Using Record Linkage to Improve Race Data Quality for American Indians and Alaska Natives in Two Pacific Northwest State Hospital Discharge Databases

Auditing Learned Associations in Deep Learning Approaches to Extract Race and Ethnicity from Clinical Text

Complete and accurate race and ethnicity (RE) patient information is important for many areas of biomedical informatics research, such as defining and characterizing cohorts, performing quality assessments, and identifying health inequities. Patient-level RE data is often inaccurate or missing in structured sources, but can be supplemented through clinical notes and natural language processing (NLP). While NLP has made many improvements in recent years with large language models, bias remains an often-unaddressed concern, with research showing that harmful and negative language is more often used for certain racial/ethnic groups than others. We present an approach to audit the learned associations of models trained to identify RE information in clinical text by measuring the concordance between model-derived salient features and manually identified RE-related spans of text. We show that while models perform well on the surface, there exist concerning learned associations and potential for future harms from RE-identification models if left unaddressed.

Community-Level Factors are Predictors of Severe Maternal Morbidity Among American Indian and Alaska Native Pregnant People in the Pacific Northwest in a Multilevel Logistic Regression

INTRODUCTION

American Indian/Alaska Native (AI/AN) pregnant people face barriers to health and healthcare that put them at risk of pregnancy complications. Rates of severe maternal morbidity (SMM) among Indigenous pregnant people are estimated to be twice that of non-Hispanic White (NHW) pregnant people.

METHODS

Race-corrected Oregon Hospital Discharge and Washington Comprehensive Hospital Abstract Reporting System data were combined to create a joint dataset of births between 2012 and 2016. The analytic sample was composed of 12,535 AI/AN records and 313,046 NHW records. A multilevel logistic regression was used to assess the relationship between community-level, individual and pregnancy risk factors on SMM for AI/AN pregnant people.

RESULTS

At the community level, AI/AN pregnant people were more likely than NHW to live in mostly or completely rural counties with low median household income and high uninsured rates. They were more likely to use Medicaid, be in a high-risk age category, and have diabetes or obesity. During pregnancy, AI/AN pregnant people were more likely to have insufficient prenatal care (PNC), gestational diabetes, and pre-eclampsia. In the multilevel model, county accounted for 6% of model variance. Hypertension pre-eclampsia, and county rurality were significant predictors of SMM among AI/AN pregnant people. High-risk age, insufficient PNC and a low county insured rate were near-significant at p < 0.10.

DISCUSSION

Community-level factors are significant contributors to SMM risk for AI/AN pregnant people in addition to hypertension and pre-eclampsia. These findings demonstrate the need for targeted support in pregnancy to AI/AN pregnant people, particularly those who live in rural and underserved communities.

Misracialization of Indigenous people in population health and mortality studies: a scoping review to establish promising practices

Indigenous people are often misracialized as other racial or ethnic identities in population health research. This misclassification leads to underestimation of Indigenous-specific mortality and health metrics, and subsequently, inadequate resource allocation. In recognition of this problem, investigators around the world have devised analytic methods to address racial misclassification of Indigenous people. We carried out a scoping review based on searches in PubMed, Web of Science, and the Native Health Database for empirical studies published after 2000 that include Indigenous-specific estimates of health or mortality and that take analytic steps to rectify racial misclassification of Indigenous people. We then considered the weaknesses and strengths of implemented analytic approaches, with a focus on methods used in the US context. To do this, we extracted information from 97 articles and compared the analytic approaches used. The most common approach to address Indigenous misclassification is to use data linkage; other methods include geographic restriction to areas where misclassification is less common, exclusion of some subgroups, imputation, aggregation, and electronic health record abstraction. We identified 4 primary limitations of these approaches: (1) combining data sources that use inconsistent processes and/or sources of race and ethnicity information; (2) conflating race, ethnicity, and nationality; (3) applying insufficient algorithms to bridge, impute, or link race and ethnicity information; and (4) assuming the hyperlocality of Indigenous people. Although there is no perfect solution to the issue of Indigenous misclassification in population-based studies, a review of this literature provided information on promising practices to consider.

