Provision of disaggregated race and ethnicity information in hospital visits

Abstract

Introduction

Addressing and measuring health racialized inequities requires health care datasets to include race and ethnicity variables.¹ Commonly used racial designations, such as American Indian or Alaska Native (AIAN), Asian, Black or African American, Native Hawaiian or Pacific Islander (NHPI), and White, may be too broad and mask within-group disparities.^2-6 For instance, although Asian patients grouped together were comparable to or healthier than non-Hispanic White patients, Filipino patients were more likely than non-Hispanic White patients to be in fair or poor health, be obese or overweight, and have certain chronic illnesses.³

The need for data disaggregation in race and ethnicity has increasingly been recognized.^4,7-10 In 2023, 4 states passed laws requiring state agencies to disaggregate race and ethnicity.¹¹ In 2024, the Office of Management and Budget (OMB) changed rules to make collecting detailed race and ethnicity the default (Appendix I).¹² The rules additionally add Middle Eastern or North African (MENA) as a new minimum designation and combine ethnicity and race in a single question.¹² Since organizations including the Joint Commission¹³ use the OMB racial and ethnic designations, these changes may encourage data disaggregation by health providers.

However, few studies have examined responses to laws encouraging data disaggregation. Health care organizations might vary in data collection,¹⁴ or patients could decline to provide more detail.

This descriptive study uses New York State discharge data to examine whether hospitals collect and patients report detailed race and ethnic group information when state policies support this. Since 2014, the New York Statewide Planning and Research Cooperative System (SPARCS) encouraged state-licensed hospitals to collect detailed race and ethnic group information for patients identifying as Asian, NHPI, Hispanic, Latinx, or of Spanish origin.¹⁵ For instance, for “race,” patients could select Asian or Chinese, Filipino, or Indonesian, and for “ethnicity,” they could select Hispanic or Mexican, South American, or Puerto Rican. The SPARCS did not specifically encourage detailed ethnic group information for patients identifying with other races—for example, AIAN, Black, or White. This study compares whether patients identifying as Hispanic, Asian, or NHPI, compared with those identifying as AIAN, Black, or White, were more likely to provide detailed race and ethnic group information. We further examine which hospitals collected detailed data.

Data and methods

We merged New York all-payer hospital discharge data (2016–2022) from SPARCS with American Hospital Association data on hospital characteristics (2016–2022) and 5-year 2021 American Community Survey data using hospital county. The SPARCS data contain all inpatient and emergency department visits to state-licensed hospitals. We excluded non–acute care facilities. Analyses report visit-responses, as patients could select more than 1 race or ethnicity group. For the race analysis, we excluded visit-responses for “Other” race, as no further detail was possible. “Other” race was included for analyses for Hispanic patients since SPARCS data treat race and ethnicity as 2 separate questions. In sensitivity analyses, we examined trends by year (Appendix II).

We examined the percentage of visit-responses where detailed race and ethnic group were provided. We categorized detailed ethnic group information by a patient's identified “major” race group, using the 1997 OMB definitions for race (in the place when the data were collected).¹⁶

We calculated the percentage of hospitals collecting detailed race and ethnic group information. We investigated 4 potential reasons that hospitals might collect, and patients might report, more detailed race and ethnic group information.

First, patients living in more diverse areas might be more likely to provide detailed race and ethnic group information. To examine this, we compared the county-level percentage of Hispanic, Asian, or NHPI population with the corresponding percentage of patients reporting detailed ethnic group information for hospitals in that county. We both mapped and calculated the county-level correlation between these measures.

Second, hospitals with greater awareness of the importance of data disaggregation may be more likely to collect this information. We tested this by excluding New York City from our results, as hospitals in New York City may be more aware of its importance (Appendix III and IV).

Third, hospitals may be more likely to collect more detailed race and ethnic group information if they had a higher percentage of patients who were Hispanic, Asian, or NHPI; hospitals treating fewer of these patients might not see the need to collect granular information.¹⁷ To examine this, we compared whether the hospitals collecting detailed race or ethnic group information had a larger share of visit-responses for each major race group compared with those that did not collect detailed race or ethnic group information.

Fourth, hospitals with more resources might be more likely to collect more detailed race and ethnic group information. We examined differences in hospital and community characteristics for hospitals that collected detailed race and ethnic group data vs those that did not. As we were interested in factors enabling data collection, we examined hospital size, academic hospital status, median household income, area deprivation index, and urbanicity. All comparisons used chi-square or t tests, as appropriate.

