Predicting Race And Ethnicity To Ensure Equitable Algorithms For Health Care Decision Making

Cabreros, Irineo; Agniel, Denis; Martino, Steven C.; Damberg, Cheryl L.; Elliott, Marc N.

doi:10.1377/hlthaff.2022.00095

Cited by 11 publications

(12 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The extent of this missingness is complex and varies across sources, ranging from 30-70%, 23 but can largely be attributed to issues in data transmission and collection rather than simply patient abstention. 42,43…”

Section: Discussionmentioning

confidence: 99%

Genetic Ancestry and Somatic Mutations in Lung Adenocarcinoma: Insights from Real-World Clinico-Genomic Data

Rhead,

Pouliot,

Guinney

et al. 2024

Preprint

View full text Add to dashboard Cite

Background: Lung cancer presents a significant global health challenge, with disparities in incidence and outcomes across races and ethnicities. These disparities underscore the need to explore the molecular landscapes of lung cancer in relation to ancestry. Here, we leverage data from a real-world clinico-genomic database to discover associations between molecular profiles and genetic ancestry or race/ethnicity categories. Methods: We utilized data from a cohort of 13,196 primarily late-stage non-small cell lung adenocarcinoma (LUAD) patients, sequenced with the Tempus xT NGS 648-gene panel, of which normal tissue was also sequenced for 6,520 cases. Genetic ancestry proportions were estimated using ancestry informative markers. Race and ethnicity categories were imputed using an ancestry-backed method, resulting in the assignment of 568 Hispanic/Latino, 892 non-Hispanic (NH) Asian, 1,581 NH Black, and 10,063 NH White individuals. Multiple imputation addressed missing data on smoking status. Logistic regression models assessed associations between ancestry proportions and somatic variants in 23 LUAD-related genes, adjusting for a false discovery rate of 5%. Analyzed mutations included copy number alterations, gene fusions, protein-altering SNVs and indels, and actionable or predicted driver mutations. Results: Our analysis confirmed previously reported associations, such as a positive correlation between East Asian (EAS) ancestry and EGFR (OR per doubling ancestry=1.1) and a negative correlation with KRAS driver mutations (OR=0.96), while European ancestry exhibited the opposite relationship (OR=0.93 and 1.08, correspondingly; all p<0.0001). We also verified a positive association with EGFR driver mutations (OR=2) and a negative one with KRAS (OR=0.46; p<0.001) among Hispanic/Latino patients and American Indigenous (AMR) genetic ancestry (OR=1.03 and 0.97, correspondingly; p<0.05). Novel associations were identified between African (AFR) and South Asian (SAS) ancestries and LUAD genes. Some associations are explained by differences in smoking status (e.g., ATM and ALK fusions), while others persist even after adjusting for smoking (e.g., EGFR, KRAS, and CDKN2A copy-number alterations). Notably, we identified a positive association between EAS ancestry and the imputed NH Asian category with driver mutations in CTNNB1 (OR=1.05 and 2.2, respectively; p<0.01), independent of smoking. These mutations are rare in NH White patients (2.4%) but are more prevalent in never-smoker NH Asian patients with predominant EAS ancestry (8.5%). Conclusion: This study underscores the value of clinico-genomic databases in revealing associations between LUAD mutational profiles and genetic ancestry, shedding light on lung cancer disparities. Identification of a previously unappreciated association between EAS with CTNNB1, a potential biomarker for spindle assembly checkpoint kinase (TTK) inhibitors effectiveness and prognosis in LUAD, emphasizes the value of studying diverse populations in cancer research, paving the way for more equitable lung cancer treatments.

show abstract

Section: Discussionmentioning

confidence: 99%

Genetic Ancestry and Somatic Mutations in Lung Adenocarcinoma: Insights from Real-World Clinico-Genomic Data

Rhead,

Pouliot,

Guinney

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…These issues are not unsurmountable, however. There are algorithms that can be used to impute missing race and ethnicity data with high sensitivity, and geospatial-based covariates, such as the Child Opportunity Index, can be estimated from zip code, census block, and census block group–level data …”

Section: Equity and Social Determinants Of Healthmentioning

confidence: 99%

“…There are algorithms that can be used to impute missing race and ethnicity data with high sensitivity, and geospatial-based covariates, such as the Child Opportunity Index, can be estimated from zip code, census block, and census block group-level data. 9,10 The Future of BMI-Associated Health Care Expenditures To date, treatments for unhealthy BMI classes have primarily included behavioral interventions, and thus, the costs captured by studies of BMI class-related expenditures likely relate to BMI class-related comorbidities and behavioral interventions. Related to overweight and obesity, the therapeutic arsenal for treating childhood obesity may soon change, with a growing movement to use bariatric surgery and pharmacotherapy in child and adolescent populations, as is now recommended by the American Academy of Pediatrics.…”

Section: Equity and Social Determinants Of Healthmentioning

confidence: 99%

Current and Future Challenges Regarding Estimating Costs to Determine the Value of Interventions to Manage Unhealthy Weight