Improving data capture of race and ethnicity for the Food and Drug Administration Sentinel database: a narrative review

PURPOSE

The U.S. Food and Drug Administration's Sentinel System is a national medical product safety surveillance system consisting of a large multisite distributed database of administrative claims supplemented by electronic health-care record data. The program seeks to improve data capture of race and ethnicity for pharmacoepidemiology studies.

METHODS

We conducted a narrative literature review of published research on data augmentation and imputation methods to improve race and ethnicity capture in U.S. health-care systems databases. We focused on methods with limited (five-digit ZIP codes only) or full patient identifiers available to link to external sources of self-reported data. We organized the literature by themes: (1) variation in data capture of self-reported data, (2) data augmentation from external sources of self-reported data, and (3) imputation methods, including Bayesian analysis and multiple regression.

RESULTS

Researchers reduced data missingness with high validity for Asian, Black, White, and Pacific Islander racial groups and Hispanic ethnicity. Native American and multiracial groups were difficult to validate due to relatively small sample sizes.

CONCLUSIONS

Limitations on accessible self-reported data for validation will dictate methods to improve race and ethnicity data capture. We recommend methods leveraging multiple sources that account for variations in geography, age, and sex.

The accuracy of race & ethnicity data in US based healthcare databases: A systematic review

BACKGROUND

The availability and accuracy of data on a patient's race/ethnicity varies across databases. Discrepancies in data quality can negatively impact attempts to study health disparities.

METHODS

We conducted a systematic review to organize information on the accuracy of race/ethnicity data stratified by database type and by specific race/ethnicity categories.

RESULTS

The review included 43 studies. Disease registries showed consistently high levels of data completeness and accuracy. EHRs frequently showed incomplete and/or inaccurate data on the race/ethnicity of patients. Databases had high levels of accurate data for White and Black patients but relatively high levels of misclassification and incomplete data for Hispanic/Latinx patients. Asians, Pacific Islanders, and AI/ANs are the most misclassified. Systems-based interventions to increase self-reported data showed improvement in data quality.

CONCLUSION

Data on race/ethnicity that is collected with the purpose of research and quality improvement appears most reliable. Data accuracy can vary by race/ethnicity status and better collection standards are needed.

Disparities in Misclassification of Race and Ethnicity in Electronic Medical Records Among Patients with Traumatic Injury

Systems-level barriers to self-reporting of race and ethnicity reduce the integrity of data entered into the medical record and trauma registry among patients with injuries, limiting research assessing the burden of racial disparities. We sought to characterize misclassification of self-identified versus hospital-recorded racial and ethnic identity data among 10,513 patients with traumatic injuries. American Indian/Alaska Native patients (59.9%) and Native Hawaiian/Pacific Islander patients (52.4%) were most likely to be misclassified. Most Hispanic/Latin(x) patients preferred to only be identified as Hispanic/Latin(x) (73.2%) rather than a separate race category (e.g., White). Incorrect identification of race/ethnicity also has substantial implications for the perceived demographics of patient population; according to the medical record, 82.3% of the population were White, although only 70.6% were self-identified as White. The frequency of misclassification of race and ethnicity for persons of color limits research validity on racial and ethnic injury disparities.

Surveillance Methods Used to Detect, Characterize, and Monitor the COVID-19 Pandemic in Rocky Mountain Tribal Communities

OBJECTIVE

Data were essential to public health decision-making during the COVID-19 pandemic, yet no single data source was adequate for Tribes in Montana and Wyoming. We outlined data access, availability, and limitations for COVID-19 pandemic surveillance response to improve future data exchange.

MATERIALS AND METHODS

The Rocky Mountain Tribal Epidemiology Center (RMTEC) used various data sources to deliver data on the number of COVID-19 cases, deaths, and vaccinations at local, state, and regional levels to inform Tribes in Montana and Wyoming. RMTEC reviewed state, federal, and public datasets and then attached a score to each dataset for completeness of demographic information, including race, geographic level, and refresh rate.

RESULTS

The RMTEC COVID-19 response team shared data weekly on the number of COVID-19 cases, deaths, and vaccinations distributed and the percentage of the population vaccinated with Tribal health departments in Montana and Wyoming. The Indian Health Service Epidemiology Data Mart dataset scored the highest (24 of 30), followed by datasets from Montana (18 of 30) and Wyoming (22 of 30). Publicly available datasets scored low largely due to data aggregation across larger geographic areas and lack of demographic variables.

