Decreasing Racial Disparities in Preventable Emergency Department Visits Through Hospital Health Information Technology Patient Engagement Functionalities

Abstract

Introduction:

Hospitals are a major source of care for underserved populations in the United States. However, little is known about how hospital-based health information technology (HIT) can improve the efficiency of care and reduce disparities.

Objective:

We examined the variation of preventable emergency department (ED) visits and associated racial disparities by hospital adoption of HIT patient engagement (HIT-PE) functionalities.

Methods:

This was an observational study of 6,543,514 non-Hispanic Black (Black) and non-Hispanic White (White) adult patients using 2019 datasets of seven states (Arizona, Florida, Kentucky, Maryland, North Carolina, Vermont, Wisconsin) from the State Emergency Department Databases, American Hospital Association Annual Survey & Information Technology Supplement, and Area Health Resources File.

Results:

High HIT-PE adoption was associated with lower rates of preventable ED (odds ratio [OR] = 0.992, p < 0.001). Specific HIT-PE functions such as importing medical records from other organizations into the patient portal (OR = 0.977, p < 0.001), electronically sending medical information to a third party (OR = 0.970, p < 0.001), and scheduling appointments online (OR = 0.987, p < 0.001) were also associated with reduced preventable ED rates. Black patients had higher rates of preventable ED compared with Whites (OR = 1.386, p < 0.001); however, the interaction of Black patients and high HIT-PE adoption was associated with lower rates of preventable ED (OR = 0.977, p < 0.001). Our results also showed that higher HIT-PE adoption was associated with a reduction in preventable ED visits among Black patients with comorbidities and Black patients living in low-income areas.

Conclusions:

The results of our study suggest that there is potential to reduce preventable ED rates and racial disparities through hospital-based HIT-PE functionalities.

Introduction

Health information technology (HIT) provides the platform required to facilitate the electronic communication of providers and patients and is crucial to improving the quality and efficiency of health care.¹ Major national health initiatives such as Healthy People 2030 outlined the mission to increase digital health and considered improving health and digital literacy as one of the important tools to achieve health equity.² Previous research showed that HIT could address racial disparities by capturing meaningful data and facilitating efficient health care delivery for racial and ethnic minorities.³ HIT can also facilitate patient engagement (PE), which is defined as services to help patients use their own abilities to manage their health and health behaviors, which may lead to improved health outcomes and reduced health-related costs.⁴ Recent research showed that an increase in hospital adoption of HIT-PE capabilities led to a significant decrease in readmission rates.⁵

The objective of this study was to examine whether hospital-based HIT-PE functionalities were associated with preventable emergency department (ED) visits. ED use has steadily increased from 352.8 per 1,000 persons in 1997 to 448.2 per 1,000 persons in 2016.^6,7 In addition, the ED has become an important safety net of health care for many patients who are medically underserved and/or on Medicaid.⁶ Nearly half of all ED visits can be classified as potentially preventable or preventable,⁸ and these preventable ED visits present a great cost to the health care system, as research shows that $4.4 billion can be saved annually if the ED patients were treated at urgent care or retail clinics.⁹ We hypothesize that hospital-based HIT-PE functionalities are associated with lower odds of preventable ED visits because HIT-PE can likely improve communication of medical information with medical providers and other health care providers outside clinical settings.

We also hypothesize that hospitals' adoptions of HIT-PE functionalities are associated with reduced racial disparities in preventable ED visits. When compared with White patients, Black and Hispanic patients were more likely to have ED visits at least three times over 6 months and less likely to have a usual source of care.¹⁰ In addition, Black patients reported lower rates of PE owing to factors such as higher rates of perceived racism in health care settings compared with White patients.¹¹ Therefore, HIT-PE has the potential to benefit Black patients and facilitate better patient–clinician relationships and increase PE.¹² Finally, we hypothesize that Black patients with chronic conditions or from low socioeconomic status communities can benefit more from HIT-PE to lower the odds of preventable ED visits. HIT-PE can assist these populations by improving communication, access to health information, and supporting chronic condition self-management.^13,14

Materials and Methods

DATA

Our study sample included non-Hispanic Black (Black) and non-Hispanic White (White) adult patients who had a routine ED visit in 2019. We used the Healthcare Cost and Utilization Project data from seven states (Arizona, Florida, Kentucky, Maryland, North Carolina, Vermont, and Wisconsin) from the State Emergency Department Databases (SEDD).¹⁵ We then linked the American Hospital Association Annual Survey & Information Technology Supplement (AHAIT) to obtain hospital characteristics and hospital-based HIT variables^16,17 and also the 2019 Area Health Resources File (AHRF) to obtain county-level characteristics.

