, Houston, TX, USA
| | - Rania Abdelkhaleq
- Department of Neurology, The University of Texas Health Science Center at Houston, Houston, TX, USA
| | - Sunil A Sheth
- Department of Neurology, The University of Texas Health Science Center at Houston, Houston, TX, USA; Memorial Hermann Hospital-Texas Medical Center, Houston, TX, USA
| | - Peng R Chen
- Vivian L. Smith Department of Neurosurgery, The University of Texas Health Science Center at Houston, Houston, TX, USA; Memorial Hermann Hospital-Texas Medical Center, Houston, TX, USA
| | - Ankush Chandra
- Vivian L. Smith Department of Neurosurgery, The University of Texas Health Science Center at Houston, Houston, TX, USA; Memorial Hermann Hospital-Texas Medical Center, Houston, TX, USA
| | - Leomar Y Ballester
- Vivian L. Smith Department of Neurosurgery, The University of Texas Health Science Center at Houston, Houston, TX, USA; Department of Pathology and Laboratory Medicine, The University of Texas Health Science Center at Houston, Houston, TX, USA; Memorial Hermann Hospital-Texas Medical Center, Houston, TX, USA
| | - Yoshua Esquenazi
- Vivian L. Smith Department of Neurosurgery, The University of Texas Health Science Center at Houston, Houston, TX, USA; Memorial Hermann Hospital-Texas Medical Center, Houston, TX, USA; Center for Precision Health, School of Biomedical Informatics, The University of Texas Health Science Center at Houston, Houston, TX, USA.
| |
Collapse
|
6
|
Racial/ethnic disparities in penile squamous cell carcinoma incidences, clinical characteristics, and outcomes: A population-based study, 2004-2016. Urol Oncol 2020; 38:688.e11-688.e19. [PMID: 32340796 DOI: 10.1016/j.urolonc.2020.03.003] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/20/2019] [Revised: 02/23/2020] [Accepted: 03/05/2020] [Indexed: 12/11/2022]
Abstract
PURPOSE This study assessed the impact of race/ethnicity on penile squamous cell carcinoma (PSCC) incidence rates, clinical characteristics, and outcomes. MATERIALS AND METHODS Surveillance, Epidemiology and End Results data from 2004 to 2016 was used for this study. We evaluated racial/ethnic differences in clinical characteristics using χ2 tests. Overall survival (OS) and PSCC-specific survival (PSCC-SS) were estimated using the Kaplan-Meier method, and differences were determined using the log-rank test. Cox regression models were performed to assess independent predictors for PSCC patient survival. RESULTS A total of 2,720 PSCC patients were included for incidence analysis, and 2,438 patients were identified for the χ2 testing and survival analyses.The overall incidence of PSCC during 2004 to 2016 was 0.30 per 100,000. Only non-Hispanic white (NHW) patients had a statistically significant increase in age-adjusted incidence rates (annual percent change = 2.26, 95% confidence interval [CI]: 0.78-3.76; P = 0.01). In univariate analysis, race/ethnicity was an independent prognostic factor for OS and PSCC-SS. After adjusting for age, marital status, income, grade, TNM (tumor, node, metastasis) stage, and treatment strategies, non-Hispanic black patients still had a statistically significant hazard ratio of 1.35 (95% CI: 1.08-1.68; P = 0.007) for OS, and a hazard ratio of 1.36 (95% CI: 1.01-1.82; P = 0.045) for PSCC-SS compared to NHW. CONCLUSION NHW patients had a statistically significant increase in age-adjusted incidence rate during the period 2004 to 2016. Race/ethnicity is an independent prognostic factor for OS and PSCC-SS, and non-Hispanic black were proven to have unfavorable OS and PSCC-SS compared with NHW.
Collapse
|