PRACTICE IMPLICATIONS

The absence of data on race and ethnicity from publicly available data and lack of access to real-time data limited RMTEC's ability to provide Tribal-specific updates on COVID-19 cases, deaths, and vaccinations to Tribal health departments. RMTEC should be fully funded to provide the necessary resources for data management and the capacity to respond to data requests from Tribal health departments and their programs to address current and future pandemics. Federal and state agencies should also be educated on Tribal Epidemiology Centers' public health authority status to improve access to infectious disease data among those agencies.

Racial Misclassification and Disparities in Neonatal Abstinence Syndrome Among American Indians and Alaska Natives

OBJECTIVES

Maternal substance misuse can result in neonatal abstinence syndrome (NAS), a drug withdrawal process in newborns exposed in utero to drugs. This study aimed to examine the effect of racial misclassification of American Indians and Alaska Natives (AI/AN) on rates of NAS in two hospital discharge datasets in the Pacific Northwest.

METHODS

We conducted probabilistic record linkages between the Northwest Tribal Registry and Oregon and Washington hospital discharge datasets to correct racial misclassification of AI/AN people. We assessed outcomes using International Classification of Disease, Ninth Revision/Tenth Revision, Clinical Modification (ICD-9-CM or ICD-10-CM) diagnosis codes.

RESULTS

Linkage increased ascertainment of NAS cases among AI/AN by 8.8% in Oregon and by 18.1% in Washington. AI/AN newborns were 1.5 and 3.9 times more likely to be diagnosed with NAS than NHW newborns in Oregon and Washington, respectively. The results showed that newborns residing in rural Washington were 1.4 times more likely to be diagnosed with NAS than those living in urban areas.

CONCLUSIONS

Correct racial classification is an important factor in improving data quality for AI/AN populations and establishing accurate surveillance to help address the disproportionate burden of neonatal abstinence syndrome among AI/AN. The results highlight the need for programing efforts tailored by insurance status and rurality for pregnant women using substances.

Disparities in Hospitalized Chronic Obstructive Pulmonary Disease Exacerbations Between American Indians and Non-Hispanic Whites

BACKGROUND

The prevalence of chronic obstructive pulmonary disease (COPD) is high in American Indian (AI) populations, as are diabetes and obesity, which are common COPD comorbidities. However, COPD research among AI populations is limited.

STUDY DESIGN AND METHODS

We conducted a retrospective study to investigate potential health disparities and risk factors among AI and non-Hispanic White (NHW) patients with COPD exacerbations hospitalized at the University of Oklahoma Medical Center between July 2001 and June 2020. Demographics, clinical variables, and outcomes were collected.

RESULTS

A total of 76 AI patients and 304 NHW patients were included. AI patients had more comorbidities than did NHW patients (4.3 versus.3.1, p<0.001). In multiple variable analyses, AI race was associated with higher odds of needing intensive care unit (ICU) care ( odds ratio [OR], 2.37, 95% confidence interval [CI], 1.36--4.16, p=0.002) and invasive mechanical ventilator use (OR, 2.75, 95% CI, 1.42-5.29, p=0.002). AI race was also associated with longer ICU stays compared with NHWs (OR, 1.43, 95% CI, 1.18--1.73, p<0.001). The average number of days on mechanical ventilator support increased by 137.3% for an AI patient compared to an NHW patient (p<0.001). AI race was not associated with discharge to other health facilities (OR, 0.98, 95% CI, 0.52-1.83, p=0.944).

INTERPRETATION

AI patients were more likely than NHW patients to need ICU care and ventilator support, have longer ICU stays, and more days on mechanical ventilator support. More studies are needed to identify reasons for these disparities and effective interventions to reduce them.

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.

