Racial Differences in Data Quality and Completeness: Spinal Cord Injury Model Systems' Experiences

Improvements in data completeness in health information systems reveal racial inequalities: longitudinal national data from hospital admissions in Brazil 2010-2022

BACKGROUND

Race and ethnicity are important drivers of health inequalities worldwide. However, the recording of race/ethnicity in data systems is frequently insufficient, particularly in low- and middle-income countries. The aim of this study is to descriptively analyse trends in data completeness in race/color records in hospital admissions and the rates of hospitalizations by various causes for Blacks and Whites individuals.

METHODS

We conducted a longitudinal analysis, examining hospital admission data from Brazil's Hospital Information System (SIH) between 2010 and 2022, and analysed trends in reporting completeness and racial inequalities. These hospitalization records were examined based on year, quarter, cause of admission (using International Classification of Diseases (ICD-10) codes), and race/color (categorized as Black, White, or missing). We examined the patterns in hospitalization rates and the prevalence of missing data over a period of time.

RESULTS

Over the study period, there was a notable improvement in data completeness regarding race/color in hospital admissions in Brazil. The proportion of missing values on race decreased from 34.7% in 2010 to 21.2% in 2020. As data completeness improved, racial inequalities in hospitalization rates became more evident - across several causes, including assaults, tuberculosis, hypertensive diseases, at-risk hospitalizations during pregnancy and motorcycle accidents.

CONCLUSIONS

The study highlights the critical role of data quality in identifying and addressing racial health inequalities. Improved data completeness has revealed previously hidden inequalities in health records, emphasizing the need for comprehensive data collection to inform equitable health policies and interventions. Policymakers working in areas where socioeconomic data reporting (including on race and ethnicity) is suboptimal, should address data completeness to fully understand the scale of health inequalities.

Racial Differences in Accuracy of Predictive Models for High-Flow Nasal Cannula Failure in COVID-19

OBJECTIVES

To develop and validate machine learning (ML) models to predict high-flow nasal cannula (HFNC) failure in COVID-19, compare their performance to the respiratory rate-oxygenation (ROX) index, and evaluate model accuracy by self-reported race.

DESIGN

Retrospective cohort study.

SETTING

Four Emory University Hospitals in Atlanta, GA.

PATIENTS

Adult patients hospitalized with COVID-19 between March 2020 and April 2022 who received HFNC therapy within 24 hours of ICU admission were included.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Four types of supervised ML models were developed for predicting HFNC failure (defined as intubation or death within 7 d of HFNC initiation), using routine clinical variables from the first 24 hours of ICU admission. Models were trained on the first 60% (n = 594) of admissions and validated on the latter 40% (n = 390) of admissions to simulate prospective implementation. Among 984 patients included, 317 patients (32.2%) developed HFNC failure. eXtreme Gradient Boosting (XGB) model had the highest area under the receiver-operator characteristic curve (AUROC) for predicting HFNC failure (0.707), and was the only model with significantly better performance than the ROX index (AUROC 0.616). XGB model had significantly worse performance in Black patients compared with White patients (AUROC 0.663 vs. 0.808, p = 0.02). Racial differences in the XGB model were reduced and no longer statistically significant when restricted to patients with nonmissing arterial blood gas data, and when XGB model was developed to predict mortality (rather than the composite outcome of failure, which could be influenced by biased clinical decisions for intubation).

CONCLUSIONS

Our XGB model had better discrimination for predicting HFNC failure in COVID-19 than the ROX index, but had racial differences in accuracy of predictions. Further studies are needed to understand and mitigate potential sources of biases in clinical ML models and to improve their equitability.

State insurance mandates and racial and ethnic inequities in assisted reproductive technology utilization

OBJECTIVE

To examine whether the (1) scope of state-mandated insurance coverage for assisted reproductive technology (ART) and (2) proportion of the population eligible for this coverage are associated with reductions in racial/ethnic inequities in ART utilization.

DESIGN

National cross-sectional, ecologic study.

SUBJECTS

We employed estimates from the US Census Bureau of all women 20-44 years of age living in the US in 2018. Data on the number of women who initiated an ART cycle during that year that were reported to the US Centers for Disease Control and Prevention were obtained from the National ART Surveillance System.

EXPOSURE

State mandates were classified according to the scope of required coverage for fertility services: Comprehensive, Limited, and No Mandate.

MAIN OUTCOME MEASURES

Race and ethnic-specific ART utilization rates, defined as the number of women undergoing ≥1 ART cycles per 10,000 women, were the primary outcomes. As state mandates do not apply to all insurance plans, Comprehensive Mandate utilization rates were recalculated using denominators corrected for the estimated proportions of populations eligible for coverage.