PREVENTABLE ED VISITS

We calculated preventable ED visits using the New York University (NYU) ED Algorithm, which used ICD diagnosis codes to calculate the probability of an ED visit being classified as preventable by summing the categories of nonemergent within 12 h, emergent but primary care treatable within 12 h, and emergent but preventable/avoidable if timely and effective ambulatory care had been received.¹⁸ We classified an ED visit to be preventable if the probability of being preventable is at least 75% and excluded diagnoses determined to be unclassified based on previous research methodologies.^19–21

HEALTH INFORMATION TECHNOLOGY PATIENT ENGAGEMENT

We focused on the measures of hospital-based HIT aimed to improve outpatient PE. The AHA HIT survey reported 13 HIT-PE functionalities, including whether the patient could view information from their health/medical record online, designate a family member or caregiver to access their health/medical information on behalf of the patient (e.g., proxy access), schedule appointments online, and so on.^22,23 The complete list of these HIT-PE functionalities is listed in the notes of Table 1. We calculated the total number of adopted functionalities and categorized the HIT-PE adoption into two levels: low HIT-PE, if the total number of PE adopted was below the median, that is, 0–10 total functionalities; and high HIT-PE, if the total number of PE adopted was above the median, that is, 11–13 total functionalities. We also tested individual HIT-PE functionalities as given in Table 5.

Table 1.

Study Population Characteristics by Race

VARIABLE	NON-HISPANIC WHITE (%)	NON-HISPANIC BLACK (%)	p
Hospital patient engagement functionalities			<0.001
Low/below median (0–10)	2,041,038 (45.17)	816,867 (40.34)
High/above median (11–13)	2,477,491 (54.83)	1,208,118 (59.66)
Preventable ED visit status			<0.001
Not preventable	2,509,598 (55.54)	936,833 (46.26)
Preventable	2,008,931 (44.46)	1,088,152 (53.74)
Gender			<0.001
Male	1,921,071 (42.52)	797,658 (39.39)
Female	2,597,458 (57.48)	1,227,327 (60.61)
Age group			<0.001
18–34 years old	1,242,394 (27.50)	837,397 (41.35)
35–64 years old	2,029,388 (44.91)	955,850 (47.20)
65+ years old	1,246,747 (27.59)	231,738 (11.45)
Primary payer			<0.001
Medicare	1,397,478 (30.93)	327,745 (16.19)
Medicaid	808,269 (17.89)	576,169 (28.45)
Private	1,447,795 (32.04)	581,173 (28.70)
Other	864,987 (19.14)	539,898 (26.66)
Elixhauser comorbidities			<0.001
0	2,195,488 (48.59)	1,054,120 (52.06)
≥1	2,323,041 (51.41)	970,865 (47.94)
Time index			<0.001
Q1	1,139,445 (25.22)	508,366 (25.10)
Q2	1,130,754 (25.02)	509,859 (25.18)
Q3	1,136,075 (25.14)	504,879 (24.94)
Q4	1,112,255 (24.62)	501,881 (24.78)
Income quartile by ZIP code			<0.001
0–25th percentile	1,565,907 (34.66)	1,037,071 (51.21)
26th–50th percentile	1,470,438 (32.54)	550,392 (27.18)
51th–75th percentile	976,906 (21.62)	301,445 (14.89)
76th–100th percentile	505,278 (11.18)	136,077 (6.72)
County average percent African American			<0.001
Below average (<15.50%)	2,874,358 (63.61)	420,076 (20.74)
Above average (≥15.50%)	1,644,171 (36.39)	1,604,909 (79.26)
Urban/rural residence			<0.001
Urban	3,734,116 (82.64)	1,824,146 (90.08)
Rural	784,413 (17.36)	200,839 (9.92)
Hospital beds			<0.001
<200	1,647,409 (36.46)	557,073 (27.51)
200–299	803,082 (17.77)	354,346 (17.50)
300–499	1,058,083 (23.42)	472,963 (23.36)
≥500	1,009,955 (22.35)	640,603 (31.63)
Hospital teaching status			<0.001
No	3,976,922 (88.01)	1,679,541 (82.94)
Yes	541,607 (11.99)	345,444 (17.06)
Hospital ownership			<0.001
For profit	821,404 (18.18)	351,653 (17.37)
Not-for-profit	3,010,039 (66.62)	1,250,373 (61.75)
Government	687,086 (15.21)	422,959 (20.89)
States			<0.001
Arizona	487,304 (10.78)	62,373 (3.08)
Florida	1,888,594 (41.80)	862,363 (42.59)
Kentucky	467,181 (10.34)	69,139 (3.41)
Maryland	288,177 (6.38)	314,060 (15.51)
North Carolina	862,839 (19.10)	599,183 (29.59)
Vermont	36,011 (0.80)	666 (0.03)
Wisconsin	488,423 (10.81)	117,201 (5.79)
Overall	4,518,529	2,024,985