COVID-19 Pandemic and Indigenous Representation in Public Health Data

Public Health 3.0 calls for the inclusion of new partners and novel data to bring systemic change to the US public health landscape. The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic has illuminated significant data gaps influenced by ongoing colonial legacies of racism and erasure. American Indian and Alaska Native (AI/AN) populations and communities have been disproportionately affected by incomplete public health data and by the COVID-19 pandemic itself. Our findings indicate that only 26 US states were able to calculate COVID-19‒related death rates for AI/AN populations. Given that 37 states have Indian Health Service locations, we argue that public health researchers and practitioners should have a far larger data set of aggregated public health information on AI/AN populations. Despite enormous obstacles, local Tribal facilities have created effective community responses to COVID-19 testing, tracking, and vaccine administration. Their knowledge can lead the way to a healthier nation. Federal and state governments and health agencies must learn to responsibly support Tribal efforts, collect data from AI/AN persons in partnership with Indian Health Service and Tribal governments, and communicate effectively with Tribal authorities to ensure Indigenous data sovereignty. (Am J Public Health. 2021;111(S3): S208-S214. https://doi.org/10.2105/AJPH.2021.306415).

Participatory Approaches to Addressing Missing COVID-19 Race and Ethnicity Data

Varying dimensions of social, environmental, and economic vulnerability can lead to drastically different health outcomes. The novel coronavirus (SARS-CoV-19) pandemic exposes how the intersection of these vulnerabilities with individual behavior, healthcare access, and pre-existing conditions can lead to disproportionate risks of morbidity and mortality from the virus-induced illness, COVID-19. The available data shows that those who are black, indigenous, and people of color (BIPOC) bear the brunt of this risk; however, missing data on race/ethnicity from federal, state, and local agencies impedes nuanced understanding of health disparities. In this commentary, we summarize the link between racism and COVID-19 disparities and the extent of missing data on race/ethnicity in critical COVID-19 reporting. In addition, we provide an overview of the current literature on missing demographic data in the US and hypothesize how racism contributes to nonresponse in health reporting broadly. Finally, we argue that health departments and healthcare systems must engage communities of color to co-develop race/ethnicity data collection processes as part of a comprehensive strategy for achieving health equity.

Indigenous data sovereignty and COVID-19 data issues for American Indian and Alaska Native Tribes and populations

Indigenous Peoples in the United States have been experiencing disproportionate impacts of COVID-19. American Indian and Alaska Native persons are more likely to be infected, experience complications, and die from coronavirus. Evidence suggests that Indigenous persons have 3.5 times the incidence rate of non-Hispanic/Latinx whites. Unfortunately, this is likely a gross underestimate because of a lack of reliable and accurate COVID-19 data for American Indian and Alaska Native populations. Multiple factors contribute to poor data quality including the lack of Indigenous representation in the data and rampant racial misclassification at both the individual and group levels. The current pandemic has shed light on multiple pre-existing issues related to Indigenous data sovereignty in data collection and management. We discuss the importance of centring Indigenous data sovereignty in the systemic efforts to increase COVID-19 data availability and quality. The federal and state governments must support and promote Tribes' rights to access data. Federal and state governments should also focus on bolstering their data availability and quality for aggregated data on AIAN populations and for providing disaggregated Tribal data to Tribes. Given the pivotal moment in the United States with ongoing and parallel pandemics of coronavirus and racism, we urge demographers and population scientists to reflect on the role of structural racism in data, data collection and analysis.

Variation in racial/ethnic disparities in COVID-19 mortality by age in the United States: A cross-sectional study

BACKGROUND

In the United States, non-Hispanic Black (NHB), Hispanic, and non-Hispanic American Indian/Alaska Native (NHAIAN) populations experience excess COVID-19 mortality, compared to the non-Hispanic White (NHW) population, but racial/ethnic differences in age at death are not known. The release of national COVID-19 death data by racial/ethnic group now permits analysis of age-specific mortality rates for these groups and the non-Hispanic Asian or Pacific Islander (NHAPI) population. Our objectives were to examine variation in age-specific COVID-19 mortality rates by racial/ethnicity and to calculate the impact of this mortality using years of potential life lost (YPLL).