The study was approved as exempt by the Pennsylvania State University Institutional Review Board.

Limitations

This study has limitations. First, we could not compare pre-post data because, prior to the 2014 requirement, the data did not include detailed race or ethnic group information. However, during the study period, the number of hospitals collecting detailed race or ethnic group information increased (Appendix II). Second, the data do not detail how race and ethnicity are collected. Although SPARCS requires that race and ethnicity data be through patient self-report,¹⁵ in practice, hospitals may vary in how they collect these data, which may result in data quality problems.^18-21 Third, the study is descriptive so does not control for factors influencing detailed race and ethnic group provision. For instance, patients may react negatively when asked about race during emergency department (ED) registration processes,^21,22 and staff training may be required. Fourth, patients' interpretation of race and ethnicity may differ from how it is coded in the data.²¹ For instance, because SPARCS asks separate questions about race and ethnicity, we report these separately, but some individuals may view race and ethnicity as interchangeable. Fifth, this study may have limited generalizability, both because it is focused only on New York State and because both the state and organizations there may be more enthusiastic about collecting more detailed race and ethnic group information, particularly for Asian and NHPI residents.^23-26

Results

From 2016 to 2022, there were 58 188 733 visits and 59 078 774 visit-responses made to 190 nonfederal acute care hospitals in New York. Excluding patients with “other” race, 2.4% of visit-responses were by patients who provided more detailed ethnic group information, whereas the remainder provided only the broad racial category (Figure 1). Patients who identified as Asians and NHPI were significantly more likely to report detailed ethnic group information than patients identified as White, Black or African American, or AIAN (P < .001). Across all hospitals, 42.3% of visit-responses by patients identifying as Asian and 57.8% by patients identifying as NHPI reported detailed data, compared to 0.003% of visit-responses by patients who identified as White, 0.1% of visit-responses by Black or African American individuals, and 2.3% of visit-responses by AIAN individuals. A total of 40.1% of patients who identified as Hispanic provided detailed ethnicity data (Figure 1). The most common detailed race and ethnic group information among patients who identified as Asian, NHPI, and Hispanic is shown in Appendix V.

Figure 1.

Percentage of visits with a detailed race and ethnic group response in New York State, 2016–2022. Source: Authors' analysis. The figure shows the visit-responses with more detailed race and ethnic group information over all visit-responses for that major race group at all hospitals. Note that each person can respond with multiple responses; each would be counted separately.

Open in new tabDownload slide

We examined 4 reasons that patients and hospitals might be more likely to provide detailed race and ethnic group information. As described in the Data and methods, these are as follows: patients living in more racially and ethnically diverse areas, hospitals' awareness of the importance, hospitals treating a higher proportion of racially and ethnically diverse patients, and hospitals having more enabling resources to collect disaggregated data.

To examine whether patients in more diverse areas were more likely to provide detailed race and ethnic group information, we mapped the percentage of patients providing detailed race or ethnic group information with a higher county density of Asian, NHPI, and Hispanic individuals. Figure 2 suggests that counties where Asian, NHPI, and Hispanic patients are more likely to provide detailed ethnic group (larger dots) are not always counties with a higher percentage of Asian, NHPI, or Hispanic individuals (darker colors). The correlation between the 2 was similarly low (Asian: r = 0.19922; NHPI: 0.06825; Hispanic: 0.19631).

Figure 2.

A–C: Patients who identified as Hispanic, Asian, or NHPI who reported detailed ethnic group information compared with the county-level population of Hispanic, Asian, or NHPI individuals, 2021. Source: Authors' analysis. The figure shows the percentage of individuals identifying as Hispanic, Asian, or NHPI living in each county (color) vs the percentage of patients who identified as Hispanic, Asian, or NHPI who provided detailed ethnic group information (size of dots). Counties with fewer than 15 patients who identified as Hispanic, Asian, or NHPI and who provided detailed ethnic group information are excluded from this figure (ie, the dot does not appear). Abbreviation: NHPI, Native Hawaiian or Pacific Islander.

Open in new tabDownload slide

To examine if hospital awareness of the importance of data disaggregation explained the results, we examined results after excluding New York City, and results were similar (Appendix IV).