Note: Authors' analysis of the 2019 State Emergency Department Databases, Area Health Resources File, and AHAIT. p Value is based on the χ² test. HIT-PE is the use of hospital-based HIT functionalities for the active involvement of patients and measured using 13 functionalities in the AHAIT. (1) Patient can view information from their health/medical record online, (2) patient can download information from their medical record, (3) import their medical records from other organizations into the patient portal, (4) electronically transmit (send) health/medical information from their patient portal to a third party, (5) patient can request an amendment to change/update their medical record online, (6) designate a family member or caregiver to access their health/medical information on behalf of the patient (e.g., proxy access), (7) view clinical notes online, (8) access their health/medical information using applications (apps) configured to meet the API specifications in your EHR, (9) patient can submit self-generated data, (10) send/receive secure message with providers, (11) patient can pay bills online, (12) patient can request refills for prescriptions online, (13) patient can schedule appointments online. N = 6,543,514.

AHAIT, American Hospital Association Annual Survey & Information Technology Supplement; API, application programming interfaces; ED, emergency department; EHR, Electronic Health Record; HIT-PE, health information technology patient engagement.

Table 5.

Select Health Information Technology Patient Engagement Functionalities on Preventable Emergency Department Visits Adjusted for Other Covariates

VARIABLE	OR	SE	p
Importing medical records from other organizations into the patient portal
No	Ref
Yes	0.977	0.002	<0.001
Electronically transmit (send) medical information from patient portal to a third party
No	Ref
Yes	0.970	0.003	<0.001
Request amendment to change/update their medical record online
No	Ref
Yes	0.946	0.003	<0.001
Designate family member or caregiver for proxy access to health/medical information on behalf of patient
No	Ref
Yes	0.968	0.005	<0.001
View clinical notes online from the patient portal
No	Ref
Yes	0.990	0.003	<0.001
Pay bills online
No	Ref
Yes	0.929	0.005	<0.001
Schedule appointments online
No	Ref
Yes	0.987	0.003	<0.001

Note: Authors' analysis of the 2019 State Emergency Department Databases, Area Health Resources File, and AHAIT. p Value is based on the χ² test. N = 6,543,514.

ANALYSIS

Using the Andersen Behavioral Model of Health Services Use and the National Institute on Minority Health and Health Disparities (NIMHD) research framework, we set the model specification.^24,25 The following covariates were included: patient's race, gender, age, any Elixhauser comorbidities, ZIP code income, quarter of the year being discharged, health insurance, hospital number of beds, hospital teaching status, hospital ownership status, county urban/rural status, county percent African American and state fixed effects. We also included the interaction of Black patients with HIT-PE to examine the possible variation of HIT-PE and preventable ED visit by race. Finally, we applied our model among subpopulations who might additionally benefit from HIT-PE, including individuals with chronic conditions and individuals who resided in low-income neighborhoods.

In this study, we focused on White and Black adult patients with complete information measures. Our analysis excluded federal and specialty hospitals, which is in line with HIT research guidelines from the Office of the National Coordinator (ONC) for HIT, because the included hospitals are more similar in financing mechanisms and patient population characteristics compared with federal and specialty hospitals.^26,27 Previous studies showed that hospitals with more than one primary EMR (Electronic Medical Record) or EHR (Electronic Health Record) were more likely to experience problems with patient safety such as synchronizing accurate medical information, providing patient information, and sending health information, and also were less likely to implement patient HIT access capabilities.^27,28 Therefore, we excluded hospitals with more than one primary EMR/EHR system across outpatient sites to keep our hospital HIT sample more homogeneous.

The sample size of this linked data set was 8,211,888. We then excluded patients who had an unclassified preventable ED status. Our final sample size included 6,543,514 patients with 4,518,529 White patients and 2,024,985 Black patients who have routine discharges that did not result in a hospital admission. The study has obtained the approval from the Institutional Review Board of the University of Maryland.

