Racial and Ethnic Bias in Risk Prediction Models for Colorectal Cancer Recurrence When Race and Ethnicity Are Omitted as Predictors.

Importance: Including race and ethnicity as a predictor in clinical risk prediction algorithms has received increased scrutiny, but there continues to be a lack of empirical studies addressing whether simply omitting race and ethnicity from the algorithms will ultimately affect decision-making for patients of minoritized racial and ethnic groups. Objective: To examine whether including race and ethnicity as a predictor in a colorectal cancer recurrence risk algorithm is associated with racial bias, defined as racial and ethnic differences in model accuracy that could potentially lead to unequal treatment. Design, Setting, and Participants: This retrospective prognostic study was conducted using data from a large integrated health care system in Southern California for patients with colorectal cancer who received primary treatment between 2008 and 2013 and follow-up until December 31, 2018. Data were analyzed from January 2021 to June 2022. Main Outcomes and Measures: Four Cox proportional hazards regression prediction models were fitted to predict time from surveillance start to cancer recurrence: (1) a race-neutral model that explicitly excluded race and ethnicity as a predictor, (2) a race-sensitive model that included race and ethnicity, (3) a model with 2-way interactions between clinical predictors and race and ethnicity, and (4) separate models by race and ethnicity. Algorithmic fairness was assessed using model calibration, discriminative ability, false-positive and false-negative rates, positive predictive value (PPV), and negative predictive value (NPV). Results: The study cohort included 4230 patients (mean [SD] age, 65.3 [12.5] years; 2034 [48.1%] female; 490 [11.6%] Asian, Hawaiian, or Pacific Islander; 554 [13.1%] Black or African American; 937 [22.1%] Hispanic; and 2249 [53.1%] non-Hispanic White). The race-neutral model had worse calibration, NPV, and false-negative rates among racial and ethnic minority subgroups than non-Hispanic White individuals (eg, false-negative rate for Hispanic patients: 12.0% [95% CI, 6.0%-18.6%]; for non-Hispanic White patients: 3.1% [95% CI, 0.8%-6.2%]). Adding race and ethnicity as a predictor improved algorithmic fairness in calibration slope, discriminative ability, PPV, and false-negative rates (eg, false-negative rate for Hispanic patients: 9.2% [95% CI, 3.9%-14.9%]; for non-Hispanic White patients: 7.9% [95% CI, 4.3%-11.9%]). Inclusion of race interaction terms or using race-stratified models did not improve model fairness, likely due to small sample sizes in subgroups. Conclusions and Relevance: In this prognostic study of the racial bias in a cancer recurrence risk algorithm, removing race and ethnicity as a predictor worsened algorithmic fairness in multiple measures, which could lead to inappropriate care recommendations for patients who belong to minoritized racial and ethnic groups. Clinical algorithm development should include evaluation of fairness criteria to understand the potential consequences of removing race and ethnicity for health inequities.

Racial Disparities in the Ascertainment of Cancer Recurrence in Electronic Health Records.

PURPOSE: There is growing interest in using computable phenotypes or proxies to identify important clinical outcomes, such as cancer recurrence, in rich electronic health records data. However, the race/ethnicity-specific accuracies of these proxies remain unclear. We examined whether the accuracy of a proxy for colorectal cancer (CRC) recurrence differed by race/ethnicity and the possible mechanisms that drove the differences. METHODS: Using data from a large integrated health care system, we identified a stratified random sample of 282 Black/African American (AA), Hispanic, and non-Hispanic White (NHW) patients with CRC who received primary treatment. Patient 5-year recurrence status was estimated using a utilization-based proxy and evaluated against the true recurrence status obtained using detailed chart review and by race/ethnicity. We used covariate-adjusted probit regression models to estimate the associations between race/ethnicity and misclassification. RESULTS: The recurrence proxy had excellent overall accuracy (positive predictive value [PPV] 89.4%; negative predictive value 96.5%; mean difference in timing 1.96 months); however, accuracy varied by race/ethnicity. Compared with NHW patients, PPV was 14.9% lower (95% CI, 2.53 to 28.6) among Hispanic patients and 4.3% lower (95% CI, -4.8 to 14.8) among Black/AA patients. The proxy disproportionately inflated the 5-year recurrence incidence for Hispanic patients by 10.6% (95% CI, 4.2 to 18.2). Compared with NHW patients, proxy recurrences for Hispanic patients were almost three times as likely to have been misclassified as positive (adjusted risk ratio 2.91 [95% CI, 1.21 to 8.31]). Higher false positives among racial/ethnic minorities may be related to higher prevalence of noncancerous lung-related problems and substantial delays in primary treatment because of insufficient patient-provider communication and abnormal treatment patterns. CONCLUSION: Using a proxy with worse accuracy among racial/ethnic minority patients to estimate population health may misdirect resources and support erroneous conclusions around treatment benefit for these patients.

Development and assessment of a natural language processing model to identify residential instability in electronic health records' unstructured data: a comparison of 3 integrated healthcare delivery systems.

OBJECTIVE: To evaluate whether a natural language processing (NLP) algorithm could be adapted to extract, with acceptable validity, markers of residential instability (ie, homelessness and housing insecurity) from electronic health records (EHRs) of 3 healthcare systems. MATERIALS AND METHODS: We included patients 18 years and older who received care at 1 of 3 healthcare systems from 2016 through 2020 and had at least 1 free-text note in the EHR during this period. We conducted the study independently; the NLP algorithm logic and method of validity assessment were identical across sites. The approach to the development of the gold standard for assessment of validity differed across sites. Using the EntityRuler module of spaCy 2.3 Python toolkit, we created a rule-based NLP system made up of expert-developed patterns indicating residential instability at the lead site and enriched the NLP system using insight gained from its application at the other 2 sites. We adapted the algorithm at each site then validated the algorithm using a split-sample approach. We assessed the performance of the algorithm by measures of positive predictive value (precision), sensitivity (recall), and specificity. RESULTS: The NLP algorithm performed with moderate precision (0.45, 0.73, and 1.0) at 3 sites. The sensitivity and specificity of the NLP algorithm varied across 3 sites (sensitivity: 0.68, 0.85, and 0.96; specificity: 0.69, 0.89, and 1.0). DISCUSSION: The performance of this NLP algorithm to identify residential instability in 3 different healthcare systems suggests the algorithm is generally valid and applicable in other healthcare systems with similar EHRs. CONCLUSION: The NLP approach developed in this project is adaptable and can be modified to extract types of social needs other than residential instability from EHRs across different healthcare systems.