Not all hospitals collected detailed race and ethnic group information. Significantly more hospitals collected detailed race and ethnic group information for patients identified as Hispanic (76.2%), Asian (73.2%), and NHPI (76.8%) compared with 2 hospitals that collected race for AIAN, Black/African American, and White patients (data not shown) (P < .00001). Notably, 96.2% of patients identifying as AIAN at the single hospital collecting detailed ethnic group information for AIAN provided detailed information.

Among all visit-responses by patients identified as Asian, 91.6% of visit-responses were in hospitals that collected detailed ethnic group information for patients who identified as Asian (data not shown). This percentage was 91.5% for patients identifying as NHPI, 87.7% for patients identifying as Hispanic, 2.4% for patients identifying as AIAN, 2.0% for patients identifying as Black or African American, and 0.7% for patients identifying as White.

Finally, we examined hospital and community characteristics associated with collecting detailed race and ethnic group information. Compared with hospitals that did not collect detailed race and ethnic group information for Asian, NHPI, or Hispanic patients, hospitals that did were more likely to be large hospitals, in more urban areas, and in areas with a lower area deprivation index (Appendix VI). Results were similar after excluding New York City, except there was no difference in area deprivation index (Appendix IV).

Conclusion

The SPARCS encouraged the collection of detailed race and ethnic group data for Hispanic, Asian, and NHPI individuals. Hospitals and patients were more likely to report detailed race and ethnic group data for these patients. For instance, 42.3% and 57.8% of patients identifying as Asian and NHPI, respectively, provided detailed ethnic group information, compared to only 2.4% of patients overall.

Although 70% of hospitals collected detailed race and ethnic group information on Asian, NHPI, or Hispanic patients, these hospitals served approximately 90% of patients identifying as Asian, NHPI, or Hispanic. This was because they were hospitals serving more patients, in more urban areas, and serving less disadvantaged counties. Modifying data collection for more detailed race and ethnic group information is labor-intensive, taking 1 hospital system in New York several years.¹⁴ The hospital system met with SPARCS, modified registration systems, and educated and trained staff.¹⁴ Better-resourced hospitals may be better equipped to make these changes, raising questions about the need for greater resources with the OMB 2024 update. Resources could be allocated through the Centers for Medicare and Medicaid Services (CMS) Electronic Health Record (EHR) Incentive Programs and requirements under meaningful use, which include race and ethnicity data in their criteria.

The OMB update makes collecting detailed ethnic group information the default for all major race groups and Hispanic ethnicity.¹² Only 2 hospitals collected detailed ethnic group information in our study beyond Asian, NHPI, and Hispanic. At the 1 hospital collecting detailed ethnic group information on AIAN patients, 96.2% of AIAN patients provided more detail, suggesting high willingness to provide this information.

The specific detailed race and ethnicity categories under the OMB update are the 6 largest ethnic groups in the United States. For instance, under “Asian,” individuals can select “Asian” or Chinese, Asian Indian, Filipino, Vietnamese, Korean, or Japanese categories. Notably, these detailed race and ethnic group categories were not the most numerous visit-responses in SPARCS data. Thus, organizations may want to engage with their communities and tailor the detailed race and ethnic group categories for their local populations.²⁷

Health equity has become increasingly important in health policy, and growing research suggests the importance of data disaggregation for race. Given the new OMB rule, hospitals and health systems may be concerned about whether patients actually provide disaggregated race and ethnicity information. This study suggests that they would. Also, it suggests that laws supporting data disaggregation may encourage hospitals, although it is possible that hospitals decided on their own to collect these data for Asian, Hispanic, and NHPI patients, particularly because state and New York City policies supported collection of detailed race and ethnic group information.^23-26

Acknowledgments

Previous versions of this manuscript were presented as poster presentations at the AcademyHealth Annual Research Meeting (Baltimore, June 2024) and the American Public Health Association Annual Meeting and Expo (Minneapolis, October 2024). This publication was produced from raw data purchased from or provided by the New York State Department of Health (NYSDOH). However, the calculations, metrics, conclusions derived, and views expressed herein are those of the author(s) and do not reflect the conclusions or views of NYSDOH. The NYSDOH, its employees, officers, and agents make no representation, warranty, or guarantee as to the accuracy, completeness, currency, or suitability of the information provided here.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

This work was supported by the National Institute on Minority Health and Health Disparities (R01MD017495), National Center for Advancing Translational Sciences (UL1 TR002014) (C.H.), and Robert Wood Johnson Foundation (79779) (N.A.P.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health (NIH) or the Robert Wood Johnson Foundation.

Notes

Author notes