RESULTS

Across all mandate categories, Non-Hispanic (NH) Asian and NH White populations had the highest ART utilization rates, whereas the lowest rates were among Hispanic, NH Black, and NH Other/Multiple Races populations. Compared with the NH Asian reference group, the NH Black population had smaller inequities in the Comprehensive Mandate group than the No Mandate group (rate ratio [RR 0.33 [0.28-0.38] vs. RR 0.23 [0.22-0.24]). Using the Comprehensive Mandate group for each race/ethnicity as the reference, the NH Black and NH Other/Multiple Races populations showed the largest relative differences in utilization between the No Mandate and Comprehensive Mandate groups (RR 0.39 [0.37-0.41] and 0.33 [0.28-0.38], respectively). Within the Comprehensive Mandate group, the disparities in the Hispanic and NH Black populations moved toward the null after correcting for state-mandated insurance eligibility.

CONCLUSIONS

Racial/ethnic inequities in ART utilization were reduced in states with comprehensive infertility coverage mandates. Inequities were further attenuated after correcting for mandate eligibility. Mandates alone, however, were not sufficient to eliminate disparities. These findings can inform future strategies aimed at improving ART access under a social justice framework.

Racial and Ethnical Discrepancies and Similarities in the Epidemiology, Survival, and Neurological Outcomes After Acute Traumatic Spinal Cord Injury: A Retrospective Cohort Study Using Data from the NASCIS-1 Trial

Background

Little is known about the impact of race/ethnicity on the clinical and neurological outcomes after acute traumatic spinal cord injury (tSCI).

Objectives

This study examined the influence of race/ethnicity on the individuals' survival and neurological recovery within the first year after tSCI.

Methods

The 306 cases enrolled in the First National Acute Spinal Cord Injury Study (NASCIS-1) were grouped as African American individuals (n = 84), non-Hispanic White individuals (n = 159), and other races/ethnicities that included Hispanic individuals (n = 60) and Asian individuals (n = 3). Outcome measures included survival and neurological recovery within the first year after tSCI. Data analyses were adjusted for major potential confounders.

Results

There were 39 females and 267 males with mean age of 31 years who mostly sustained cervical severe tSCI after motor vehicle accidents or falls. The three groups were comparable regarding sex distribution, level and severity of tSCI, level of consciousness at admission, and total received dose of methylprednisolone. African American individuals were significantly older than non-Hispanic White individuals (p = .0238). African American individuals and individuals of other races/ethnicities more often had a tSCI with open wound caused by missile and water-related accidents than non-Hispanic White individuals (p < .0001). Survival rates within the first year after tSCI were comparable among the three groups (p = .3191). Among the survivors, there were no significant differences among the three groups regarding motor and pinprick and light-touch sensory recovery (p > .0500).

Conclusions

The results of this study suggest that, while there were few differences among the racial/ethnical groups regarding the epidemiology of tSCI, race/ethnicity did not influence survival rate or neurological recovery within the first year post-tSCI.

A Scoping Review of Approaches to Improving Quality of Data Relating to Health Inequalities

Identifying and monitoring of health inequalities requires good-quality data. The aim of this work is to systematically review the evidence base on approaches taken within the healthcare context to improve the quality of data for the identification and monitoring of health inequalities and describe the evidence base on the effectiveness of such approaches or recommendations. Peer-reviewed scientific journal publications, as well as grey literature, were included in this review if they described approaches and/or made recommendations to improve data quality relating to the identification and monitoring of health inequalities. A thematic analysis was undertaken of included papers to identify themes, and a narrative synthesis approach was used to summarise findings. Fifty-seven papers were included describing a variety of approaches. These approaches were grouped under four themes: policy and legislation, wider actions that enable implementation of policies, data collection instruments and systems, and methodological approaches. Our findings indicate that a variety of mechanisms can be used to improve the quality of data on health inequalities at different stages (prior to, during, and after data collection). These findings can inform us of actions that can be taken by those working in local health and care services on approaches to improving the quality of data on health inequalities.

The quality of social determinants data in the electronic health record: a systematic review

OBJECTIVE

The aim of this study was to collect and synthesize evidence regarding data quality problems encountered when working with variables related to social determinants of health (SDoH).

MATERIALS AND METHODS

We conducted a systematic review of the literature on social determinants research and data quality and then iteratively identified themes in the literature using a content analysis process.

RESULTS

The most commonly represented quality issue associated with SDoH data is plausibility (n = 31, 41%). Factors related to race and ethnicity have the largest body of literature (n = 40, 53%). The first theme, noted in 62% (n = 47) of articles, is that bias or validity issues often result from data quality problems. The most frequently identified validity issue is misclassification bias (n = 23, 30%). The second theme is that many of the articles suggest methods for mitigating the issues resulting from poor social determinants data quality. We grouped these into 5 suggestions: avoid complete case analysis, impute data, rely on multiple sources, use validated software tools, and select addresses thoughtfully.

DISCUSSION

The type of data quality problem varies depending on the variable, and each problem is associated with particular forms of analytical error. Problems encountered with the quality of SDoH data are rarely distributed randomly. Data from Hispanic patients are more prone to issues with plausibility and misclassification than data from other racial/ethnic groups.

CONCLUSION

Consideration of data quality and evidence-based quality improvement methods may help prevent bias and improve the validity of research conducted with SDoH data.