Results

Figure 1 provides the preventable ED rate by race and levels of exposure to HIT-PE. Being treated in hospitals with high HIT-PE adoption was associated with a 2.52% point reduction (45.84–43.32%) of preventable ED among White patients (p < 0.001), and a 2.69% point reduction among Black patients (55.34–52.65%, p < 0.001).

Fig. 1.

Black/White preventable ED rates by HIT-PE. ED, emergency department; HIT-PE, health information technology patient engagement.

Table 1 provides the study population characteristics for White and Black patients. Results showed significant differences by race for all covariates. Black patients were significantly more likely to undergo a preventable ED visit (53.74% vs. 44.46%, p < 0.001), live in an urban county (90.08% vs. 82.64% p < 0.001), live in a county with an above-average percentage of African Americans (79.26% vs. 36.39%, p < 0.001), and go to a hospital with high HIT-PE (59.66% vs. 54.83%, p < 0.001) than White patients. Black patients were also more likely to be younger than 65 years of age, live in a ZIP code in the lowest income quartile, have Medicaid, go to a teaching hospital, go to a government-owned hospital, and go to a large hospital with at least 500 beds while less likely to have comorbidities.

Table 2 provides the results of the multivariate logistic regression. Black patients were more likely to have a preventable ED visit (odds ratio [OR] = 1.386, p < 0.001) compared with White patients. Patients who went to a hospital with high HIT-PE were less likely to have a preventable ED visit compared with patients who went to a hospital with low HIT-PE (OR = 0.992, p < 0.001). The interaction of Black patients and high HIT-PE shows a significant decrease in the odds of having a preventable ED visit (OR = 0.977, p < 0.001). Other significant variables show that younger adult patient groups, living in a ZIP code that is not in the highest quartile of income, nonprivate insurance users, having comorbidities, discharge quarter, teaching hospital status, for-profit hospital status, rural county residence, and state fixed effects were associated with higher odds of having a preventable ED visit.

Table 2.

Logistic Regression of Preventable Emergency Department Visits by Health Information Technology Patient Engagement Functionalities

VARIABLE	OR	SE	p
Race
Non-Hispanic White	Ref
Non-Hispanic Black	1.386	0.004	<0.001
Hospital HIT-PE functionalities
Low/below median (0–10)	Ref
High/above median (11–13)	0.992	0.002	<0.001
Black*high HIT-PE
No	Ref
Yes	0.977	0.003	<0.001
Gender
Male	Ref
Female	1.242	0.002	<0.001
Age group
18–34 years old	Ref
35–64 years old	0.914	0.002	<0.001
65+ years old	0.853	0.003	<0.001
Primary payer
Private	Ref
Medicare	1.147	0.003	<0.001
Medicaid	1.272	0.003	<0.001
Other	1.203	0.003	<0.001
Elixhauser comorbidities
0	Ref
≥1	1.072	0.002	<0.001
Time index
Q1	Ref
Q2	0.901	0.002	<0.001
Q3	0.871	0.002	<0.001
Q4	0.966	0.002	<0.001
Income quartile by ZIP code
76th–100th percentile	Ref
0–25th percentile	1.189	0.004	<0.001
26th–50th percentile	1.140	0.003	<0.001
51th–75th percentile	1.106	0.003	<0.001
County average percent African American
Below average (<15.50%)	Ref
Above average (≥15.50%)	1.038	0.002	<0.001
Urban/rural residence
Urban	Ref
Rural	1.012	0.003	<0.001
Hospital beds
<200	Ref
200–299	0.954	0.002	<0.001
300–499	0.907	0.002	<0.001
≥500	0.836	0.002	<0.001
Hospital teaching status
No	Ref
Yes	1.049	0.003	<0.001
Hospital ownership
For profit	Ref
Not-for-profit	0.956	0.003	<0.001
Government	0.949	0.003	<0.001
States
North Carolina	Ref
Arizona	0.906	0.003	<0.001
Florida	0.966	0.002	<0.001
Kentucky	0.907	0.003	<0.001
Maryland	0.790	0.003	<0.001
Vermont	0.885	0.010	<0.001
Wisconsin	0.885	0.003	<0.001

Note: Authors' analysis of the 2019 State Emergency Department Databases, Area Health Resources File, and AHAIT. p Value is based on the χ² test. HIT-PE is the use of hospital-based HIT functionalities for the active involvement of patients and measured using 13 functionalities in the AHAIT. N = 6,543,514.