2023

View full text Add to dashboard Cite

Large systems-level changes are needed to address the high prevalence of childhood obesity in the US. In a world with many competing priorities, stakeholders need to understand the value of investing in childhood obesity policy changes to prioritize funding clinical and policy interventions. The study by Kumar et al 1 in this issue of JAMA Pediatrics aims to estimate the association between children's body mass index (BMI) class and medical expenditures in the US. Using electronic medical record (EMR) and claims data for privately insured youth, this cross-sectional study found that, relative to healthy weight, having underweight, overweight, and moderate and severe obesity were associated with higher total health care expenditures among those aged 2 to 19 years in 2018.The study by Kumar et al 1 contributes to other literature that uses nationally representative data to estimate health care costs attributable to underweight and/or excess weight. [2][3][4][5] One strength of many of these studies is the holistic way in which they focus on both underweight and overweight. The primary contribution of the present study is the use of measured BMI derived from EMR records; most national weight class-related claims-based analyses rely on International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes, which have low specificity for identifying obesity, 6 or self-reported BMI, which is often biased. 7 Based on their findings, Kumar et al 1 suggest that there may be some economic value of interventions or treatments aimed at reducing BMI-associated health risks. Below, we outline 3 themes that should be considered to fully understand the implications of the findings of this and similar studies of BMI class-related expenditures.

show abstract

“…If the RTI race code disproportionately misclassifies beneficiaries from certain race and ethnicity groups, there is risk of additional algorithmic biases being introduced when reporting any outcomes of interest stratified by predicted race/ethnicity. 14,15 Instead, the use of self-reported race and ethnicity should be considered the gold-standard for identifying beneficiary race and ethnicity. 16,17 Recent studies have assessed the validity of EDB and RTI race/ ethnicity coding in Medicare data using 2015 home health 18 and 2018-2020 nursing home 19 assessments; however, the populations studied are specific to beneficiaries receiving postacute care, and it is unclear how consistently beneficiary race and ethnicity is collected during assessment visits compared to self-report.…”

Section: Introductionmentioning

confidence: 99%

Assessing the validity of race and ethnicity coding in administrative Medicare data for reporting outcomes among Medicare advantage beneficiaries from 2015 to 2017

Huang

Meyers

2023

Health Services Research

View full text Add to dashboard Cite

ObjectiveTo assess the validity of race/ethnicity coding in Medicare data and whether misclassification errors lead to biased outcome reporting by race/ethnicity among Medicare Advantage beneficiaries.Data Sources and Study SettingIn this national study of Medicare Advantage beneficiaries, we analyzed individual‐level data from the Health Outcomes Survey (HOS) and the Consumer Assessment of Healthcare Providers and Systems (CAHPS), race/ethnicity codes from the Medicare Master Beneficiary Summary File (MBSF), and outcomes from the Medicare Provider Analysis and Review (MedPAR) files from 2015 to 2017.Study DesignWe used self‐reported beneficiary race/ethnicity to validate the Medicare Enrollment Database (EDB) and Research Triangle Institute (RTI) race/ethnicity codes. We measured the sensitivity, specificity, and positive and negative predictive values of the Medicare EDB and RTI codes compared to self‐report. For outcomes, we compared annualized hospital admission, 30‐day, and 90‐day readmission rates.Data Collection/Extraction MethodsData for Medicare Advantage beneficiaries who completed either the HOS or CAHPS survey were linked to MBSF and MedPAR files. Validity was assessed for both self‐reported multiracial and single‐race beneficiaries.Principal FindingsFor beneficiaries enrolled in Medicare Advantage, the EDB and RTI race/ethnicity codes have high validity for identifying non‐Hispanic White or Black beneficiaries, but lower sensitivity for beneficiaries self‐reported Hispanic any race (EDB: 28.3%, RTI: 85.9%) or non‐Hispanic Asian American or Native Hawaiian Pacific Islander (EDB: 56.1%, RTI: 72.1%). Only 8.7% of beneficiaries self‐reported non‐Hispanic American Indian Alaska Native are correctly identified by either Medicare code, resulting in underreported annualized hospitalization rates (EDB: 31.5%, RTI: 31.6% vs. self‐report: 34.6%). We find variation in 30‐day readmission rates for Hispanic beneficiaries across race categories, which is not measured by Medicare race/ethnicity coding.ConclusionsCurrent Medicare race/ethnicity codes misclassify and bias outcomes for non‐Hispanic AIAN beneficiaries, who are more likely to select multiple racial identities. Revisions to race/ethnicity categories are needed to better represent multiracial/ethnic identities among Medicare Advantage beneficiaries.

show abstract

Predicting Race And Ethnicity To Ensure Equitable Algorithms For Health Care Decision Making

Cited by 11 publications

References 34 publications

Genetic Ancestry and Somatic Mutations in Lung Adenocarcinoma: Insights from Real-World Clinico-Genomic Data

Genetic Ancestry and Somatic Mutations in Lung Adenocarcinoma: Insights from Real-World Clinico-Genomic Data

Current and Future Challenges Regarding Estimating Costs to Determine the Value of Interventions to Manage Unhealthy Weight

Assessing the validity of race and ethnicity coding in administrative Medicare data for reporting outcomes among Medicare advantage beneficiaries from 2015 to 2017

Contact Info

Product

Resources

About