METHODS AND FINDINGS

This cross-sectional study used the recently publicly available data on US COVID-19 deaths with reported race/ethnicity, for the time period February 1, 2020, to July 22, 2020. Population data were drawn from the US Census. As of July 22, 2020, the number of COVID-19 deaths equaled 68,377 for NHW, 29,476 for NHB, 23,256 for Hispanic, 1,143 for NHAIAN, and 6,468 for NHAPI populations; the corresponding population sizes were 186.4 million, 40.6 million, 2.6 million, 19.5 million, and 57.7 million. Age-standardized rate ratios relative to NHW were 3.6 (95% CI 3.5, 3.8; p < 0.001) for NHB, 2.8 (95% CI 2.7, 3.0; p < 0.001) for Hispanic, 2.2 (95% CI 1.8, 2.6; p < 0.001) for NHAIAN, and 1.6 (95% CI 1.4, 1.7; p < 0.001) for NHAP populations. By contrast, NHB rate ratios relative to NHW were 7.1 (95% CI 5.8, 8.7; p < 0.001) for persons aged 25-34 years, 9.0 (95% CI 7.9, 10.2; p < 0.001) for persons aged 35-44 years, and 7.4 (95% CI 6.9, 7.9; p < 0.001) for persons aged 45-54 years. Even at older ages, NHB rate ratios were between 2.0 and 5.7. Similarly, rate ratios for the Hispanic versus NHW population were 7.0 (95% CI 5.8, 8.7; p < 0.001), 8.8 (95% CI 7.8, 9.9; p < 0.001), and 7.0 (95% CI 6.6, 7.5; p < 0.001) for the corresponding age strata above, with remaining rate ratios ranging from 1.4 to 5.0. Rate ratios for NHAIAN were similarly high through age 74 years. Among NHAPI persons, rate ratios ranged from 2.0 to 2.8 for persons aged 25-74 years and were 1.6 and 1.2 for persons aged 75-84 and 85+ years, respectively. As a consequence, more YPLL before age 65 were experienced by the NHB and Hispanic populations than the NHW population-despite the fact that the NHW population is larger-with a ratio of 4.6:1 and 3.2:1, respectively, for NHB and Hispanic persons. Study limitations include likely lag time in receipt of completed death certificates received by the Centers for Disease Control and Prevention for transmission to NCHS, with consequent lag in capturing the total number of deaths compared to data reported on state dashboards.

CONCLUSIONS

In this study, we observed racial variation in age-specific mortality rates not fully captured with examination of age-standardized rates alone. These findings suggest the importance of examining age-specific mortality rates and underscores how age standardization can obscure extreme variations within age strata. To avoid overlooking such variation, data that permit age-specific analyses should be routinely publicly available.

Identification of American Indians and Alaska Natives in Public Health Data Sets: A Comparison Using Linkage-Corrected Washington State Death Certificates

CONTEXT

Efforts to address disparities experienced by American Indians/Alaska Natives (AI/ANs) have been hampered by a lack of accurate and timely health data. One challenge to obtaining accurate data is determining who "counts" as AI/AN in health and administrative data sets.

OBJECTIVE

To compare the effects of definition and misclassification of AI/AN on estimates of all-cause and cause-specific mortality for AI/AN in Washington during 2015-2016.

DESIGN

Secondary analysis of death certificate data from Washington State. Data were corrected for AI/AN racial misclassification through probabilistic linkage with the Northwest Tribal Registry. Counts and age-adjusted rates were calculated and compared for 6 definitions of AI/AN. Comparisons were made with the non-Hispanic white population to identify disparities.

SETTING

Washington State.

PARTICIPANTS

AI/AN and non-Hispanic white residents of Washington State who died in 2015 and 2016.

MAIN OUTCOME MEASURES

Counts and age-adjusted rates for all-cause mortality and mortality from cardiovascular diseases, cancer, and unintentional injuries.

RESULTS

The most conservative single-race definition of AI/AN identified 1502 AI/AN deaths in Washington State during 2015-2016. The least conservative multiple-race definition of AI/AN identified 2473 AI/AN deaths, with an age-adjusted mortality rate that was 48% higher than the most conservative definition. Correcting misclassified AI/AN records through probabilistic linkage significantly increased mortality rate estimates by 11%. Regardless of definition used, AI/AN in Washington had significantly higher all-cause mortality rates than non-Hispanic whites in the state.

CONCLUSIONS

Reporting single-race versus multiple-race AI/AN had the most consequential effect on mortality counts and rates. Correction of misclassified AI/AN records resulted in small but statistically significant increases in AI/AN mortality rates. Researchers and practitioners should consult with AI/AN communities on the complex issues surrounding AI/AN identity to obtain the best method for identifying AI/AN in health data sets.