OR, odds ratio; SE, standard error.

Tables 3 and 4 provide additional analyses when comparing racial disparities for patients with comorbidities and patients living in low-income ZIP codes. Black patients with comorbidities were more likely to have a preventable ED visit compared with White patients with comorbidities (OR = 1.404, p < 0.001), whereas all patients with comorbidities who went to a hospital with high HIT-PE were less likely to have a preventable ED visit (OR = 0.973, p < 0.001). The interaction of Black patients who had comorbidities and high HIT-PE showed a significant decrease in the odds of having a preventable ED visit (OR = 0.974, p < 0.001). In addition, Black patients living in low-income ZIP codes were more likely to have a preventable ED visit compared with White patients living in low-income ZIP codes (OR = 1.373, p < 0.001), whereas all patients living in low-income ZIP codes who went to a hospital with high HIT-PE were less likely to have a preventable ED visit (OR = 0.978, p < 0.001). The interaction of Black patients who lived in low-income ZIP codes and high HIT-PE showed a significant decrease in the odds of having a preventable ED visit (OR = 0.969, p < 0.001).

Table 3.

Subanalysis of Study Population with Chronic Conditions

VARIABLE	OR	SE	p
Race
Non-Hispanic White	Ref
Non-Hispanic Black	1.404	0.006	<0.001
Hospital HIT-PE functionalities
Below (0–10)	Ref
Above (11–13)	0.973	0.003	<0.001
Black*high HIT-PE
No	Ref
Yes	0.974	0.005	<0.001

Note: Authors' analysis of the 2019 State Emergency Department Databases, Area Health Resources File, and AHAIT. p Value is based on the χ² test. HIT-PE is the use of hospital-based HIT functionalities for the active involvement of patients and measured using 13 functionalities in the AHAIT. N = 3,293,906.

Table 4.

Subanalysis of Study Population Living in a Low-Income ZIP Code

VARIABLE	OR	SE	p
Race
Non-Hispanic White	Ref
Non-Hispanic Black	1.373	0.006	<0.001
Hospital HIT-PE functionalities
Below (0–10)	Ref
Above (11–13)	0.978	0.004	<0.001
Black*high HIT-PE
No	Ref
Yes	0.969	0.005	<0.001

Note: Authors' analysis of the 2019 State Emergency Department Databases, Area Health Resources File, and AHAIT. p Value is based on the χ² test. HIT-PE is the use of hospital-based HIT functionalities for the active involvement of patients and measured using 13 functionalities in the AHAIT. N = 2,602,978.

Table 5 tests the association of individual HIT-PE functionalities. Specific HIT-PE functionalities such as the ability to import medical records from other organizations into the patient portal (OR = 0.977, p < 0.001), electronically sending medical information to a third party (OR = 0.970, p < 0.001), requesting an amendment to change/update their medical information online (OR = 0.946, p < 0.001), designate a family member or caregiver proxy access to health/medical information on behalf of the patient (OR = 0.968, p < 0.001), view clinical notes online from the patient portal (OR = 0.990, p < 0.001), pay bills online (OR = 0.929, p < 0.001), and schedule appointments online (OR = 0.987, p < 0.001) were significantly associated with lower odds of preventable ED visits.

Discussion

Our results show the evidence that hospital-based HIT for PE purposes was associated with lower odds of preventable ED visits and reduced disparities in preventable ED visits among Black and White patients. We speculate that HIT-PE can decrease preventable ED visits and associated racial disparities by improving patient activation. Patient activation is defined as patients' willingness to take action to manage their health and care options and is an essential component of PE. However, racial minorities tend to report lower levels of patient activation that leads to less trust in medical providers and less efficiency in treatments.^29,30 Improving health literacy and education has been found to mediate racial disparities in patient activation, and HIT-PE can provide the platform for clinicians to improve patient health education in these areas.³¹

The results also suggest that HIT-PE is associated with reduced preventable ED through improved third-party communication of medical information and accessing medical information with centralized databases. The capability to import medical records from other organizations into the patient portal and electronically send medical information from their patient portal to third parties were found to be significantly associated with lower rates of preventable ED visits. EHR, a primary source of medical data, could be imported into the patient portal for clinical use.³² Medical information such as laboratory data can be sent to ambulatory care, and such data exchange, possibly through centralized databases, can produce better patient follow-up and engagement regarding their post-ED treatment options.³²

Requesting an amendment to change/update patients' medical records online was also found to be significantly associated with lower preventable ED rates. This finding supports policies advocated by the American Academy of Professional Coders, which state that it is important for patients to be able to ask to change their medical record because of mistakes and provide notes to discuss for follow-up consultations and order medical services.³³ In addition, the HIT-PE functionality of the ability to designate family members or caregivers for proxy access to health/medical information on behalf of the patient was significantly associated with lower preventable ED rates. This is especially important for patients with comorbidities such as cancer or dementia because it allows family members and caregivers to engage in communication and decision-making with clinicians.³⁴

HIT's ability to allow patients to view clinical notes online through the patient portal is important for PE. Previous research showed that patients who had access to their clinical notes had higher perceived insight into their health conditions and treatment plans, and allowing patients to review notes can build greater trust in understanding provider assessments.³⁵ HIT's ability that allows patients to pay bills online has been shown to be correlated with positive patient experiences.³⁶ With the ability to pay bills online, patients can be better informed about the costs of treatment and reduce confusion about the costs of care. Finally, HIT's ability to allow patients to schedule appointments online is crucial for follow-up care. It is especially important for patients with comorbidities, as previous research showed that scheduling appointments online is associated with better care management.³⁷

The results of our study also suggest that HIT-PE can help reduce the racial gap among patients with comorbidities. Prevention and management of chronic conditions are a main target of policies to reduce health and socioeconomic disparities and achieve equitable outcomes.³⁸ A systematic review found that HIT could help facilitate the communication of a patient's history, symptoms, and opinions to the provider, and such communications helped with the management of patient conditions (such as in-depth discussions about symptoms and quality of life).³⁹ This communication supported by HIT-PE can be especially important for patients with chronic conditions to create an appropriate long-term treatment plan.³⁹

We also found evidence that patients living in low-income communities had worse preventable ED rates and that HIT-PE could help reduce the racial gap for patients living in low-income areas. Low-income communities of color are especially vulnerable to health disparities, and community-based organizations provide a much-needed infrastructure to promote a culture of health.⁴⁰ Although HIT-PE has great potential to improve health delivery and outcomes, many disadvantaged patients and communities have uneven access to these technologies, as hospitals that serve low-income communities tend to adopt HIT at lower rates.⁴¹

Although Black patients were more likely to be treated in hospitals with high HIT-PE, they also had a higher rate of preventable ED rates. Yet without high HIT-PE, the rate of preventable EDs among Black patients would be even higher, and as such, strengthening HIT-PE functionalities for Black patients becomes critical for achieving health equity. The results of our study call for more attention to racial disparities and future studies may control for unobserved factors such as the disease severity or cultural preference that can be associated with the racial variation of encountering preventable ED visits.

There are several limitations to our study. First, the results of our study showed an associative relationship but not a causal link. Second, although our study controlled for comprehensive multilevel measures using the NIMHD framework and evidence from the literature, we might still have omitted measures, such as disease severity and culture preference, owing to the data availability. Third, this study measured the intent-to-treat effect of HIT and does not measure patients' actual use of HIT-PE functionalities. We used the hospital-reported HIT measures. Future studies may also include HIT-PE at the community and primary care setting. Fourth, our study only included hospitals that responded to the AHA HIT survey. The included hospitals in our sample could serve different patient populations compared with excluded hospitals. Finally, our analysis covered hospital discharges from multiple states and the results may not be nationally representative.

Conclusions

Although many hospitals have made improvements in HIT-PE, about half of our study population went to a hospital with low HIT-PE. Encouraging investments in hospital-based HIT-PE as well as educating health providers in the use of HIT-PE is crucial to realizing the full potential of HIT. Our study showed that going to a hospital that has high HIT-PE can reduce the odds of having a preventable ED visit and reduce the racial disparity between non-Hispanic Black and non-Hispanic White patients. Future studies can examine specific arrangements of HIT that can target specific health needs of racial and ethnic minorities to be able to realize the full potential of HIT benefits.

Authors' Contributions

Authorship confirmation/contribution statement: N.W.: conceptualization, data curation, formal analysis, investigation, software, visualization, writing—original draft, writing—review and editing. J.C.: conceptualization, funding acquisition, methodology, project administration, resources, supervision, validation, writing—review and editing.

Disclosure Statement

No competing financial interests exist.

Funding Information

Funding was provided by the National Institutes of Health, NIMHD grant R01MD011523 and National Institutes of Health, National Institute on Aging grant R01AG062315.