Trends in Cardiovascular Disease Morbidity and Mortality in American Indians Over 25 Years: The Strong Heart Study

Background American Indians experience high rates of cardiovascular disease. We evaluated whether cardiovascular disease incidence, mortality, and prevalence changed over 25 years among American Indians aged 30 to 85. Methods and Results The SHS (Strong Heart Study) and SHFS (Strong Heart Family Study) are prospective studies of cardiovascular disease in American Indians. Participants enrolled in 1989 to 1990 or 2000 to 2003 with birth years from 1915 to 1984 were followed for cardiovascular disease events through 2013. We used Poisson regression to analyze data for 5627 individuals aged 30 to 85 years during follow-up. Outcomes reflect change in age-specific cardiovascular disease incidence, mortality, and prevalence, stratified by sex. To illustrate generational change, 5-year relative risk compared most recent birth years for ages 45, 55, 65, and 75 to same-aged counterparts born 1 generation (23-25 years) earlier. At all ages, cardiovascular disease incidence was lower for people with more recent birth years. Cardiovascular disease mortality declined consistently among men, while prevalence declined among women. Generational comparisons were similar for women aged 45 to 75 (relative risk, 0.39-0.46), but among men magnitudes strengthened from age 45 to 75 (relative risk, 0.91-0.39). For cardiovascular disease mortality, risk was lower in the most recent versus the earliest birth years for women (relative risk, 0.56-0.83) and men (relative risk, 0.40-0.54), but results for women were inconclusive. Conclusions Cardiovascular disease incidence declined over a generation in an American Indian cohort. Mortality declined more for men, while prevalence declined more for women. These trends might reflect more improvement in case survival among men compared with women.

The US Census and the People's Health: Public Health Engagement From Enslavement and "Indians Not Taxed" to Census Tracts and Health Equity (1790-2018)

Public health professionals have long played a vital-albeit underappreciated-role in shaping, not simply using, US Census data, so as to provide the factual evidence required for good governance and health equity. Since its advent in 1790, the US Census has constituted a key political instrument, given the novel mandate of the US Constitution to allocate political representation via a national decennial census. US Census approaches to categorizing and enumerating people and places have profound implications for every branch and level of government and the resources and representation accorded across and within US states. Using a health equity lens to consider how public health has featured in each generation's political battles waged over and with census data, this essay considers three illustrations of public health's engagement with the enduring ramifications of three foundational elements of the US Census: its treatment of slavery, Indigenous populations, and the politics of place. This history underscores how public health has major stakes in the values and vision for governance that produces and uses census data.

Racial Differences in Hospitalizations Due to Injuries in South Dakota Children and Adolescents

OBJECTIVES

To determine racial differences and trends in pediatric injury hospitalization rates in a rural state.

METHODS

Hospital inpatient discharge data (2009-2014) for South Dakota residents aged 0-19 years were used to calculate annual hospitalization rates due to injuries. Race-, age-, and sex-specific rates were calculated, and trends over time were determined.

RESULTS

Between 2009 and 2014, there were 3701 pediatric hospitalizations (1008 American Indian [AI]; 2303 white) due to injuries at an average rate of 269/100,000 (95% CI 260-280/100,000). Injury hospitalization rates were higher for AI than white children (532 vs. 213 per 100,000, respectively; p < 0.001). Rates for both AI and white children increased between 2009 and 2014 (both, p < 0.001). Suicide attempts were the predominant manner of injury in both the 10-14- and 15-19-year age groups, with AI adolescents having 3.5 and 3.2 times higher rates than white adolescents. Among AI adolescents aged 15 to 19 years, hospitalizations due to homicide-related injuries were 12.6 times higher than that of white children. Injury hospitalization rates among females recently exceeded that of males, due primarily to an increase in attempted suicides. Mechanism and nature of hospitalized injuries were consistent with the high rate of suicide-related admissions.

CONCLUSION

South Dakota AI children have disproportionately higher hospitalization rates due to unintentional and attempted suicide- and homicide-related injuries, and the rate differences between AI and white children are increasing over time. Injury hospitalization rates among females have increased more rapidly and recently surpassed that of males.